Coming soon! Our webinar just ended. Check back soon to watch the video.
How Gotion Monitors its EV Battery Solution with InfluxDB, Grafana and AWS
Session date: 2021-11-02 08:00:00 (Pacific Time)
Meet Gotion — a leading EV battery and battery management system. Car manufacturers use Gotion’s platform to extract sensor data from electric car batteries, which is used to enable predictive maintenance and provide a better customer experience. Hear how they use a time series database to collect and cleanse their data to find outliers. By analyzing time-stamped data, they determine if the anomalies are cause for concern or if they can be ignored and scrubbed.
Join this InfluxData webinar as Tony Li dives into:
- Gotion’s approach to IIoT monitoring (battery life, health, temperate and voltage)
- Their ability to provide better forecasting and machine learning models
- How InfluxDB and Python are used to extract data from batteries, cars, testing facilities, manufacturing plants and simulation centers
Watch the Webinar
Watch the webinar “How Gotion Monitors its EV Battery Solution with InfluxDB, Grafana and AWS” by filling out the form and clicking on the Watch Webinar button on the right. This will open the recording.
Here is an unedited transcript of the webinar “How Gotion Monitors its EV Battery Solution with InfluxDB, Grafana and AWS”. This is provided for those who prefer to read than watch the webinar. Please note that the transcript is raw. We apologize for any transcribing errors.
- Caitlin Croft: Customer Marketing Manager, InfluxData
- Tony Li, Ph.D.: Manager of Software and Data, Gotion
Caitlin Croft: 00:00:02.285 Hello, everyone, and welcome to today’s webinar. My name is Caitlin Croft. I’m very excited to have Tony Li from Gotion here to talk about how Gotion is using InfluxDB, Grafana, and AWS as part of their monitoring solution. What’s really cool is they’re monitoring electric vehicle batteries, so I know I’m super excited to learn more about this, and I’m really excited to see you all here today. Just a couple of friendly reminders. Please post any questions you have for Tony in the Q&A or the chat. We will be monitoring both, and we’ll answer questions at the end. This session will be recorded and will be made available later today or tomorrow morning, so the slides as well as the recording of the webinar. We also have tons of events throughout the month, so please be sure to check out the event page on influxdata.com. There’s tons of fantastic events. Next week we have the monthly InfluxDB IOx Tech Talks. Last week we had InfluxDays, so all the recordings will be made available for that shortly. And also, if you ever need help with InfluxDB, please be sure to check out the forums and the community Slack channel. There’s tons of people out there who are willing to help. All right. I think without further ado, I’m going to hand things off to Tony. Once again, welcome everyone to today’s webinar. My name is Caitlin Croft, and I’m going to hand things off to Tony.
Tony Li: 00:01:41.416 Thanks, Caitlin. Really appreciate it. So hello, guys. My name is Xiaojun Li, or you can call me Tony. I’m a software manager at Gotion here. So today’s topic with InfluxDB, we talk about how Gotion use InfluxDB to build up our EV battery data solution. This mainly involves InfluxDB, Grafana, and AWS. So I’ll go ahead to the next slides. Summary for today’s presentation. At first, we talk about the battery data problem we have here at Gotion. Secondly, we jump into the reason why we choose InfluxDB as our database. In the later part of this presentation, we talk about two applications. One is the dashboard we built for our user to view their data. Secondly is, how do we use the database to develop our algorithms? Then the last part, we talk about the future [of work?], our vision, yeah, for the future of our database.
