KR-20260063340-A - Battery system for AI-based battery fault diagnosis and remaining useful life prediction

Abstract

The present invention relates to a battery system for AI-based battery fault diagnosis and remaining life prediction, and more specifically, to a battery system for AI-based battery fault diagnosis and remaining life prediction capable of collecting battery status information and diagnosing battery faults and predicting remaining life using an AI-based analysis model. The battery system for AI-based battery fault diagnosis and remaining life prediction according to the present invention can diagnose faults and predict and estimate the lifespan of battery cells and battery packs through artificial intelligence big data analysis of battery status information, thereby preventing accidents caused by batteries and increasing battery utilization efficiency. The battery system for AI-based battery fault diagnosis and remaining life prediction according to the present invention can extend the lifespan of battery cells and battery packs and optimize performance by continuously monitoring and managing the status of battery cells and battery packs, thereby promoting the growth of battery systems and the development of battery technology. This patent was obtained with the support of the Regional Key Industry Development (Regional Specialization Project Support Program).

Inventors

• 박수련
• 황선호
• 윤진형
• 왕다빈

Assignees

• 주식회사 현대쏠라텍

Dates

Publication Date

20260507

Application Date

20241030

Claims (5)

1. A battery pack comprising a plurality of lithium iron phosphate battery cells formed in a cylindrical shape, a holder unit having a plurality of individual storage spaces for storing each of the battery cells individually, and a connection unit installed in the holder unit to connect the battery cells stored in the holder unit in a series and parallel combination; A sensor unit for detecting individual state information including voltage, current, and temperature of each of the above battery cells; A battery management unit comprising: a data storage unit that receives individual state information detected by the sensor unit, converts the received individual state information into individual state data and stores it, and stores charging history data and discharging history data for each of the battery cells that have previously undergone the process; a fault diagnosis unit that diagnoses whether there is a fault in each of the battery cells through an AI-based fault diagnosis model that has been machine-learned in advance to diagnose whether there is a fault in each of the battery cells using the individual state data and history data stored in the data storage unit; and a remaining life prediction unit that predicts the remaining life of each of the battery cells through an AI-based remaining life prediction model that has been machine-learned in advance to predict the remaining life of each of the battery cells using the individual state data and history data stored in the data storage unit; A switch unit installed on a power cable connected to the battery pack to supply power from the battery pack to a load or to supply external power to the battery pack, and electrically opening and closing the power cable; A battery system for AI-based battery fault diagnosis and remaining life prediction, characterized by comprising: a controller that receives a fault diagnosis result from the battery management unit and controls the opening and closing operation of the switch unit according to the received fault diagnosis result.
2. In paragraph 1, On both sides of the individual storage space facing both longitudinal sides of the battery cell, a first connecting electrode and a second connecting electrode are formed, respectively, to electrically contact the positive and negative electrodes of the battery cell. A battery system for AI-based battery fault diagnosis and remaining life prediction, further comprising: a connection separation unit installed inside the individual storage space facing the circumferential surface of the battery cell, which separates the battery cell from at least one of the first connection electrode and the second connection electrode when the temperature of the battery cell reaches a set overheating temperature.
3. In paragraph 2, A battery system for AI-based battery fault diagnosis and remaining life prediction, characterized in that the connection separation part is installed inside the individual storage space in contact with the circumferential surface of the battery cell and includes an overheat separation part formed of a bimetal material such that when the temperature of the battery cell reaches a set overheating temperature, an interference force is generated to push the battery cell out from the individual space part.
4. In paragraph 2, A battery system for AI-based battery fault diagnosis and remaining life prediction, characterized in that the connection separation part is installed inside the individual storage space in contact with the circumferential surface of the battery cell and includes an overheat separation part formed of a shape memory alloy material such that when the temperature of the battery cell reaches a set overheating temperature, an interference force is generated to push the battery cell out from the individual space part.
5. In either Paragraph 3 or Paragraph 5, The sensor unit further includes a microphone capable of measuring a separation sound generated when the battery cell is separated from at least one of the first connection electrode or the second connection electrode by the connection separation unit. A battery system for AI-based battery fault diagnosis and remaining life prediction, characterized in that when a separation sound is measured in the sensor unit, the controller outputs separation information of the battery cell to the display unit or transmits it to a configured user terminal.

