KR20230120168A - Waste battery thermal runaway prevention system using ai prediction model and sensor monitoring - Google Patents

Abstract

The waste battery thermal runaway prevention system monitors abnormal heat generation and predicts thermal runaway in advance based on the voltage value, temperature value, humidity value, and off-gas emission value measured in the waste battery. The waste battery thermal runaway prevention system includes a sensor module, a camera module, a data transmission device, a server, and an output module. The voltage sensor is coupled to the OBD terminal of the waste battery to measure the voltage value of the waste battery. The temperature sensor contacts one surface of the waste battery to measure a temperature value. A thermal runaway prediction model is stored in the server. The thermal runaway prediction model is pre-learned based on voltage and temperature values. The server predicts the possibility of thermal runaway of the waste battery using a thermal runaway prediction model based on the voltage value, the temperature value, and the humidity value.

Description

Waste battery thermal runaway prevention system using AI prediction model and sensor monitoring {WASTE BATTERY THERMAL RUNAWAY PREVENTION SYSTEM USING AI PREDICTION MODEL AND SENSOR MONITORING}

The present invention relates to a waste battery thermal runaway prevention system capable of integratedly managing waste batteries and predicting the occurrence of threats, and specifically, with respect to the sensed voltage value, temperature value, humidity value, and off-gas emission value It relates to a waste battery thermal runaway prevention system that predicts thermal runaway of a waste battery by applying a pre-learned thermal runaway prediction model using an artificial intelligence algorithm.

Due to the vitalization of the electric vehicle market, waste battery emissions are increasing. Accordingly, it is expected that the recycling market of waste batteries will grow in earnest. In this process, the importance of technology for safely storing waste batteries over the long/short term is also increasing. In the case of actual ESS (Energy Storage System) facilities, 26 explosions and fires have occurred in the past two years.

Specifically, there are three major causes of fire and explosion that may occur while storing waste batteries (thermal threat due to overheating, electrical threat due to overcharging/discharging and short circuit, and physical threat due to external shock). do. Due to the above three causes, thermal runaway occurs through the steps of ABUSE, offgas, smoke, and fire.

In order to prevent the thermal runaway phenomenon, a technology for monitoring the temperature in the ABUSE stage of the waste battery and a technology for detecting gas in the off-gas stage have been introduced. However, the conventional techniques have a problem in that only some (voltage only or temperature only) of three factors (voltage, temperature, and off-gas) capable of predicting thermal runaway are monitored. That is, there is a growing demand for a technology for predicting whether thermal runaway will occur by comprehensively collecting the three factors (voltage, temperature, off-gas) directly or indirectly related to fire.

The present invention is to solve the above conventional problems, in addition to the temperature and off-gas of the waste battery, the voltage and ambient humidity of the waste battery are monitored, thereby providing a waste battery thermal runaway prevention system with an advanced threat situation detection function. intended to provide

In addition, an object of the present invention is to provide a waste battery thermal runaway prevention system that predicts thermal runaway through artificial intelligence learning based on the measured voltage, temperature, off-gas, and humidity of the waste battery.

A thermal runaway prevention system according to an embodiment of the present invention may include a sensor module, a camera module, a data transmission device, a server, and an output module.

The sensor module may include a voltage sensor and a temperature sensor. The voltage sensor may measure the voltage value of the waste battery. A temperature sensor may be disposed adjacent to the waste battery to measure a temperature value. The sensor module may generate sensing data including the voltage value and the temperature value.

The camera module may generate image data by acquiring images of the waste battery and the surrounding environment in which the waste battery is stored.

A data transmission device may collect and wirelessly transmit the sensing data and the image data.

The server may include a pre-learned thermal runaway prediction model based on the temperature value and the voltage value. The server may receive the sensing data from the data transmission device. The server may receive the image data from the camera module. The server may calculate the possibility of thermal runaway of the waste battery using the thermal runaway prediction model based on the sensing data.

The output module may receive the sensing data, the image data, and the possibility of thermal runaway from the server, and output a step alarm including at least one of image, text, and sound in response thereto.