Tony Li: 00:02:54.214 So let’s jump into the first part. We talk a little bit about Gotion, the company itself. So Gotion is a global company. We’re mainly focused on battery technologies. We have offices in California, Ohio, Shanghai, Singapore, and Japan. Different locations actually focus on different aspects for battery technologies. For example, in California, we focus on the product so called BMS. I will talk about this quite a lot in the future slides. BMS stands for battery management system. Basically, it’s a piece of hardware and software that goes onto an electrical vehicle or a [inaudible] system that use batteries. So if you’re driving a Tesla or a GM Volt, for example, there would be a BMS inside the car that’s monitoring that battery status. It’s also telling you how long the battery can last, how many miles do you have left? So this is a key part of a modern EV, electrical vehicle. So here in California, like I mentioned, we focus on the BMS electronic software. Then the other locations — for example, Ohio, they focus on the chemistry parts. Similar goes for Japan and Singapore.
Tony Li: 00:04:21.292 Okay. I will stop here talking about the company. Then next I will talk about the battery data problem with face here at Gotion. So like I mentioned before, we have the battery management system [as?] our product we sell to the customers, but mostly our EV manufacturers. We do get a lot of data. So normally the battery management systems will monitor all the battery cells. The batteries we have, basically they are made up of three levels. Cell is the basic level that’s similar to what you have at home. For example, one type of the battery cells is called the cylindrical cell, that’s very close, maybe a little bit larger, but close to the regular AAA cells we have at home. And these cells were made into a module where they will have a case and all the thermal components, basically monitoring stuff, embedded into the module. And the customer will basically ask us to make it to a pack. So from the cell to pack, we’re talking about hundreds of cells, possibly, and the BMS is monitoring all the information for the cells. So these including hundreds of battery voltage, maybe a dozen sensor for temperatures, and at least one current sensor. So we are getting these data from different sources, like testing center, where you have the battery pack being tested or even the cell being tested. Then also, you have vehicle data that’s coming from the automotive testing or just customers online. Yeah. Remote sensing.
Tony Li: 00:06:15.118 Then we also have the energy storage part that’s also [use the?] BMS. It’s monitoring even more cells, more battery packs. We’re talking about gigawatt hours of battery that’s being deployed. So data really come into different data formats. For example, the testing center normally use their dedicated test software that generates either a CSV or text file. For vehicle side, there could be more data. They use, for example, a standard automotive test data source format called [MPF?]. And also, they include a little bit more data compared to other storage. For example, there will be vehicle locations. Normally they GPS information or speeds. And it could be other information like charger, motor. So we try to have a uniform place to store this data and to leverage our algorithms and analytics based on these data. Most of these data are time series.
Tony Li: 00:07:25.227 Okay. Let’s go to the next slides. So here is the reason we choose InfluxDB as our database. First of all, it’s dedicated time series database, so it is number one in the market. So we do have a high confidence it will be continuously updated in the future. Secondly, it does have a very SQL-like [coding?] language, which is friendly to use. Most our data scientist or data engineer, they’re very familiar with the SQL and it’s easy to jump into InfluxDB. And of course, it has very native integration with Grafana, which is our main front-end tool. Also, this is very important for us. It has a native Python API developed by the InfluxData team, I believe. And also we find that there’s also MATLAB API. These two are the main algorithm [development?] environments that we use here at Gotion, so supporting these will help us with the future algorithm development. And of course, it is very easy to implement on a cloud platform such as AWS.
Tony Li: 00:08:42.550 So previously, we have evaluated a different time series database, for example, Graphite. We found that [when it’s?] compared to Graphite, InfluxDB is relatively easy to get started with and has better API support, for example, it has a MATLAB API. So on the right side, you can see is Gotion’s data platform solution. So we have different data sources. They’re coming into the time series database, which is InfluxDB. Then, we also gathering that metadata. For example, the testing locations, what kind of chemistry this battery is using, the tester name, whether it’s a cell or a pack. All these information, they will be storing them at metadatabase which is used in SQL. So it captures the relations between these times series. On top of that, we developed our algorithms this way, mostly involve with — right now we working on anomaly detections. So I will talk a little bit about this part later in the later slides, including the thermal anomaly detections we have developed at Gotion. In the future, we plan to do the battery SOH. SOH stands for state of health. It basically tells you how many years left for your battery there. And also, we plan to do the battery quality management. This is about tracing, tracking the lifecycle data of a battery cell so that we can better match to build a pack, for example, and the battery to recycle the batteries based on the usage they have. So of course, on top of that, we’re going to build the data visualization. We’ll talk a little bit about the dashboard in the later slides. And we plan to do the battery data analytics and the [inaudible] as well.
