Guest of this issue
Liu Zhaotao Founder of Zhihe Technology
Liu Zhaotao, the founder of Zhihe Technology and a serial entrepreneur, has rich experience in construction machinery leasing, sales, management and engineering construction. Zhihe Technology was established in 2014. With the mission of "continuously improving construction efficiency", it serves construction companies and provides mature solutions for digital management of construction machinery in the digital transformation of enterprises. During this period, it participated in the research and development of 75 authorized invention patents.
host
Tian ChaoGeneral Manager of Tencent Cloud Enterprise Center
Tian Chao, General Manager of Tencent Cloud Enterprise Center and General Manager of Operation Center, is responsible for Tencent Cloud user growth, DNSPod business and enterprise application related products. At the same time, he is also a senior user growth expert and big data technology expert. He used to be the general manager of Yongyingbao growth platform and the vice president of Mobike technology. Long-term commitment to research on enterprise digitalization.
1
Tian Chao : Zhihe Technology is a manufacturer of digital solutions for engineering equipment management. In our impression, industries are heavy and tiring, and the input and output are not high. Why do you choose to enter the industrial field? How did you discover the market opportunity for Industrial IoT?
Liu Zhaotao :Although many people now include digital construction sites in the scope of the Industrial Internet, strictly speaking,we belong to the engineering field, not the industrial field.
Why do I place special emphasis on the "engineering" industry? Because the engineering industry is indeed more difficult to operate than the industrial industry. The digitalization level of the engineering industry ranks last in the industry . It is still in the transition stage from handicraft industry to mechanization, but this just shows that this industry needs to be changed. This is what I founded. The original intention of Zhihe Technology.
I used to be in the business of leasing engineering equipment, and later, by chance, I became a construction site contractor. During this period, I found that the management of this industry is very difficult and the efficiency is very low .
Nowadays, many construction sites do not buy equipment themselves, but rent equipment. If I rent 50 equipment, I need to deal with 50 equipment owners, and I have to wait for the driver to come, do training, etc., which is more difficult to manage than other industries. After the start of construction, I have to check how much work the workers have done every day, how much fuel they have consumed, and whether they have been lazy during the period. Finally, according to their workload, I will issue manual invoices to them for settlement.
There are very large management loopholes in this process . Even if I send my own people to watch these workers go to work, there will inevitably be omissions. In addition, the contractor may not be on site temporarily, or the contractor may not be able to manage it. As a result, some workers have no work to do, which will lead to laziness and waste of resources.
The pain points in the engineering industry have always existed . A little investment in technical means can solve some existing problems on the construction site and produce immediate returns, so I think this industry is very worthwhile for us to do.
2
Tian Chao :In the eyes of many traditional enterprises, the cost of digital transformation projects is high, the cycle is long, and the difficulty is great. In the end, it may be reversed. The people's court ordered reorganization, so the acceptance of digital transformation is not high. Have you also encountered a wall in this regard in the process of promoting digital solutions? How do you drive construction companies to be willing to spend money on equipment IoT? You mainly serve central state-owned enterprises. They are more cautious in transformation. How do you gain the trust of these central state-owned enterprises?
Liu Zhaotao :When we promote to customers,we don't specifically emphasize the lofty concept of "digitalization". Our approach is to tell customers what we are doing and what problems we can solve.
Many customers have been engaged in the engineering industry for a long time, they have long experienced the pain points of the industry, and as experts, they can know whether their problems can really be solved by looking at the solution. As long as they are willing to try, we will let them do a pilot first, and it will be easy to fully promote it after it is effective .
From the data point of view, 86% of our revenue comes from the repurchase of old customers , which shows that our plan has indeed brought them results. Judging from customer feedback, we can save at least 15% fuel consumption on the construction site, and the workload of employees can be reduced by more than half .
Talk about a very typical customer case of ours— Shandong Road and Bridge . After using our products, they can clearly monitor the usage of the equipment, and can also dispatch tasks scientifically in real time. In the end, they saved nearly 70 million yuan in equipment rental costs, and saved more than 100 million yuan in equipment costs and fuel consumption costs.