The sensor module of the thermal runaway prevention system according to an embodiment of the present invention may further include an off-gas detection sensor and a humidity sensor.

An off-gas detection sensor may be disposed adjacent to the waste battery to measure an off-gas emission value.

A humidity sensor may be disposed adjacent to the waste battery to measure a humidity value.

The sensing data may further include the off-gas emission value and the humidity value.

In the thermal runaway prevention system according to an embodiment of the present invention, the voltage sensor is coupled to the OBD terminal of the waste battery to measure the voltage value of the waste battery. The temperature sensor may be disposed to contact the surface of the waste battery to measure the temperature of the surface of the waste battery. The off-gas detection sensor may measure at least one of carbon monoxide, alcohol, oxygen, fine dust, and composite smoke.

The thermal runaway prediction model is pre-learned based on input data determined by assigning a first weight to the voltage value, a second weight to the temperature value, and a third weight to the off-gas emission value. can The first weight may be greater than the second weight, and the second weight may be greater than the third weight.

If the humidity value measured by the humidity sensor is greater than or equal to a predetermined value, the possibility of thermal runaway is multiplied by a first value greater than 1, and if the humidity value is less than the predetermined value, a second value less than 1 is associated with the possibility of thermal runaway. can be multiplied to calculate the final thermal runaway probability.

When the temperature value is greater than or equal to 50 degrees Celsius and less than 120 degrees Celsius, or the emission value of the off-gas is greater than or equal to 0.02 mol, the server may generate a first output signal. The output module may receive the first output signal and output a first stage alarm warning that there is a precursor of thermal runaway of the waste battery.

When the temperature value is greater than 120 degrees Celsius and less than 220 degrees Celsius, or the temperature value increases by more than 0.25 degrees Celsius per second, the server may generate a second output signal. The output module may receive the second output signal and output a second stage alarm warning that thermal runaway of the waste battery is highly likely to occur after 630 seconds to 690 seconds.

When the temperature value is 220 degrees Celsius or higher, the server may generate a third output signal. The output module may receive a third output signal and output a third stage alarm warning that the thermal runaway of the waste battery is highly likely to occur within a few seconds.

The thermal runaway prediction model may be designed as a layer based on a Recurrent Neural Network (RNN). A Recurrent Neural Network (RNN) includes a forget gate to remove the input data, an update gate to reflect the input data, an output gate to output a result, and It may be composed of a Long Short Term Memory (LSTM) cell including a store gate for performing the next learning step.

The output module of the thermal runaway prevention system according to an embodiment of the present invention may include at least one of a PC, a smart phone, a tablet PC, a PDA, and a speaker.

The data transmission device of the thermal runaway prevention system according to an embodiment of the present invention may be a synchronous LoRa modem.

A thermal runaway prevention method according to an embodiment of the present invention may include a preliminary learning step, a monitoring step, a thermal runaway possibility calculation step, and an output step.

In the pre-learning step, thermal runaway prediction learned based on the voltage value of the waste battery measured by the voltage sensor, the temperature value of the waste battery measured by the temperature sensor, and the off-gas emission value of the waste battery measured by the off-gas detection sensor A model can be created.

In the monitoring step, it may be determined whether or not the waste battery is abnormal by monitoring the voltage value, the temperature value, and the off-gas emission value.

In the thermal runaway possibility calculation step, the thermal runaway possibility may be calculated using the thermal runaway prediction model based on the voltage value, the temperature value, and the offgas emission value.

In the output step, at least one of an image, text, and sound corresponding to the possibility of thermal runaway and whether or not the waste battery is abnormal may be output as a step alarm.

The preliminary learning step of the thermal runaway prevention method according to an embodiment of the present invention may include a sensing data collection step, an input data generation step, and a predictive model learning step.

In the sensing data collection step, a voltage value, a temperature value, an off-gas emission value, and a humidity value may be collected. The voltage value may be measured by the voltage sensor coupled to the OBD PORT of the waste battery. The temperature value may be measured by the temperature sensor disposed to contact the surface of the waste battery. The off-gas emission value may include an emission amount of at least one of carbon monoxide, alcohol, oxygen, fine dust, and composite smoke measured by an off-gas detection sensor disposed adjacent to the waste battery. The humidity value may be measured by a humidity sensor disposed adjacent to the waste battery.