Tony Li: 00:10:42.888 Okay, so let’s go to the next slides. Okay. Yeah. So here we talk about the applications. The first one I will mention is the battery dashboard with [inaudible] here at Gotion. Secondly, we talk about one algorithm we developed, which is the thermal anomaly detection. So let’s jump into the battery dashboard. We call it battery volts. So this is a one stop battery database. Here the components we used for this application are Grafana, which is using AWS EC2. And we also have the InfluxDB storing in an AWS EC2. Then the relational database we use is a PostgreSQL. That’s using AWS RDS. So as I interact [with the?] dashboard, the users basically be able to first curate the database based on various criteria filters, so for example, based on the chemistry type, the capacity of the battery, and what kind of testing method we have. Then from there, they can directly view the time series. In the next slides, I’m going to talk about how it looks like for the dashboard. So basically, you can see here—
let me see if I can turn on the eraser here. Yeah. Okay. So see here we can select the parameter [in?] selection box based on the chemistry type, cell capacity or pack capacity, and who has uploaded data.
Tony Li: 00:12:26.375 So in the main table we have here is all the time series being filtered out. And you may notice in the very right column, we have the link, which will let you click — if you click it, it will lead you to the times series data, actually. So this is just one table. We also have the different tables for cells and packs. Yeah. So here we can see if you click to the link here, this will bring you to the times series. Basically, the user will be able to select other fields they want to view, for example, if we want to see the current temperature of [all?] [inaudible] products. And from here, we’re just basically using Grafana’s time series panel. So a few simple tips on how we made this dashboard. First, how to make a kind of interactive dashboard. At Grafana, we use the Grafana variables and InfluxQL a lot. So first, use the SQL in Grafana’s dashboard to create some variables. So, for example, actually using InfluxQL, for example, we created these variable like [inaudible] Select a bucket, Select fields. And the users will be able to — using a dropdown box, like I mentioned earlier, like I showed earlier, to select the data they want. Then the last [long?] query basically is listed here at the bottom. That’s actually display the times series based on these selections. So this is a very simple example.
Tony Li: 00:14:18.390 Next, I will talk about a little bit interesting parts that we use the Grafana dashboard to support second language display. So as I mentioned earlier, we have different office at different locations. For example, the office in Shanghai, they would prefer in Chinese if the dashboard can be displayed in Chinese. So we have developed a little simple system that allows users to view the different language interface. So how it works is if you look at the right side, we have a Postgres table that’s basically mapping all the English name with the Chinese names. Now on the Grafana part of — first of all, [inaudible] query, the existing selected time series, like I showed here, with a correspondent database and a series, though we got the existing field keys. And using this field keys as a variable, we do a second query. Actually it’s a SQL query that search for the matching Chinese translations in the table I mentioned. So this is using Grafana advanced [select?] syntax, which they allow users basically to select one column as the value, the other column as the display text. So here we’re using the second language as the display text. The first language, which is the English key in this case, as the value. So after user select the text in Chinese, the actually returns results from this query, which is another Grafana variable, with English. And this results will be used again to do the final query for the time series. This is how we achieve translation, basically, without much effort.
