US-12627142-B2 - Thermal-runaway warning method, system, and terminal for power station

US12627142B2US 12627142 B2US12627142 B2US 12627142B2US-12627142-B2

Abstract

A thermal-runaway warning method, system, and terminal for a power station are provided. The method comprises: obtaining collected data of a battery module comprised in the power station, wherein the collected data comprises collected temperature values of the battery module collected by temperature measuring devices; normalizing the collected data to obtain a standard dataset; training a CNNLSTM-based temperature predicting model based on the standard dataset to obtain a trained CNNLSTM-based temperature predicting model; and obtaining a predicted temperature value of the battery module within an output time window based on the trained CNNLSTM-based temperature predicting model, and determining whether there is a risk of thermal runaway in the power station based on the predicted temperature value. The present disclosed thermal-runaway warning method, system, and terminal for a power station realize effective warning of thermal runaway for a power station by predicting the temperature of battery modules in the power station.

Inventors

Haowen REN
Enhai Zhao
Peng Ding
Weikun Wu
DECHENG WANG
Yuan Feng
Wei Song
Guopeng Zhou
Zonglin Cai
Xiao Yan

Assignees

Makesense Energy Technology Co., Limited.

Dates

Publication Date: 20260512
Application Date: 20230920
Priority Date: 20220926

Claims (10)

1 . A thermal-runaway warning method for a power station, comprising: obtaining collected data from a battery module comprised in the power station, wherein the collected data comprises temperature values of the battery module acquired by a plurality of temperature measuring devices; normalizing the collected data to obtain a standard dataset; training a convolutional neural network and long-short term memory (CNNLSTM)-based temperature predicting model based on the standard dataset to obtain a trained CNNLSTM-based temperature predicting model; and obtaining a predicted temperature value of the battery module within an output time window based on the trained CNNLSTM-based temperature predicting model, and determining whether there is a risk of thermal runaway in the power station based on the predicted temperature value.
2 . The thermal-runaway warning method for the power station according to claim 1 , wherein training the CNNLSTM-based temperature predicting model based on the standard dataset comprises: constructing the CNNLSTM-based temperature predicting model, wherein the CNNLSTM-based temperature predicting model comprises a CNN layer, an LSTM layer, and a fully connected layer, connected in sequence; intercepting the standard dataset using a sliding window approach to obtain a standard sub-dataset within an input time window; obtaining a sub-group of the collected temperature values within the output time window corresponding to the standard sub-dataset obtained within the input time window; and training the CNNLSTM-based temperature predicting model based on the standard sub-dataset and the sub-group of the collected temperature values to enable the CNNLSTM-based temperature predicting model to obtain the predicted temperature value within the output time window based on the standard sub-dataset obtained within the input time window.
3 . The thermal-runaway warning method for the power station according to claim 2 , wherein the CNN layer comprises a first convolutional module and at least one second convolutional module connected in sequence, wherein the first convolutional module comprises a first one-dimensional convolutional layer and a first activation layer connected in series, wherein the at least one second convolutional module comprises a second one-dimensional convolutional layer, a batch normalization layer, and a second activation layer connected in series.
4 . The thermal-runaway warning method for the power station according to claim 1 , wherein determining whether there is the risk of thermal runaway in the power station based on the predicted temperature value comprises: calculating a reconstruction error of the predicted temperature value; and when the reconstruction error is greater than a predetermined error threshold, determining that there is the risk of thermal runaway in the power station.
5 . The thermal-runaway warning method for the power station according to claim 4 , wherein calculating the reconstruction error of the predicted temperature value comprises: encoding the predicted temperature value to obtain an encoded temperature value; decoding the encoded temperature value to obtain a decoded temperature value; and calculating a difference between the predicted temperature value and the decoded temperature value, wherein the difference is the reconstruction error of the predicted temperature value.
6 . The thermal-runaway warning method for the power station according to claim 1 , wherein determining whether there is the risk of thermal runaway in the power station based on the predicted temperature value comprises: calculating a deviation value between the predicted temperature value and a corresponding one of the collected temperature values; and when the deviation value is greater than a predetermined deviation threshold, concluding that there is the risk of thermal runaway in the power station.
7 . The thermal-runaway warning method for the power station according to claim 1 , further comprising: issuing a warning message when it is concluded that there is the risk of thermal runaway in the power station.
8 . A thermal-runaway warning system for a power station, comprising an acquisition module, a normalization module, a training module, and a judgment module; wherein the acquisition module obtains collected data of a battery module comprised in the power station, wherein the collected data comprises temperature values of the battery module acquired by a plurality of temperature measuring devices; wherein the normalization module normalizes the collected data to obtain a standard dataset; wherein the training module trains a CNNLSTM-based temperature predicting model based on the standard dataset to obtain a trained CNNLSTM-based temperature predicting model; and wherein the judgment module obtains a predicted temperature value of the battery module within an output time window based on the trained CNNLSTM-based temperature predicting model, and determining whether there is a risk of thermal runaway in the power station based on the predicted temperature value.
9 . A non-transitory storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the thermal-runaway warning method for the power station according to claim 1 .
10 . A terminal for communicating a thermal-runaway warning message for a power station, comprising a processor and a memory; wherein the memory is configured to store a computer program; wherein the processor is for executing the computer program to cause the thermal-runaway warning message communicated on the terminal for the power station to implement the thermal-runaway warning method for the power station according to claim 1 .