A first weight is assigned to the voltage value, a second weight less than the first weight is assigned to the temperature value, and a third weight less than the second weight is assigned to the off-gas emission value to determine input data. can

In the prediction model learning step, a thermal runaway prediction model learned based on the input data may be generated.

The thermal runaway prediction model may be designed as a Recurrent Neural Network (RNN)-based layer composed of Long Short Term Memory (LSTM) cells. A Long Short Term Memory (LSTM) cell has a forget gate to remove the input data, an update gate to reflect the input data, and an output gate to output the result. , and a store gate for performing the next learning step.

In the calculating the possibility of thermal runaway, if the humidity value is greater than or equal to a predetermined value, the possibility of thermal runaway is multiplied by a first value greater than 1, and if the humidity value is less than the predetermined value, the possibility of thermal runaway is multiplied by a first value less than 1. The final thermal runaway probability multiplied by a value of 2 can be calculated.

In the monitoring step, the server may generate a first output signal when the temperature value is greater than 50 degrees Celsius and less than 120 degrees Celsius, or the offgas emission value is greater than 0.02 mol. The server may generate a second output signal when the temperature value is greater than 120 degrees Celsius and less than 220 degrees Celsius, or when the temperature value increases by more than 0.25 degrees Celsius per second. When the temperature value is 220 degrees Celsius or higher, the server may generate a third output signal.

In the output step, the output module may receive the first output signal from the server and output a first step alarm warning that there is a precursor symptom of thermal runaway of the waste battery. The server may receive the second output signal and output a second stage alarm warning that the thermal runaway of the waste battery is highly likely to occur after 630 seconds to 690 seconds. The server may receive the third output signal and output a third stage alarm warning that the thermal runaway of the waste battery is highly likely to occur within a few seconds.

According to one embodiment of the present invention, since the voltage and ambient humidity of the waste battery are monitored in addition to the temperature and off-gas of the waste battery, a waste battery thermal runaway prevention system with an advanced threat situation detection function can be provided.

According to one embodiment of the present invention, based on the voltage, temperature, off-gas, and humidity of the waste battery, it is possible to provide a waste battery thermal runaway prevention system that predicts thermal runaway through artificial intelligence learning.

1 schematically illustrates a thermal runaway prediction system.
FIG. 2 is an enlarged view of the battery rack shown in FIG. 1 .
3 is a block diagram showing the configuration of a thermal runaway prediction system.
4 is a flowchart of a thermal runaway prediction method.
Figure 5 is a flow chart showing the pre-learning step shown in Figure 4 in detail.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings.

In the drawings, proportions and dimensions of components may be exaggerated for effective description of technical content.

The term "includes" is intended to indicate that there are features, numbers, steps, operations, components, parts, or combinations thereof described in the specification, but one or more other features, numbers, steps, operations, or configurations. It should be understood that it does not preclude the possibility of the presence or addition of elements, parts or combinations thereof.

1 schematically illustrates a thermal runaway prediction system 1000 . FIG. 2 is an enlarged view of the battery rack (BTL) shown in FIG. 1 . 3 is a block diagram showing the configuration of a thermal runaway prediction system 1000.

Thermal runaway refers to a fire caused by an internal short-circuit caused by an increase in the internal temperature of a battery due to thermal, electrical, or physical stress and melting of a separator inside the battery. The thermal runaway prediction system 1000 of the present invention detects whether thermal runaway will occur in real time based on the voltage value, temperature value, humidity value, off-gas emission value, etc. of a waste battery (BT) being stored and predicts in advance will be.

1 to 3, the thermal runaway prediction system 1000 of the present invention includes a sensor module 100, a camera module 200, a data transmission device 300, a server 400, and an output module 500. can include Waste batteries (BT) may be stacked and stored in a battery rack (BTL).

The sensor module 100 may include a voltage sensor 110 , a temperature sensor 120 , a humidity sensor 130 , and an off-gas detection sensor 140 .