Tony Li: 00:16:20.699 Okay. Let’s go to the next slides. So here I will talk about one example for the algorithm [inaudible] outputs. It’s called thermal anomaly detection. So a little bit of background information. Here we’re looking at one phenomenon called a thermal runaway. So I don’t know if you have heard of — are following the news, these events happens quite a lot on different OEMs, especially at the early stage of their EVs. It happens on Tesla. It happens on GM’s Volts or even on Toyota’s Prius. Basically, this is a very critical issue for batteries. People have concerns about batteries all of a sudden catch on fire. The real mechanism behind this events, they’re complicated. It could be caused by, for example, mechanical issue, most likely a manufacturer issue in the cells, or it could be electrical issue, a short, or even a thermal issue. For example, if the temperature is too high — if you park your car under the Texas Sun for six to eight hours, it could leads to this problem. And when the thermal runaway happens, normally it’s very difficult to be put out by a regular, let’s say, fire extinguisher. Yeah, it’s basically what cause fire explosions. And the only solution for the customer is to just run away. So it’s very critical if we can somehow give an early warning or predict if this event’s going to happen.
Tony Li: 00:18:01.323 So the difficulties in predicting are similar in a way. Basically, if you look at the data, we have, data from database, these data are collected from BMS, battery management system, to our car database. So they usually have these problems. For example, you have intermittent invalid data. You have long-term data loss. This is usually caused by people turning off their cars or the wireless transmission module no longer working properly. So you don’t know the real cause for it. So with this issues of the data we got, it’s very difficult to, let’s say, implement a rule-based or a manual-based method. We come with here is a [inaudible] method. This is where we utilize the InfluxDB to store our data, with a large data set so we be able to apply our data to a method. So let me go to the next slide. Yeah. So this part is a little bit complicated. I will not go to the details about the algorithm itself. There is a paper published if you guys are interested. Please click on the slides, which I believe will be shared later. This paper is listed here.
Tony Li: 00:19:21.975 So general workflow for this algorithm is we have a [inaudible] machine that’s gathering data from the database. These data are curating the segments. For example, every five minutes, we will try to curate some data. These data will be temperature measurements on the car. Then after processing these data, all of these temperature measurements will be clustered based on their shapes. For example, let’s say you have 10 temperature measurements on your car, and you’re seeing one of these measurements starting to behaving differently in terms of shape compared to the other measurements. In this case, we compare it to the previous cluster. We find out the anomaly. Then we starting to be more certain that this single measurement is actually behaving different than the others, which could be indicating [some more?] anomaly events. So here are just the basic information about how this algorithm works.
Tony Li: 00:20:37.178 Then next, I will talk about how we implement it with InfluxDB and the other tools. So in this application, we still using Grafana as our front-end. The user basically would be able to see how these anomaly detections are marked on the time series. And also, they will be able to trigger a algorithm to analyze certain section of the time series. For example, they can select one car from throughout the whole year and try to analyze it, see if the algorithm can detect some anomaly. So I will show example in the later slides. Of course, InfluxDB is used to store the times series using this application. Then we have the Airflow actually using as our pipeline tool that basically Airflow will receive the commands from Grafana and it will curate data from InfluxDB. And [inaudible] algorithms [gather?] results. Later, the Airflow pipeline machine will return the results to InfluxDB as a time series, as well. And during this process, all the metadata will be recorded in a PostgreSQL database. For example, when is the last time we actually analyzed the time series? How many times series has been analyzed? How many are there? These metadata will be stored in PostgreSQL. Yeah.
Tony Li: 00:22:10.971 So here is an example of one test case, basically, from a EV. So this electric vehicle is using our Gotion BMS. As you can see here, we have about 13 temperature measurements. I believe this is — the data is recording 2019. So around October, late October 2019, when one of the temperature measurements rise to really high value, it’s actually exceeding the typical threshold for this type of battery, we call the [FP?] lithium [inaudible]. And they have a threshold of 65 degrees over that. So [we indicate?] some system failure. So here, let’s see how the algorithm behaves. So with the algorithm applied to this time series, what we found is it’s able to return the analytical results and send a warning around 2:15 PM. So if you look at the graph on the left side here, the little blue line is where the algorithm was able to detect the anomaly as it’s indicated here. And this is way before the real temperature sensor go to very high value. So for the BMS, the onboard BMS, to detect anomalies are actually around 3:45 PM. So in this particular test case, we can see here this new algorithm was able to detect the thermal anomaly by one and a half an hour earlier. Again, this is using other tools we gathered here, especially InfluxDB as the main [inaudible] database.
