当前位置：网站首页>Edge box + time series database, technology selection behind Midea's digital platform iBuilding

Edge box + time series database, technology selection behind Midea's digital platform iBuilding

2022-08-03 20:03:00 【Taosi data TDengine】

小 T 导读：在 2021 楼宇科技 TRUE 大会上,Beauty of hvac and building department issued a digital platform for the first time iBuilding,以“Floppy hard drive”Ways to empower the construction industry.作为一个全新的项目,iBuilding Be cautious in database selection,Compare several Database 产品之后,just made my choice.This article shares their database selection thinking and landing experience.

政策背景

根据 2021 年 12 Month by the us accused of intelligent building joint ou think-tank released jointly《China Building Automation White Paper》,2021 The annual output value of China's building intelligent market is about 7238.2 亿元,Combining the development trend of industry in recent years,经过初步估算,2016-2021 Intelligent building in China market is rising year by year,stock size close to 5000 亿元,Add more than 2200 亿元.

Low-carbon due to intelligent building、Energy efficiency advantages,At the same time, it can bring more comfortable experience to people's life,In addition, the government has standardized the intelligent construction of buildings.、scientific guidance,Intelligent building will have a very good development prospect in the future.从目标来看,Building intelligence meets the development needs of digitalization and intelligence in the construction industry,The future will continue to help the Chinese construction industry transformation and upgrading,In order to comply with the national requirements for energy conservation, emission reduction and digital economy.

业务介绍

随着 5G 时代的到来,On the one hand, Midea is continuing to build industrial Internet products,On the other hand, technological empowerment is constantly being carried out,Development of greener integration solutions,Provide new ideas for industry and manufacturing.

As one of the five business segments of Midea Group,Midea HVAC and Building Division established“A leader in HVAC and building smart ecological integration solutions”的发展愿景,Designed to meet complex building needs with intelligently integrated industry solutions,Currently mainly involved in central air conditioning、电梯、Building control and other fields.在 2021 楼宇科技 TRUE 大会上,Beauty of hvac and building department issued a digital platform for the first time iBuilding,以“Floppy hard drive”Ways to empower the construction industry.

作为一个全新的项目,We compared the relational database, respectively（Relational Database）As well as the mainstream of temporal database（Time Series Database）,包括 InfluxDB、TDengine、MySQL 等.对比关系型数据库 MySQL 来说,在这个场景下,We don't need complex queries,However, it requires efficient storage and large-scale data pull.And with the temporal database InfluxDB 对比,TDengine The performance of the stand-alone version is much better than InfluxDB.因此,In a comprehensive assessment of the fit、查询、After writing and storage and other comprehensive ability,我们最终选择了 TDengine 这款产品.

iBuilding 项目属于“智慧楼宇”的一部分,The project itself is used for edge side to large refrigeration equipment（中央空调）The intelligent monitoring and interaction.具体应用场景是：Dozens of buildings involved in the project,Each has some large centrifugal chillers（10 台左右）,We have deployed a TDengine 到 ARM64 系统上.通过 Python 程序,Data acquisition system will first,然后把数据写入 TDengine ,Finally, the data uploaded to the cloud TDengine 进行处理.

具体实践

以其中一个 Database 环境为例：

我们根据 TDengine “一个设备一张表,A kind of equipment a super table”The modeling principle of,Created the following table,The indicators of the two types of equipment are: 97 和 199 ,Data column starts with float 和 int 为主,设备每 5s Submit a batch of data：

To the side of the edge data collection,由于资源有限,Therefore, the use of resource data has become a very important indicator..这方面 TDengine 表现非常好,Further help us reduce costs and increase efficiency.

Our edge box hosting the database service is configured as 2GB 内存,4C CPU,ARM64 位的系统.Due to the small number of subtables,以及 TDengine Write the characteristics of relatively fixed memory,The current memory usage is less than 200MB.Database daily CPU 消耗比较低,大概在 3%-5% 左右,Conservative estimate even if write volume expands 50-100 倍,也没有问题.

应用效果

ask for a device 70 天前到 40 days ago,Device power usage over time,Use without data prev 值填充.结果如下：

Query for a month before the sum of several indicators of a certain equipment,Sort by timestamp in descending order.query about 19 万行数据,耗时 0.4s.结果如下：

‍

经验汇总

在使用 TDengine 的过程中,We also had some small problems,比如：Our environment is numerous,But the client and server must keep the exact same version,Upgrading will be more complicated.再比如：monitoring library log 中的 dn 表的 disk_used Semantics are not practical TDengine 对磁盘的占用,But the data files in the file system's total footprint,In some cases, users may mistakenly believe that it is TDengine 的空间占用,result in unexpected,就像下图一样：

later we and TDengine Community staff discussed the situation together,You think you can add a column,dedicated to statistics TDengine the size of the data file,然后把它与 disk_used、disk_total Normalize and unify naming together,to prevent users from misunderstanding.

目前 TDengine The official has been actively processing optimization.This is also a great value of the open source community,Everyone can participate,Let the product continue to iterate,发展地更好.

写在最后

当前,TDengine Mainly used in the monitoring business of central air-conditioning refrigeration equipment,as a pilot,This scene has achieved good results.But because the unit is expensive、成本较高,So by this kind of intellectualized operation platform dynamically generated operating instructions are still need to be careful,So this feature has not been officially opened yet..

In the aspect of intelligent building,We also have a lot of work to do,Monitoring from the edge side、to command control、Then to the integrated service of edge-cloud collaboration,We will continue to explore and dig in these scenarios TDengine 的潜力.