Voltage sensor 110 may be coupled to the OBD terminal of the waste battery (BT). The voltage sensor 110 may measure the voltage value of the waste battery BT. Until thermal runaway, the voltage value of the waste battery (BT) goes through a series of changes. At the beginning of thermal runaway, 1) the voltage value rises gently, 2) the voltage value decreases through an inflection point, 3) after a certain period of time, 4) the voltage value rises rapidly, leading to thermal runaway. Using this series of voltage value changes, thermal runaway of the waste battery BT can be predicted in advance by monitoring the voltage value at all times.

The temperature sensor 120 may be disposed adjacent to the waste battery BT. The temperature sensor 120 may measure a temperature value. In one embodiment of the present invention, the temperature sensor 120 may be disposed to contact the surface of the waste battery (BT). In addition, the temperature sensor 120 may measure the temperature of the surface of the waste battery BT. Until thermal runaway, the temperature value of the waste battery (BT) goes through a series of changes. 1) After the temperature value of the cell surface is maintained at about 100℃ 2) for a certain period of time 3) the temperature rises rapidly and thermal runaway occurs. Using this series of temperature value changes, thermal runaway of the waste battery (BT) can be predicted in advance by monitoring the temperature value at all times.

The humidity sensor 130 may be disposed adjacent to the waste battery BT. Specifically, it may be disposed on a battery rack (BTL). The humidity sensor 130 may measure a humidity value around. When the humidity of the space storing the waste battery BT is high, dew condensation may occur inside the waste battery BT. Condensation combines with dust in the storage space to weaken the insulation of the waste battery (BT) and cause a short circuit, increasing the risk of fire. Therefore, it is important to continuously monitor and manage the humidity value in the storage space.

The off-gas detection sensor 140 may be disposed adjacent to the waste battery BT. The off-gas detection sensor 140 may measure an off-gas emission value. When a battery is subjected to thermal, physical, or electrical stress, the temperature inside the battery rises, and the electrolyte is vaporized and vented, which is called off-gas. Therefore, the thermal runaway phenomenon can be prevented in advance by detecting the off-gas emission value in real time.

In another embodiment of the present invention, the off-gas detection sensor 140 may detect at least one of carbon monoxide, alcohol, oxygen, fine dust, and composite smoke.

The sensing data DT1 may include a voltage value, a temperature value, an off-gas emission value, and a humidity value.

The camera module 200 may generate image data DT2 by acquiring an image of the waste battery BT and the surrounding environment in which the waste battery BT is stored. Therefore, a user using the thermal runaway prediction system 1000 of the present invention can visually monitor the internal situation of the storage and respond promptly by utilizing the image obtained from the camera module 200 .

In another embodiment of the present invention, by applying a deep learning algorithm to an image acquired by the camera module 200, it is possible to prevent a fire and a precursor symptom of a fire situation. For example, when steam is generated from a waste battery, it can be detected from an image, automatically contact a nearby fire station and management personnel, and take action to respond to a fire in an early stage.

The data transmission device 300 may collect sensing data DT1 and image data DT2 and transmit them wirelessly. In an embodiment of the invention, data transmission device 300 may be a synchronous LoRa modem. The data transmission device 300 of the present invention can maintain high stability by preventing wireless collisions that may occur by using multiple types of sensors through time synchronization. The data transmission device 300 can communicate at the correct time with minimum delay by wirelessly transmitting on a timeslot basis.

Also, the data transmission device 300 may mix the sensing data DT1 and the image data DT2 and provide the mixed data to the user. That is, images and data corresponding to the images may be comprehensively provided to the user on the basis of time. In addition, since wireless communication of the LoRa method is utilized through the data transmission device 300, the thermal runaway prediction system 1000 of the present invention can be used in a wide area of up to 3 km, and the installation and maintenance costs are low. There is an advantage to that.

However, the data transmission device 300 of the present invention is not limited thereto. A modem (MODEM) capable of performing modulation and demodulation of a signal corresponding to the sensing data DT1 and transmitting it wirelessly and in real time is sufficient.