Tony Li: 00:24:08.917 Okay. Next, this will be the last slides for my presentation. So what is the future step for Gotion’s data platform? Our vision is that in the future, the batteries will be more available to EVs to ENERGY STAR System. They will be used in lot of cases for renewable energies. But the battery manufacturing process is very complicated. So you’re starting from the cells to the modules to the pack. Different teams mostly are involved. So cells may mostly be made by the battery manufacturers. Modules, that will be module designers. This anomaly could involve the OEMs, I mean, the car manufacturers. And in the end, there will be a pack assembly that’s usually done by both parties, so that if we can trace or track the bearing information, the data, along these lines, we’ll be able to better facilitate our, for example, algorithm developments. This involves, for example, state of charge you will see on vehicles, on the EV, for example, how many miles it has left, how many battery charge you have left. So with more data, we will more accurate [inaudible] algorithms and [for controls?] so this will involve with the similar runaway events. With more data, we will get more accurate algorithms to predict when these faulty events will happen and provide safer battery to the customers.
Tony Li: 00:25:51.382 And then the last parts, and not least, is the quality management. For example, one of the applications using the power data is to track all the cell’s lifecycle data. So as you know, the battery manufacturing process is not very environmental friendly, right? Extracting lithium, it may cause a lot of pollutions. So what we can do is track the battery data for a single cell and try to recycle it. It’s based on the usage of millions of cells. So cells with a similar usage will be remaking to battery pack and used as a secondary life. So, for example, if you have a EV battery that is being recycled, you can use for ENERGY STAR System. So in order to facilitate this process, it’s crucial to store this data. Yeah. So at the center of this operation, what we see here is the battery database to store these time series, which is very crucial for us. Yeah, that’s all my slides here. I’m open to the question now. Thank you.
Caitlin Croft: 00:27:01.411 Thank you, Tony. That was great. So the first question I have for you is, are you guys considering using Flux?
Tony Li: 00:27:12.571 Can you say that again?
Caitlin Croft: 00:27:14.191 Are you considering using Flux? I know you mentioned you’re using InfluxQL.
Tony Li: 00:27:19.649 Yes. Right now we are using InfluxDB 1.8 database. So we’ll be upgrading to 2.0 and eventually we’re using Flux.
Caitlin Croft: 00:27:30.853 Okay, cool. So you are you are looking at upgrading to 2.0 and looking at Flux. Yeah. The reason why I ask is I could see Flux being useful when you’re aggregating the anomaly detection data. So you could even join the data and down sample and all that. How did you set up the threshold? Is it subjective?
Tony Li: 00:27:58.122 So you’re talking about, I believe, the [inaudible]. Let me share the screen here. [These are more?] anomaly detection. Maybe it’s here. Yeah. So the threshold actually is not set by our team. So this usually come from the battery cell designers. They are [chemists?]. They were decide what the threshold for the — they will do a bunch of testing to push the battery to the limits, and they find out this temperature of the threshold for this battery. But the algorithm developed here, we are not actually using threshold. As you can see, the algorithm actually detect the anomaly not much early than before it reached the threshold. Yeah.
Caitlin Croft: 00:28:50.302 And how are you setting up alerting? So you have the threshold. It hit it. The anomaly was detected just above 20 degrees. How are you getting the alerts and who’s getting the alerts?
Tony Li: 00:29:03.611 Yeah. So this algorithm basically looking at the shape of the signal. So at 25 degrees, you can see here one of the green line here is already behaving — like the shape of this line is already very different from the others. That’s where the algorithm can detect the anomaly. Yeah. So the alerts, it’s now [impending?]. But our plan is that obviously you can go [to?] Grafana, take a look the results there. The plan is that we would be able to set up a mobile text warning. When these alerts happen, it basically send out alerts to our test engineers or who is in charge of the data center. Yeah.