Description

Battery system for AI-based battery fault diagnosis and remaining useful life prediction The present invention relates to a battery system for AI-based battery fault diagnosis and remaining life prediction, and more specifically, to a battery system for AI-based battery fault diagnosis and remaining life prediction capable of collecting battery status information and diagnosing battery faults and predicting remaining life using an AI-based analysis model. Recently, as interest in environmental protection has increased, electric vehicles (EV, HEV, PHEV) are gaining attention, and smart grid businesses utilizing energy storage systems (ESS) are attracting interest due to the expansion of renewable energy use and the increase in electricity demand. In particular, with the expansion of electric vehicles, interest in the recycling of secondary batteries generated when electric vehicles are scrapped is increasing explosively, and in order to reuse these batteries that have been used more than once for other purposes, technology is required to more easily estimate the lifespan of unspecified batteries. Fundamentally, batteries age as they undergo repeated charging and discharging cycles, and this aging process can be accelerated by various other factors, such as battery temperature, charging method, current variations, and depth of discharge. For a battery management system to optimally manage and operate batteries, it is necessary to accurately estimate their lifespan based on aging. In particular, the battery management system must be able to accurately predict the battery's State of Charge (SOC) and State of Health (SOH). Conventional battery life estimation methods involve counting the number of charge-discharge cycles to predict the degree of aging and lifespan of the battery in comparison to the number of charge-discharge cycles guaranteed by the battery manufacturer. However, batteries used in actual machinery do not end in clear charge-discharge cycles of complete charging and complete discharging; instead, they may discharge while partially charged or recharge after partially discharging, making it impossible to accurately count the number of charge-discharge cycles. Consequently, there are limitations in accurately estimating the degree of aging and lifespan of a battery based on the number of charge-discharge cycles. As another conventional method for estimating battery life, Korean Registered Patent No. 10-0740113 discloses "a method for determining battery life and a battery management system using the same." The aforementioned "method for determining battery life and a battery management system using the same" measures the pack current and pack voltage for a battery pack formed by multiple battery cells into a single pack to calculate the pack internal resistance, calculates the maximum output of the battery pack using the calculated pack internal resistance, and estimates the battery life by comparing the calculated maximum output with a reference output. As another conventional method for estimating battery life, Korean Published Patent No. 10-2012-0075756 discloses "Method and apparatus for calculating remaining life of a secondary battery." The aforementioned "Method and apparatus for calculating remaining life of a secondary battery" describes a technology for calculating the battery life using parameters such as the battery's internal resistance, state of charge (SOC), and conductance. However, the internal resistance calculated based on battery current and voltage consists of actual internal resistance and contact resistance; since contact resistance depends on the contact condition between the battery and the wiring rather than battery degradation, it is very difficult to accurately calculate the internal resistance resulting from battery degradation based solely on current and voltage. Furthermore, it is very difficult to calculate remaining capacity or accurately estimate battery lifespan based on internal resistance. Therefore, there is a need for a method to predict and estimate battery life that is more accurate and reliable than conventional battery life estimation methods. FIG. 1 is a block diagram schematically illustrating a battery system for AI-based battery fault diagnosis and remaining life prediction according to the present invention. FIG. 2 is a perspective view showing a battery pack of a battery system for AI-based battery fault diagnosis and remaining life prediction according to the present invention. FIGS. 3 and 4 are cross-sectional views showing the operating state of the first overheating separation unit of a battery system for AI-based battery fault diagnosis and remaining life prediction according to the present invention. FIGS. 5 and 6 are cross-sectional views showing the operating state of a second overheating separation unit of a battery system for AI-based battery fault diagnosis and remaining life prediction according to the present invention. FIGS. 7 and 8 are