The server 400 may include a thermal runaway prediction model 410 . The server 400 may receive and store the sensing data DT1 from the data transmission device 300 . The server 400 may receive and store image data DT2 from the camera module 200 . A thermal runaway prediction model 410 may be stored in the server 400 . The server 400 may calculate the remaining life of the waste battery BT, the ignition time, and the possibility of thermal runaway of the waste battery BT using the thermal runaway prediction model 410 based on the sensing data DT1. there is.

The thermal runaway prediction model 410 may be trained in advance based on the temperature value and the voltage value. Specifically, the thermal runaway prediction model 410 assigns a first weight to a voltage value, a second weight to a temperature value, and a third weight to an off-gas emission value, based on input data determined in advance. can be learned The first weight may be greater than the second weight, and the second weight may be greater than the third weight. However, the present invention is not limited thereto, and the magnitude relationship and specific values between the first weight, the second weight, and the third weight may be changed as needed by a designer.

The thermal runaway prediction model 410 may be designed as a layer based on a Recurrent Neural Network (RNN). In Recurrent Neural Network (RNN), hidden layers form a chain structure. In addition, since there is a loop in which the operation result of the hidden layer is connected to the same hidden layer again, it has a structure in which past data affects future operation. In other words, Recurrent Neural Network (RNN) forms a structure in which hidden nodes are connected to directed edges and circulate.

A Recurrent Neural Network (RNN) may be composed of Long Short Term Memory (LSTM) cells. A Long Short Term Memory (LSTM) cell has a structure in which several gates are added to the hidden layer by solving the long-term dependency problem. That is, Long Short Term Memory (LSTM) cells are added to solve the vanishing gradient problem that occurs when performing iterative learning using gradient descent.

A long short term memory (LSTM) cell may include a forget gate, an update gate, an output gate, and a store gate.

A forget gate can remove data. That is, the forget gate may determine how much to forget the previous cell state.

An update gate may reflect input data. That is, the update gate may determine how much previous data is to be reflected.

An output gate can determine how much of a result to output.

The Store Gate can perform the next learning step.

If the humidity value measured by the humidity sensor 130 is greater than or equal to a predetermined value, the server 400 multiplies the possibility of thermal runaway by a first value greater than 1, and if the humidity value is less than the predetermined value, the server 400 multiplies the possibility of thermal runaway. The possibility of final thermal runaway may be calculated by multiplying by a second value smaller than 1. In another embodiment of the present invention, the predetermined value may be between 55 and 65%. However, the present invention is not limited thereto, and a designer may change a predetermined value as needed.

The output module 500 may include at least one of a PC, a smart phone, a tablet PC, a PDA, and a speaker. The output module 500 may receive sensing data DT1, image data DT2, and integrated data DT3 including the possibility of thermal runaway. The output module 500 may output a step alarm including at least one of image, text, and sound in response to the integrated data DT3.

The stage alarm may include a first stage alarm, a second stage alarm, and a third stage alarm. The step alarm may be provided in the form of at least one of images, texts, and sounds on the output module 500 .

When the temperature value measured by the temperature sensor 120 is greater than or equal to 50 degrees Celsius and less than 120 degrees Celsius, or the off-gas emission value measured by the off-gas detection sensor 140 is greater than or equal to 0.02 mol, the server 400 outputs a first output signal. can create The output module 500 may receive the first output signal and output a first stage alarm warning that there is a precursor of thermal runaway of the waste battery BT. In addition, the output module 500 may send a message to a pre-stored personal communication device of a person in charge or provide a checklist and response manual for preventing thermal runaway.

When the temperature value is greater than 120 degrees Celsius and less than 220 degrees Celsius, or the temperature value increases by more than 0.25 degrees Celsius per second, the server 400 may generate a second output signal. The output module 500 may receive the second output signal and output a second stage alarm warning that there is a high possibility of thermal runaway of the waste battery BT after 630 to 690 seconds. In addition, the output module 500 may send a message to a pre-stored personal communication device of a person in charge or send a warning message through an indoor speaker.