Description

CROSS REFERENCE TO RELATED APPLICATION The present application claims the benefit of priority to Chinese Patent Application No. CN 202211174980.8, entitled “THERMAL-RUNAWAY WARNING METHOD, SYSTEM, AND TERMINAL FOR POWER STATION”, filed with CNIPA on Sep. 26, 2022, the disclosure of which is incorporated herein by reference in its entirety for all purposes. FIELD OF THE INVENTION The present disclosure generally relates to batteries used in power stations, and in particular to a thermal-runaway warning method, system, and terminal of an energy-storage power station. BACKGROUND OF THE INVENTION Energy-storage power stations play a crucial role in the development of new power systems. However, due to reasons such as inadequate safety measures, these stations are prone to fires, causing significant economic damage to power companies and posing a serious threat to people's lives and power companies' properties. Various factors, both internal and external to the batteries, can lead to thermal runaway in energy-storage power stations. As a result, monitoring the temperature of these power stations is essential for ensuring their safety. In current technology, the energy storage structures of large-scale power stations are composed of battery modules as stacks-clusters-boxes-cells. Energy-storage power stations collect temperature data from multiple probes within each battery box to monitor temperature changes in the battery module in real-time. This allows power stations to determine if thermal runaway has occurred based on the temperature changes and thermal abuse standards of the battery modules. Thermal runaway can escalate to a complete loss of battery control, fire or explosion in less than ten minutes, making early warning of any abnormalities crucial for safety monitoring. However, current technology has limitations, such as a single battery module is equipped with multiple temperature probes, which makes temperature monitoring based on a single point data unreliable and prone to false alarms. SUMMARY OF THE INVENTION The present disclosure provides a thermal-runaway warning method, system, and terminal for a power station, which enable effective warning of thermal runaway for a power station by predicting the temperature of battery modules in the power station. The thermal-runaway warning method for a power station comprises: obtaining collected data from a battery module in the power station, wherein the collected data comprises temperature values of the battery module acquired by a plurality of temperature measuring devices; normalizing the collected data to obtain a standard dataset; training a combined model of convolutional neural network (CNN) and long-short term memory (LSTM), or for short, CNNLSTM-based temperature predicting model from the standard dataset to generate a trained CNNLSTM-based temperature predicting model; and calculating a predicted temperature value of the battery module within an output time window given by the trained CNNLSTM-based temperature predicting model, and determining whether there is a risk of thermal runaway in the power station based on the predicted temperature value of the battery module. In one embodiment of the present disclosure, training the CNNLSTM-based temperature predicting model from the standard dataset comprises: constructing the CNNLSTM-based temperature predicting model by combining a CNN layer, an LSTM layer, and fully connected layers, wherein the CNN layer, LSTM layer, and fully connected layers are connected in sequence; intercepting the standard dataset using a sliding window approach to obtain a standard sub-dataset associated with an input time window; and obtaining a sub-group of the collected temperature values within the output time window corresponding to the standard sub-dataset; and training the CNNLSTM-based temperature predicting model based on the standard sub-dataset and the associated sub-group of the collected temperature values to enable the CNNLSTM-based temperature predicting model to obtain the predicted temperature value within the output time window associated with the standard sub-dataset within the input time window. In one embodiment of the present disclosure, the CNN layer comprises a first convolutional module and at least one second convolutional module connected in sequence, wherein the first convolutional module comprises a first one-dimensional convolutional layer and a first activation layer connected in series, wherein the second convolutional module comprises a second one-dimensional convolutional layer, a batch normalization layer, and a second activation layer connected in series. In one embodiment of the present disclosure, determining whether there is a risk of thermal runaway in the power station based on the predicted temperature value of the battery module comprises: calculating a reconstruction error of the predicted temperature value; and if the reconstruction error is greater than a predetermined error thr