Caitlin Croft: 00:29:52.484 Great. How do you pull the data timely and effectively for all of your batteries or all the battery systems, such as ESS?
Tony Li: 00:30:04.382 It’s a good question. So a little bit different — I believe a little bit different application — for different applications we have different method. For example, for the testing data, we do have a manual upload center that you can just upload a CVS file and text file. You have to obviously set up which database you want to upload it to [or the?] time series name. But that’s the other good thing about InfluxDB is that it automatically aggregate the data for you. For example, you can upload partial — for the same cell, you can upload tomorrow’s — I’m sorry. Yesterday’s data or data two months ago, that would be basically aggregated in the same time series. Then for the other applications, for example, EVs and ENERGY STAR System, we do have to work with our customers. So I won’t be able to [start?] and talk about the details because there’s part that’s concerning our customers. Basically, the customers will have a wireless transmission module in their system. The BMS will send this information to that module. They will control this module and they will send to, for example, like a gateway or a local gateway. That gateway will actually send the data to us.
Caitlin Croft: 00:31:28.768 Great. Do your batteries have an independent capability to send data back, or they dependent on the connectivity of the product they are implemented, such as the electric vehicle?
Tony Li: 00:31:41.787 Great question. Yes, they depends on the connectivity of the products they’re in, like electric vehicle. This is mostly due to regulations, though. Our customers, they won’t like — all of the customer, they like to control their data, how the data go out, how the data collected. So they will have a different module that transmit the data wirelessly.
Caitlin Croft: 00:32:06.687 And how many metrics have you collected in InfluxDB? Is it millions or billions?
Tony Li: 00:32:15.132 We haven’t actually looked at this yet. I would say the scale right now is not large, maybe millions. Yeah.
Caitlin Croft: 00:32:23.534 Okay. Yeah. And I’m assuming you’re looking — all this data is being used to improve the longevity of the batteries, the lifespan of the batteries, as well as the battery lifespan itself before it needs charging, right?
Tony Li: 00:32:37.196 Mm-hmm.
Caitlin Croft: 00:32:38.460 Yeah. And how often are you collecting data about the batteries? Is it every minute, every five seconds?
Tony Li: 00:32:49.377 Actually, it depends on the type of the data, like you mentioned earlier. So if we go back to the very original slides, what we collecting here — yeah. So we have the battery voltage, current, and temperature. The battery voltage we collect every second, I believe. But for current, it’s much higher. We’re talking about 200 milliseconds. Temperature, a little bit lower. Yeah. We’re talking about maybe five seconds. So these are basically based on what — these data collected based on their needs. So temperature, it tend to change slower over time compared to the voltage. That’s why [inaudible] lower. Yeah.
Caitlin Croft: 00:33:31.508 How do you get your battery data into InfluxDB? Is this process kicked off manually or automatically?
Tony Li: 00:33:39.639 The start of this process actually mostly going to be kicked off manually. This is, again, due to the customer’s requirements [inaudible] [won’t?] be able to automatically fetching their data. Yeah.
Caitlin Croft: 00:33:54.974 Awesome. Well, thank you, everyone. If anyone has any more questions for Tony, please feel free to post it. I feel like I could probably ask you a ton of questions. I feel like electric cars are obviously becoming more and more popular. So I’m sure that understanding the lifespan of the batteries is going to be even more important as more and more people buy these kind of cars. What was something really kind of fun or interesting that you found from the data that you guys have been collecting?
Tony Li: 00:34:35.812 [Tell?] again?
Caitlin Croft: 00:34:37.773 What was something kind of fun or interesting that you learned about the car batteries that you maybe weren’t expecting?
Tony Li: 00:34:48.883 The [inaudible] we find interesting is there are a lot of information you can get from data, actually, because they also recording GPS locations. You can basically recreate this user, how he — because if I go a grocery store, recreate, yeah, how this user drive, like where he’s driving, where he stopped. So all these engineers were thinking, ‘Wow, this is really powerful. You can know what’s the habit of this person, everything like that.” That’s the part that I think was very interesting to us when we first look at the data.