The central state-owned enterprises accounted for 80% of our sales, because the central state-owned enterprises generally need to manage relatively large projects , and the effect of digital solutions is more obvious. If two devices are working, one boss can see it, but if hundreds or tens of thousands of devices are working at the same time, a system is needed to help manage them.
How to eliminate the concerns of central state-owned enterprises? Still rely on experimentation. You asked if we will pay back for frequent pilots? In fact, the joint promotion effect of central state-owned enterprises is very strong . Either the headquarters unit encourages subordinate units to do equipment management digitization, or the divisional units do pilots first. After seeing the effect, multiple units at the bureau level do not even need to pilot. Just use our plan.
3
Tian Chao :The Internet of Things for engineering equipment requires sensors, but different companies may purchase sensors of different brands, types, and indicators, including oil level monitors, load monitors, attitude monitors, etc. I see you There are intelligent terminals of engineering equipment that need to be linked with these sensors. There are a lot of communication protocols and different communication interfaces. How did you solve these problems?
Liu Zhaotao :This question is very professional. You must first know thatmany engineering equipment have their own sensors, so we also wanted to directly collect the above data in the early stage, but the challenge is very great, andnow we can announce that this direction will not work at all.
First, there are more than 900 brands and 40,000 models of construction machinery in China . Since manufacturers need to protect their own core data, it is almost impossible to communicate data between different devices unless the country stipulates that the entire industry must follow a certain national standard , line standard. Second, the equipment of many small factories cannot be connected to the Internet . Even if we have the communication protocol in hand, the data on site cannot be connected.
Therefore, our solution is: Let the construction site install a set of our own hardware to collect the data we want , so we don't have to be restricted by the agreements of various manufacturers.
So what data to collect? ——Collect data on what customers are willing to pay for .
At the beginning, we thought that data such as engine speed, oil temperature, water temperature, etc. could measure the health of the car itself, and these data were valuable.
But then we discovered that our customers are construction companies, and they don’t really care about these data. What they want in the management process is “3+1” : how much work does the engineering equipment do every day, how much fuel is added and how much fuel is consumed, How many times have you run, here are 3 small tickets; according to these small tickets to settle, +1, it is to increase efficiency management: whether working is lazy, hard work, how to improve work efficiency.
After clarifying this direction, we know how to make our own sensors, such as collecting acceleration and angular velocity through a six-axis sensor, measuring oil level data with a hydraulic sensor, and then combining artificial intelligence algorithms and modeling to improve data accuracy . Now we have achieved an accuracy of more than 99% , which is used to solve the 3+1 problem of on-site production and equipment management.
4
Tian Chao :Zhihe Technology has launched sensors for engineering equipment and an Internet of Things management system, but for many traditional construction companies, the Internet of Things means that they need to modify the hardware and software of engineering equipment. Transformation time. In the industrial field, a second delay of a machine will cause huge losses. In addition, enterprises may also face cost and technical problems. How do you adapt to different engineering equipment and management systems of customers? How to help customers solve the problem of renovation costs?
Liu Zhaotao :We have spent 4-5 years stepping on the pits, and found that there are various protocols for construction machinery, and there is no standard interface at all. It even takes two days to install a piece of equipment. How can the construction site be willing to wait? In addition, the construction site environment is relatively complicated, and it is impossible for the equipment to drag a very long line, and it is easy to dig the data line and cut it.
We have come to a conclusion: the difficulty of connecting a wire for IoT products on the construction site is far beyond our imagination!
Therefore, we must make our sensor equipment wireless, so as not to delay the work , the customer can "snap" the monitor on the roof of the car. Our oil level monitor replaces the original fuel tank cap and is directly screwed on above. Only when the installation is simple can the digital construction site be truly implemented.
The system integration you mentioned is not a big problem in the engineering industry . Because the informatization level of the industry itself is not high, they don’t use common project management software such as ERP in the market much, and most of them still manage with excel sheets. Therefore, our products have a separate set of software, which is very important for them. It doesn't matter.