When the temperature value is 220 degrees Celsius or higher, the server 400 may generate a third output signal. The output module 500 may receive the third output signal and output a third stage alarm warning that the thermal runaway of the waste battery BT is highly likely to occur within a few seconds. In addition, the output module 500 may transmit a previously stored message to a nearby fire station or transmit a signal for controlling the operation of fire prevention facilities including fire shutters, sprinklers, and emergency alarms to the outside.

4 is a flowchart of a thermal runaway prediction method (S100). Figure 5 is a flow chart showing in detail the pre-learning step (S10) shown in Figure 4.

Referring to FIG. 4 , the thermal runaway prediction method ( S100 ) may include a pre-learning step ( S10 ), a monitoring step ( S20 ), a thermal runaway possibility calculation step ( S30 ), and an output step ( S40 ). Referring to FIG. 5 , the pre-learning step (S10) may include a sensing data collection step (S11), an input data generation step (S12), and a predictive model learning step (S13).

In the pre-learning step (S10), the voltage value of the waste battery measured by the voltage sensor 110, the temperature value of the waste battery measured by the temperature sensor 120, and the waste battery measured by the off-gas detection sensor 140 A thermal runaway prediction model 410 learned based on the offgas emission value may be generated.

In the sensing data collection step ( S11 ), a voltage value, a temperature value, and a humidity value may be collected. The voltage value can be measured by the voltage sensor 110 coupled to the OBD PORT of the waste battery. The temperature value may be measured by the temperature sensor 120 disposed to contact the surface of the waste battery. The off-gas emission value may be measured by the off-gas detection sensor 140 disposed adjacent to the waste battery. The off-gas emission value may include an emission amount of at least one of carbon monoxide, alcohol, oxygen, fine dust, and composite smoke measured by the off-gas detection sensor 140 .

The humidity value may be measured by the humidity sensor 130 disposed adjacent to the waste battery.

In the input data generation step (S12), a first weight is assigned to the voltage value, a second weight less than the first weight is assigned to the temperature value, and a third weight less than the second weight is assigned to the off-gas emission value. Input data can be determined.

In the predictive model learning step (S13), the thermal runaway prediction model 410 (FIG. 3) learned based on the input data may be generated.

The thermal runaway prediction model 410 may be designed as a layer based on a Recurrent Neural Network (RNN). In Recurrent Neural Network (RNN), hidden layers form a chain structure. In addition, since there is a loop in which the operation result of the hidden layer is connected to the same hidden layer again, it has a structure in which past data affects future operation. In other words, Recurrent Neural Network (RNN) forms a structure in which hidden nodes are connected to directed edges and circulate. A recurrent neural network (RNN) may include a long short term memory (LSTM).

A Recurrent Neural Network (RNN) may be composed of Long Short Term Memory (LSTM) cells. A Long Short Term Memory (LSTM) cell has a structure in which several gates are added to the hidden layer by solving the long-term dependency problem. That is, Long Short Term Memory (LSTM) cells are added to solve the vanishing gradient problem that occurs when performing iterative learning using gradient descent.

A long short term memory (LSTM) cell may include a forget gate, an update gate, an output gate, and a store gate.

A forget gate can remove data. That is, the forget gate may determine how much to forget the previous cell state.

An update gate may reflect input data. That is, the update gate may determine how much previous data is to be reflected.

An output gate can determine how much of a result to output.

The Store Gate can perform the next learning step.

In the monitoring step (S20), it may be determined whether or not the waste battery is abnormal by monitoring the voltage value, the temperature value, and the off-gas emission value.

When the temperature value measured by the temperature sensor 120 is greater than or equal to 50 degrees Celsius and less than 120 degrees Celsius, or the emission value of off-gas is greater than or equal to 0.02 mol, the server 400 may generate a first output signal. The server 400 may generate a second output signal when the temperature value is greater than 120 degrees Celsius and less than 220 degrees Celsius, or when the temperature value increases by more than 0.25 degrees Celsius per second. When the temperature value is 220 degrees Celsius or higher, the server 400 may generate a third output signal. The first output signal, the second output signal, and the third output signal generated by the server 400 may be transmitted to the output module 500 .

In step S30 of calculating the possibility of thermal runaway, the possibility of thermal runaway may be calculated using the thermal runaway prediction model 410 based on the voltage value, the temperature value, and the off-gas emission value.

