1 Big Data Analysis and Mass Customization Model (MCM)

1.1 Introduction of Haier COSMOPlat

COSMOPlat is an industrial Internet platform initiated and developed by Haier based on its advanced business philosophy and rich experience in the manufacturing industry. COSMO means the God of Chaos and symbolizes the search for new developments out of Chaos. As a platform with independent intellectual property rights in China, COSMOPlat designs a user-oriented platform with the model of mass customization that opens to users for full-process engagement.

Technology is the driving force of development and change; in COSMOPlat, advanced intelligent manufacturing, network engineering, IoT, digitalization engineering, big data, and artificial intelligence are further developed to provide technical support.

Relying on the technological innovation system, COSMOPlat is mainly engaged in industrial Internet platform construction and operation, industrial intelligent technology research and application, intelligent factory construction, as well as hardware and software integration and fundamentally subverts the traditional industrial system and industrial structure, enabling the overall intelligent transformation and upgrading of enterprises.

With the philosophy of mutual evolution and value sharing, COSMOPlat strives to build an open, co-creation, and win-win ecosystem and embraces innovations in mass customization and information technology fusion in different industries. COSMOPlat is becoming a platform that empowers interactions and value sharing, facilitates new industries and new ecology, and supports entrepreneurs and innovations.

Starting from the mass customization model, an industrial Internet ecosystem with COSMOPlat is rapidly extending and empowering to different industries and fields, and through the replication of mass customization solutions, COSMOPlat not only enables Haier’s own digitalization but also opens up the whole industry chain ecology around users’ needs, realizes cross-industry and cross-regional ecological empowerment, and helps enterprises realize rapid transformation and upgrading. At present, COSMOPlat has built 7 centers and benefited 15 industries in more than 20 countries worldwide.

As typical use cases of COSMOPlat’s mass customization model, two Haier interconnected factories, namely, Haier Central Air Conditioning Interconnected Factory and Haier Refrigerator Interconnected Factory in Shenyang, have been rated as global “Lighthouse Factory” by the World Economic Forum, making Haier the only enterprise who owns two “Lighthouse Factories.”

1.2 Overview of MCM in Haier COSMOPlat

The MCM is a user-oriented approach initiated by COSMOPlat to provide customized and scenario solution driven by user’s needs, building a new industrial ecology for stakeholders to achieve win-win cooperation. In this approach, users are no longer just buyers, but rather participants that engage in the whole process of the manufacturing and designers who can put their ideas in building their desired products.

In MCM, based on new technologies including AI, blockchain, big data, and 5G, an open ecosystem for resource-gathering and sharing is built. This “tropical rainforest ecosystem” in the era of IoT enables a user interaction platform for enterprises to get access to user resources, as well as a win-win and value-added platform for SMEs and ecological resources to jointly create value and share value [1]. MCM provides an overall solution covering intelligent manufacturing, interactive customization, open innovation, smart services, precision marketing, modular procurement, as well as intelligent logistics.

Based on the principle of “co-building with large enterprises and resource sharing with SMEs” [2], it promotes transformation and innovation for large traditional industries and brings digital upgrade for SMEs.

By the user community established on the platforms, a large amount of user demand data has been accumulated, autonomic machine learning is combined to process the user demand data, and neural network is used to learn the characteristics of user’s demand, enabling the machine itself for the prediction of user’s demand. As a result, we can not only have a clear mind of the user’s demand but also help them find the services that best meet their needs. In this sense, quick respond to either the user or enterprise’s need is realized.

With MCM, the production efficiency has been improved by 60%, and the non-warehousing rate of the product has reached 75%. At the same time, COSMOPlat is recognized as “the most professional junction between enterprises and intelligent manufacturing resources.” While serving its own interconnected factories, it also provides solutions and value-added services for the transformation and upgrading of manufacturing enterprises, so as to continuously improve their mass customization capabilities and offer the best experience for users.

In the implementation of MCM, an interconnected factory system with direct connection to users has been built by COSMOPlat. Users’ order information can arrive at the factory in the shortest time, and personalized products are customized with high precision and high efficiency.

1.3 MCM and Haier Interconnected Factory

Departure from the downtown of Qingdao, it takes less than 20 minutes’ drive to Haier central air conditioning interconnected factory (Fig. 26.1) located in the Sino-German Industrial Park.

Fig. 26.1
figure 1

Haier COSMOPlat interconnected factory based on MCM

Full size image

The outside appearance of this factory shows nothing extraordinary; while entering you will find a totally different world. Walking into the production workshop, there are two robots waving their orange manipulators to work on the production line in an orderly manner. Technologies including IoT interconnection, AR cooperation, and AI detection empower the robots with intelligent operation capability, which not only liberates manpower but also improves production efficiency and reduces error rate.