5
Tian Chao :Industrial production often faces more complex environments than other industries, such as tunnels, mines, etc., so how to ensure normal communication of IoT devices when facing harsh and remote environments, without delaying control and abnormal alarms, and maintaining system stability? Stable operation?
Liu Zhaotao :Why is it so difficult to digitize construction sites? The first is the data acquisition just mentioned, and the second is data transmission.
The network environment on the construction site is very poor, and there are even no signals in many places, such as the wild and deep pits. We deal with it in this way: where there is a signal, we adjust the antenna parameters better, so that the data can be sent out smoothly. In places where there is no signal, we first store the data , which can be stored for several months, and then upload the data to a place with signal.
In addition, for some customers who work in tunnels all year round, we can also regularly send people to export data , or provide some tunnel networking solutions. Of course, in the end, customers should choose after considering the investment cost.
6
Tian Chao :Your sensors can provide real-time positioning and trajectory tracking of engineering equipment, and collect a large amount of data on the type, energy consumption, and time of engineering equipment. Your customers are mainly central state-owned enterprises, and they pay more attention to information security. How do you Reassure customers of privacy and security concerns?
Liu Zhaotao :This issue was often mentioned in the early days, but now customers are basically willing to accept it.
First of all, the data collected by our sensors is not particularly sensitive data. In addition, strictly speaking, these data are not the business data of the construction company itself, but the equipment data of the unit . For car owners, there is not much need to keep these data confidential.
Of course, before we implement the digital solution, we will also ask the customer whether they accept it and whether the data is sensitive. If they do not accept it, we will not be able to provide services.
7
Tian Chao :There are already many players on the industrial Internet of Things track. Many manufacturers are focusing on the technical features of high precision and low energy consumption like Zhihe Technology. These technologies seem to be nothing new in the industry, and you call it "refresh project A new level of digitalization in the machinery industry", so what are the differences and innovations between you and your competitors? You have been immersed in the industrial Internet of Things track for many years. How do you think the industry barriers are formed?
Liu Zhaotao :The quality of data determines the depth and breadth of enterprise digital applications.How to measure the quality of data? One looks at comprehensiveness and the other at accuracy.
In terms of comprehensiveness, the site owner should manage every piece of equipment as much as possible. Our hardware does not need wiring, it can be used after being glued on , ensuring that no matter what equipment can be installed and connected to the Internet, this is comprehensive.
The difficulty of wireless hardware lies in energy consumption. If there is no wiring, where does the electricity come from? We use ultra-low-power Internet of Things technology to design power-saving strategies based on actual scenarios, equipment, and components. In addition, we adopt solar charging , even without any sun, the hardware terminal can continue to work for more than 8 months.
In terms of accuracy, the difficulty of the sensor is not its sensitivity, because the sensors we use are generally consumer-grade sensors, which have been tested by a large number of mobile phone users, and the accuracy is very high.
The difficulty of the sensor lies in the algorithm behind it . There are more than 40,000 models of engineering equipment in China. How to calculate the data of each model accurately? This is the advantage we have accumulated over the past few years.
When our customers are getting more and more, and the degree of usage is getting deeper and deeper, interactive feedback and algorithm upgrades will make the data more accurate. We summarize the four stages of digitalization in the engineering industry:
L1: In the data display stage , the data accuracy reaches more than 90%. This means that the equipment on the construction site is basically connected to the Internet, and it is clear which equipment is fake, which equipment is lazy, and so on. 80% of customers have been able to recover the cost within one month after using L1.
L2: In the stage of on-site management , the data accuracy reaches more than 95%. After reaching this level of accuracy, companies can use these data for comparison, comparison between equipment, comparison between projects, and comparison between branches, so as to design performance appraisal and process management systems.
L3: In the stage of business settlement , the data accuracy reaches over 99%. This allows the data to be directly used as the basis for financial settlement.