When the humidity value is greater than or equal to a predetermined value, the final possibility of thermal runaway may be calculated by multiplying the possibility of thermal runaway by a first value greater than 1. When the humidity value is less than the predetermined value, a final possibility of thermal runaway obtained by multiplying the possibility of thermal runaway by a second value smaller than 1 may be calculated.

In an embodiment of the present invention, the monitoring step (S20) and the thermal runaway possibility calculation step (S30) may be performed simultaneously. However, it is not limited thereto, and the order of the monitoring step (S20) and the thermal runaway possibility calculation step (S30) may be changed.

In the output step (S40), at least one of an image, text, and sound corresponding to the possibility of thermal runaway and whether or not the waste battery is abnormal may be output.

The output module 500 may receive a first output signal from the server 400 and output a first stage alarm warning that there is a precursor symptom of thermal runaway of the waste battery. The output module 500 may receive the second output signal and output a second stage alarm warning that the possibility of thermal runaway of the waste battery is high after 630 seconds to 690 seconds. The output module 500 may receive the third output signal and output a third stage alarm warning that the thermal runaway of the waste battery is highly likely to occur within a few seconds.

Although described with reference to the examples, those skilled in the art can understand that the present invention can be variously modified and changed without departing from the spirit and scope of the present invention described in the claims below. There will be. In addition, the embodiments disclosed in the present invention are not intended to limit the technical idea of the present invention, and all technical ideas within the scope of the following claims and their equivalents should be construed as being included in the scope of the present invention. .

1000: thermal runaway prediction system 100: sensor module
110: voltage sensor 120: temperature sensor
130: Humidity sensor 140: Off detection sensor
200: camera module 300: data transmission device
400: server 500: output module

Claims (7)