Caitlin Croft: 00:35:31.942 Yeah, that’s interesting. So it’s not only enabling predictive [inaudible] of the battery, but also understanding the consumer, as well, better.
Tony Li: 00:35:41.072 Right, the behavior. We think there’s a potential to analyze, for example, the individual’s behavior. How can we better change our algorithm to fit, yeah, their needs. For example, if a person is mostly — it’s quite different from a regular using the EV to do the grocery shopping, which is very light use and it’s better for the battery’s health. But if you are a taxi driver, you’re driving this car very frequently and never fully charge it. Then the battery health is going to be harmed a lot. Yeah. So this information would be very crucial once we try to recycle the battery.
Caitlin Croft: 00:36:22.315 Yep. Are you using Telegraf as the gateway between the vehicle and InfluxDB? So I guess the question is how are you getting data from the vehicles to InfluxDB?
Tony Li: 00:36:37.753 Yeah, right now we’re not using Telegraf.
Caitlin Croft: 00:36:41.895 So do you mind kind of going into, again, a little bit more of how you are getting the data into InfluxDB?
Tony Li: 00:36:48.643 Yeah. The other part — I’m not be able to talk about since it concerns our customers. What I can mention is that the manual part that we basically, again, using Airflow. So you can basically — if you have a CVS file or a text file, we have a web page that you can upload it and that will kick off a pipeline in Airflow, which [inaudible] the data and upload and aggregate to the times series.
Caitlin Croft: 00:37:19.044 Cool, thank you. So I don’t know if you can share this, but are there any cars that we may have seen on the road that have your batteries in it?
Tony Li: 00:37:27.415 [laughter] let’s see. I’m not sure the US market, actually. In China, quite a lot right now. There will be a large customer using our batteries in the US, but I can’t tell the name right now [crosstalk].
Caitlin Croft: 00:37:45.901 Okay, totally fair. I mean, I think this is such a fun use case. And more and more people are getting interested in electric vehicles. Someone is asking, where in Europe is Gotion?
Tony Li: 00:38:02.863 Oh, it’s in Germany, the Europe office. Gotion is actually working with Volkswagen building up a factory in Germany.
Caitlin Croft: 00:38:13.654 Oh, cool. Yeah. Volkswagen’s massive.
Tony Li: 00:38:18.371 Yeah.
Caitlin Croft: 00:38:22.662 Awesome. Well, thank you, everyone, for joining today’s webinar. Once again, this has been recorded and will be made available for replay later today or by tomorrow morning, as well as the slides. We have lots of these sessions and if you’re doing something cool with InfluxDB, please let me know. I would love to share that story. If you guys have any more questions for Tony that you think of afterwards, everyone should have my email. Feel free to email me. I’m happy to connect you with him, so. Oh, another question for you, Tony. Did you try using a Kapacitor, which is the anomaly detection tool built into the [inaudible]?
Tony Li: 00:39:05.011 Yes, that’s a good question. That’s? something we plan to do in the future, actually using Kapacitor. Yeah. It’s a bit better native solution than our current framework, I believe.
Caitlin Croft: 00:39:18.758 And the cool thing is, once you upgrade to 2.0, a lot of that stuff is just all under the hood, so it’s even simpler. All the visualization and alerting is kind of built into InfluxDB, so it will be even easier for you guys. Well, thank you, everyone, and I hope you have a good day.
Tony Li: 00:39:42.529 Thank you, everyone.
Tony Li, Ph.D.
Manager of Software and Data, Gotion
Xiaojun (Tony) Li is currently the R&D manager at Gotion Inc, California, USA, where he is responsible for the control and machine learning algorithms for EV and ESS batteries. Dr. Li has over 8 years of R&D experience in energy storage systems.