Machine intelligence is only the foundation. The interconnection between equipment and production process and the links between manufacturers and users are the key for intelligent manufacturing in the era of industrial Internet. In traditional factories, overstocking always remains a headache for many manufacturing enterprises. In the interconnected factory, the central air conditioner on the production line has already had an “owner” long before it was “born”; thus, non-warehousing rate of products almost reached 100%. Among Haier’s 15 existing interconnected factories, the average product non-warehousing rate is 75%.

In the interconnected factory, all elements are interconnected with users. Walking around the factory, “screens” can be seen everywhere. Product specification selection, start customization, and order confirmation are all shown on the big screen; orders from all over the world are constantly updating on the screen, and each order is made based on the user’s needs. Mass customization platform, intelligent cloud service platform, energy-saving cloud service platform, field operation platform—the abstract industrial Internet platform is thus made visualized on various interfaces to show how different production factors interact and create value.

By interconnecting with the massive data collected by various sensors in the factory, predictive maintenance, intelligent scheduling, and quality intelligent monitoring of core equipment can be realized.

In September 2018, the factory stood out from more than 1000 factories and was rated as one of the first nine “Lighthouse Factories” in the world by the World Economic Forum, which was the only Chinese factory in the list. In September 2020, Hair MCM and interconnected factory won 2020 ROI-EFESO Industry 4.0 Award, becoming the first Chinese enterprise to receive this honor.

1.4 From Mass Production to Mass Customization

For high-precision product innovation and high efficient intelligent production, Haier has built seven application platforms based on a variety of intelligent technologies. The seven platforms, covering user interaction, marketing, R&D, procurement, manufacturing, logistics, and service, are broadly explored and applied so as to realize the large-scale customized application service centered on users’ personalized needs.

The customized platform for user interaction (HaierDIY) is a customized experience platform for user community interaction. Based on the platform, users can put forward demands, ideas, and comments and conduct online interactions with the home appliances, which become the source and driver for product innovation.

The precision marketing platform is based on CRM and user community resources. Through research on big data, existing user data and the user data collected by the third party are analyzed and studied. At the same time, cluster analysis is applied to form user portrait and build label management, so as to realize precision marketing.

The open R&D platform, which consists of three key components, namely, Haier Open Partnership Ecosystem (HOPE), HID iterative R&D platform, and collaborative development platform, is built to meet various requirements in different stages of R&D technology requirement interaction, product design and implementation, and collaborative development interaction of module manufacturers.

The modular procurement platform is developed on the module vendors’ collaborative procurement platform. It is a module vendor resource service and aggregation platform built for the needs of zero distance interaction between module vendors, resources, and users, enabling module vendor on-demand design and modular supply. The procurement system adopts a distributed architecture, in which user requirements are publicly sent to global module vendors, and the system will automatically and accurately match the right vendor and push the process forward.

The intelligent production platform is equipped with 12 key software modules for intelligent production, including OES, APS, MES, MMS, EMS, TPM, SCADA, EAM, SPC, WMS, etc.; it enables intelligent production scheduling, real-time monitoring, accurate distribution, planning, and power optimization. Through its deployment, factory mass customization is supported, and million-class customization can be achieved, eliminating the distance between factories, users, and resources.

The smart logistics platform consists of smart operation and visualization, covering platform reservation management, smart logistics TMS, distribution collaboration platform, logistics trajectory visualization and intelligent vehicle management platform, etc. Eight software products are provided including VOM system, reservation platform, smart vehicle distribution system (TMS), HLES3.0, WMS, hub distribution and mobile app system, etc., with the aim to bring users the best one-stop service experience.

The smart service platform creates a new home appliance service mode, and it meets the user’s demand for timely maintenance of home appliances. Users can input information of the home appliance by one key on the platform and create and upload the exclusive file of the home appliance, which can completely replace the traditional paper warranty card, and the information will never be lost.

Based on the platforms, the following new features and highlights have been realized:

  • Interaction: autonomous learning, demand forecasting, swarm intelligence.

  • Design: system intelligent design, collaborative R&D, virtual reality verification and optimization.

  • Experience: precise procurement, decision optimization, and deep mining of cooperative resource side.

  • Pre-sale: user portrait, sales forecast; feature analysis, marketing decision.

  • Manufacturing: intelligent equipment, predictive maintenance, intelligent production scheduling, intelligent quality monitoring.

  • Iteration: user image, sales forecast, feature analysis, marketing decision.

  • Human-computer interaction, community interaction, and intelligent life scene are upgraded iteratively.

2 Data Service and Data Flow of COSMOPlat

2.1 Current Data Service

On one hand, Haier COSMOPlat connects with users through interaction with interaction sub-platform and service sub-platform. On the other hand, through the IoT sub-platform and manufacturing enterprise instrumentation, industrial equipment, industrial control system, host computer connection, and the protocol conversion function, the unified MQTT Protocol will upload data to the cloud for interaction, design, procurement, sales, production, logistics, and service.

IT provides micro-services based on Springcloud and Springboot architecture and softwares such as MES, PLM, CRM, etc. to microenterprises by SaaS subscription. They are deployed to multi-tenant-based containers and are flexibly scalable; besides, capabilities including load balancing and disaster recovery backup are also possible as new features.