A sensor module including a voltage sensor for measuring a voltage value of a waste battery and a temperature sensor disposed adjacent to the waste battery to measure a temperature value, and generating sensing data including the voltage value and the temperature value;
A camera module for generating image data by acquiring images of the waste battery and the surrounding environment in which the waste battery is stored;
a data transmission device capable of collecting and wirelessly transmitting the sensing data and the image data;
A thermal runaway prediction model pre-learned based on the temperature value and the voltage value is included, the data transmission device receives the sensing data, the camera module receives the image data, and the sensing data is received based on the sensing data. a server for calculating the possibility of thermal runaway of the waste battery using the thermal runaway prediction model; and
Waste battery thermal runaway comprising an output module that receives the sensing data, the image data, and the possibility of thermal runaway from the server, and outputs a step alarm including at least one of image, text, and sound in response thereto prevention system. According to claim 1,
The sensor module,
Further comprising an off-gas detection sensor disposed adjacent to the waste battery to measure an off-gas emission value and a humidity sensor disposed adjacent to the waste battery to measure a humidity value,
The sensing data further comprises the off-gas emission value and the humidity value. According to claim 2,
The voltage sensor is coupled to the OBD terminal of the waste battery to measure the voltage value of the waste battery,
The temperature sensor is disposed to contact the surface of the waste battery to measure the temperature of the surface of the waste battery,
The off-gas detection sensor measures at least one of carbon monoxide, alcohol, oxygen, fine dust, and composite smoke,
The thermal runaway prediction model is trained in advance based on input data determined by assigning a first weight to the voltage value, a second weight to the temperature value, and a third weight to the off-gas emission value, ,
The first weight is greater than the second weight, the second weight is greater than the third weight,
If the humidity value measured by the humidity sensor is greater than or equal to a predetermined value, the possibility of thermal runaway is multiplied by a first value greater than 1, and if the humidity value is less than the predetermined value, a second value less than 1 is associated with the possibility of thermal runaway. Multiply by to calculate the final thermal runaway probability,
When the temperature value is greater than or equal to 50 degrees Celsius and less than 120 degrees Celsius, or the off-gas emission value is greater than or equal to 0.02 mol, the server generates a first output signal, and the output module receives the first output signal to detect the waste gas. Outputs a first-stage alarm warning that there is a precursor of thermal runaway of the battery;
When the temperature value is greater than 120 degrees Celsius and less than 220 degrees Celsius, or the temperature value increases by more than 0.25 degrees Celsius per second, the server generates a second output signal, and the output module receives the second output signal. So, after 630 seconds to 690 seconds, a second stage alarm is output warning that the possibility of thermal runaway of the waste battery is high,
A third step in which the server generates a third output signal when the temperature value is 220 ° C or higher, and the output module receives the third output signal and warns that there is a high possibility of thermal runaway of the waste battery within a few seconds output an alarm,
The thermal runaway prediction model includes a forget gate to remove the input data, an update gate to reflect the input data, an output gate to output a result, and a next learning step. A waste battery thermal runaway prevention system designed as a layer based on a Recurrent Neural Network (RNN) consisting of Long Short Term Memory (LSTM) cells including a store gate to perform. According to claim 2,
The output module is a waste battery thermal runaway prevention system comprising at least one of a PC, a smartphone, a tablet PC, a PDA, and a speaker. According to claim 3,
The data transmission device is a synchronous LoRa modem waste battery thermal runaway prevention system. A dictionary in which a learned thermal runaway prediction model is generated based on the voltage value of the waste battery measured by the voltage sensor, the temperature value of the waste battery measured by the temperature sensor, and the off-gas emission value of the waste battery measured by the off-gas detection sensor. learning phase;
A monitoring step of determining whether or not the waste battery is abnormal by monitoring the voltage value, the temperature value, and the off-gas emission value;
a thermal runaway possibility calculation step of calculating a thermal runaway possibility using the thermal runaway prediction model based on the voltage value, the temperature value, and the offgas emission value; and
A waste battery thermal runaway prevention method comprising an output step of outputting at least one of an image, text, and sound corresponding to the possibility of thermal runaway and whether or not the waste battery is abnormal. According to claim 6,
The pre-learning step,
The voltage value measured by the voltage sensor coupled to the OBD PORT of the waste battery, the temperature value measured by the temperature sensor arranged to contact the surface of the waste battery, off-gas detection disposed adjacent to the waste battery Sensing in which off-gas emission values including emissions of at least one of carbon monoxide, alcohol, oxygen, fine dust, and composite smoke measured by a sensor and humidity values measured by a humidity sensor disposed adjacent to the waste battery are collected data collection step;
A first weight is assigned to the voltage value, a second weight less than the first weight is assigned to the temperature value, and a third weight less than the second weight is assigned to the off-gas emission value to determine input data. input data generation step; and
A predictive model learning step in which a thermal runaway prediction model learned based on the input data is generated;
The thermal runaway prediction model includes a forget gate to remove the input data, an update gate to reflect the input data, an output gate to output a result, and a next learning step. It is designed as a Recurrent Neural Network (RNN)-based layer composed of Long Short Term Memory (LSTM) cells including a store gate to perform
In the thermal runaway possibility calculation step,
If the humidity value is greater than or equal to a predetermined value, the probability of thermal runaway is multiplied by a first value greater than 1, and if the humidity value is less than a predetermined value, the probability of thermal runaway is multiplied by a second value less than 1, resulting in final thermal runaway. possibilities are calculated,
In the monitoring step,
When the temperature value is greater than 50 degrees Celsius and less than 120 degrees Celsius, or the offgas emission value is greater than or equal to 0.02 mol, the server generates a first output signal, and the temperature value is greater than 120 degrees Celsius and less than 220 degrees Celsius, or When the temperature value increases by 0.25 degrees Celsius or more per second, the server generates a second output signal, and when the temperature value increases by 220 degrees Celsius or more, the server generates a third output signal;
In the output stage,
The output module receives the first output signal from the server and outputs a first stage alarm for warning that there is a precursor of thermal runaway of the waste battery, and receives the second output signal from the server to set a time range of 630 seconds to 690 seconds. After seconds, a second stage alarm is output to warn that thermal runaway of the waste battery is highly likely to occur, and the server receives the third output signal to warn that the thermal runaway of the waste battery is highly likely to occur within a few seconds A method for preventing thermal runaway of a waste battery outputting a third step alarm.