L4: AI command stage . Based on high-quality data and combined with on-site practical experience, intelligent command can be realized.
Zhihe Technology Machinery Commander Fleet Management
It took us nearly 4 years to achieve an accuracy of more than 90%, and it took us the sixth year to achieve an accuracy of more than 99% . At present, it seems that our competitors can only achieve an accuracy of 60-80%. There are quite a lot of competitors in the entire IoT industry, but if it is subdivided into the engineering industry, we can say that there are almost no competing products.
8
Tian Chao :You said that you have no friends, but does this reflect from the side that the digitization of engineering equipment may not be a good business, so no one is willing to invest? I just saw your 9th anniversary in Moments last weekend. You have been doing it for 9 years. Have you made any money now?
Liu Zhaotao :I am indeed reflecting on this issue. The result of my thinking is not to say that the industry does not need digitization. This must be a pain point that has existed in the industry for many years and needs to be solved urgently.
At present, the digitization of the entire industry has not yet been done, mainly the Internet of Things in the engineering field, which is more difficult than other fields . Many friends and businessmen also saw opportunities, but they did not expect it to be so difficult.
Doing this business requires entrepreneurs to calm down and invest tens of millions to hundreds of millions in research and development expenses in the early stage . In fact, we only started to make profits last year, and we have been relying on financing and sales revenue to support the company's operations during this period.
9
Tian Chao :AIGC is undoubtedly the hottest topic this year. Do you already use AI capabilities in your digital solutions for engineering equipment? In your opinion, does ChatGPT really lead to a major change in the field of AI? What changes will the new round of AI applications bring to the construction scene?
Liu Zhaotao :When ChatGPT first came out, I was very surprised. The effect of AI application is far beyond imagination. But there is indeed some panic in the back, fearing whether ChatGPT will subvert the entire industry.
The chief scientist and AI leader of our team is a Ph.D. from Yale University and a former core engineer at Google headquarters. He quickly shared with us, allowing us to better understand how to use large models to help the industry improve efficiency.
We believe that there is still a lack of small models in the construction and civil engineering industries. In the future, when combined with large models, they can be used as assistants for engineering project managers and integrated into our L4 smart command solution.
Where does the data come from to train this small model? The level of digitalization in the engineering industry ranks last in the industry. It can be said that the industry itself does not even have data. How can we talk about training data and precipitation data?
We can use our mechanical commander and intelligent hardware to collect data on engineering equipment. After collecting these data, how do you know how to direct and manage the project manager?
Method one, we can obtain data from the project management software used on the site .
Method 2, we found that project managers actually prefer to use excel sheets, then we can make a smarter and more useful smart sheet , customers can use it to improve efficiency, and we get the data, so that the cycle can put Data accumulates.
10
Tian Chao :You once suggested that Zhihe Technology will go through three stages, from 1.0 interconnection of all machines, to 2.0 assisted driving, and then to 3.0 unmanned land. So which stage do you think you are currently in? To achieve the final 3.0 without artificial land, what do you think we still lack? How should the industry support it in the future?
Liu Zhaotao :Although we always say that we are "infrastructure maniacs", the efficiency of the engineering and construction industry is actually very low. Since we are going to improve the efficiency of the industry, I will first look back from the perspective of the future.What should the construction site efficiency be like when it is maximized? The answer is "no artificial ground", that is, our construction machinery can work by itself.
The unmanned driving of a single device will not be particularly difficult to realize, because with the support of the current unmanned driving technology, and the "unmanned driving" era of construction machinery on the construction site will definitely come earlier than road vehicles . Because the construction site is a relatively closed scene compared to the roads extending in all directions.
But it is not so easy to realize unmanned work, because dozens or even hundreds of machines are working on the construction site, and these machines lack a "command brain" that can be deployed and dispatched in a unified manner .
At present, on the construction site, construction machinery is directed by "people", and the efficiency of fleet operations greatly depends on the experience level of the construction team leader, project manager, and contractor. How do they arrange troops, how do they dispatch work, and how do they arrange the progress in different engineering projects? How to refine, precipitate, and re-optimize these excellent management experiences is the key to forming a command brain. Therefore, we need to obtain various data from the construction site for analysis and calculation.
From obtaining data, to forming a command brain, and then to realizing unmanned land, this also corresponds to our three-step strategy.
10,000 machines are interconnected. That is, 1 million+ devices are connected to the Internet (there are nearly 10 million construction machinery in China, and we are now committed to covering 10% of construction machinery) , and through 3-4 years of continuous data collection, gradually cover various possible situations Operation scenarios , such as digging fish ponds, digging earth and rocks, mining, etc., are analyzed and calculated by our algorithm, and are deposited in the database. Forms the basis of the "command brain".
2.0 assisted driving - a network on the construction site. In addition to equipment data, we also need more data on the construction site to train the brain, so in the future we need to build a "one network at the construction site" by ourselves to form a low-power transmission network and a high-precision positioning network . Collect equipment data, as well as material, quality, and safety data to help construction companies make better decisions, and can also be used to achieve high-precision positioning——check the location and status of equipment in real time after entering the site.
For us, this kills two birds with one stone. Not only has assisted driving been realized, we have also collected more data through sensors in the process, optimized our algorithm again, and upgraded our command brain.
After the successful realization of the first two stages, the arrival of the 3.0 era of no man-made land is just around the corner , when the construction site will become a high-efficiency, low-waste operation scene.
At present, Zhihe is still in the 1.0 stage and is moving towards the 2.0 stage. In the past two or three years, we have accumulated hundreds of thousands of equipment data, and we are also actively developing and trying the project of "one network at the construction site". In the future, we will continue to cultivate and let domestic engineering equipment join the wave of digitalization.
* Image source: Zhihe Technology
END
Column coordinator | Zhao Jiuzhou
Responsible Editor | Huang Qiting, Zhuang Yajie, Zhang Jie
Column Consultant | Caoheyan
Are you optimistic about the development of Internet of Things in construction sites? How do you think AI can improve the efficiency of construction sites? Welcome to share your views in the comment area~ Light up "watching" + leave a message in the comment area , Ah D will randomly select 2 fans at 15:00 pm on July 19 (Wednesday) and send Ah D a customized keychain~
"DNSPod Ten Questions" is an in-depth talk column launched by Tencent Cloud Enterprise Center. Through each issue, ten questions are asked to the guests, leading readers to stand on the shoulders of elites in the industrial Internet and technology fields, and overlook the development of major industries Trends and cutting-edge technological innovations.
The field of column guests is gradually expanding, from the initial circle of domain names and webmasters to the circle of programmers, entrepreneurs, and investors. Ding Ke, vice president of Tencent, Jiang Tao, chairman of CSDN, Dai Zhikang, founder of Discuz!, Wu Lujia of Knowledge Planet, Yang Qing, vice president of Tencent Security Academy, and other technical experts and industry leaders have left their insights in this column .
"DNSPod Ten Questions" is also extremely influential and active in the Tencent Cloud ecosystem. We are on Tencent's internal platforms - DNSPod official account, Tencent SME service official account, Tencent cloud official account, Tencent cloud host official account, Tencent cloud server official account, Tencent cloud assistant, Tencent Lewen, Tencent code guest circle, Tencent KM platform, Tencent Cloud + Community, Tencent Cloud + University and other platforms have accumulated hundreds of thousands of attentions. The total number of media readings such as agency account, penguin account, Sohu account, headline account, open source Chinese technology community, IT home, InfoQ community information site, Twitter agency account, and Facebook agency account exceeds one million.
In the future, we hope that the influence of this column will cover a more diverse audience and spread more correct ideas to the outside world. Readers are welcome to leave the guests you want to see and the questions you want to ask in the comment area. We invite you to become the questioners and voicers of the "DNSPod Ten Questions" column.
Cooperation contact:
▼Get the QR code in the background of the official account
Join the official DNSPod user group