IT manages various capabilities in the COSMOPlat main platform, including:

  • IoT layer: sub-platform of IoT cloud device management, sub-platform of edge computing, smart gateway boxes integrated with industrial protocols.

  • IaaS layer: container data center.

  • PaaS layer: sub-platform PaaS cloud management, providing cloud-based public/private cloud service; developing sub-platform developer.

  • SaaS layer: tenant center, payment center, order center, logistics center, etc.

  • BaaS (Business and Best Practice) layer: mechanism model sub-platform, big data BI sub-platform, big data transaction sub-platform, AI sub-platform, blockchain sub-platform, food security code sub-platform (provide end-to-end production, wholesale, distribution, retail, and transportation tracking of frozen food), and COSMOPlat Tiangong OS (a unified portal that enables users to access multiple industrial apps/micro-services).

2.2 Data Flow Based on COSMOPlat

The data generated in the business interfaces (portals, apps, public accounts, etc.) of microenterprises are uploaded in real time to the data mid-office of the COSMOPlat main platform, which is managed by IT.

The types of data being collected include:

  • Financial data: number of paid users, conversion rate, etc.Footnote 1

  • Ecosystem data: number of registered business users, number of registered users, etc.

  • Capability data: number of connected endpoints, number of industrial apps, etc.

  • Third-party resource indicators

IT monitors, analyzes, and governs these data to understand business landscape across the company, and it is also further co-developing a framework with external business partners of a secure, confidential and compatible data inter-operation model. The below RV (Recreational Vehicle) case will help you have a general view on how data flows between COSMOPlat and our customers, and related interactions in between (Fig. 26.2).

Fig. 26.2
figure 2

How data flows in RV sub platform

Full size image

3 Use Cases of MCM

3.1 Best Practices of Big Data Technology

There are three typical application directions of big data in manufacturing industry: One is intelligent production, which enables the production system to have the ability of agile perception, real-time analysis, independent decision-making, accurate execution, as well as learning and improvement, so as to comprehensively improve the production efficiency. The other is intelligent products, which encapsulates the trained artificial intelligence system into the hardware and endows the products with the ability of intelligent response to external changes and user needs. The third is intelligent services, which can monitor the status of products in real time, quickly respond to the needs of users, and provide value-added services such as sale by rent, billing on time, remote diagnosis, fault prediction, remote maintenance, and integrated solutions, promoting the transformation of manufacturing enterprises from providing “products” to providing “products + services.”

UC1: User Interaction

Big data combined with deep learning to make intelligent prediction of user demand.

In the past, if the research and analysis of user needs is inaccurate, it will most probably lead to a long marketing cycle, and market changes are not synchronized. In traditional ways, user needs are fragmented and personalized; user’s ideas are collected one by one, making it nearly impossible to form a feasible scheme to achieve precision marketing and data transparency; and user behavior and user tag trajectory are not visualized. In this sense, user product demand prediction needs to be changed to user scenario prediction so as to create user experience value.

In the era of IoT (Internet of Things), the consumer market is undergoing a trend change from the traditional “enterprise-centered” mode to “user-centered.” COSMOPlat has established a community interaction platform for direct communication with users, turning the uncertainty of users’ needs into certainty, and the community interaction solutions can also meet the changing needs of users. Based on the social platform and scene interaction, a large number of users with “high aspirations” can gather 100,000 or even more users in a short time. Based on the Internet platform, one user’s pain points and needs become a group of people’s (10,000, 100,000, or more). Finally, it will realize the personalized customization and large-scale sales of household appliances.

Take Haier’s infant and mom customization as an example: the mini dryer originated from the wisdom of mothers in the infant and mom community. Last year, maternal and infant groups put forward the demand for clothes dryers, which has raised heated discussion among the maternal and infant community. The interactive exposure was more than 3 million, user’s participation in the interaction reached more than 100,000 times, the 180-day interactive product design took more than 10 iterations according to the user’s interactive demand, and eight product functions were iterated. At present, the product has been put onto the market, and its sterilization rate reaches 99%+ and is capable of collecting threads and scraps. The inner barrel has the function of forward and reverse rotation and regular shaking, which can prevent clothes from winding or wrinkling. In addition, the drying process follows six professional procedures, making it possible to automatically sense the humidity of clothes so as to meet various needs of different users.

UC 2: Customized Experience

Big data combined with deep learning to create sleep curve of Haier Tianzun air conditioner

When heating season comes, many consumers will turn on their air conditioner overnight for heating. Due to the great temperature difference between day and night, the air conditioning also needs to adjust the temperature frequently, which is the headache of many air conditioning users. The birth of Haier Tianzun air conditioner provides intelligent indoor temperature solutions for users, bringing an end to the complaint of frequent temperature adjustment during sleep. By studying and analyzing different needs of the elderly, children, and people from the north and the south, Haier launched an automatic sleep curve to bring users a peaceful and comfortable sleep environment.

Users can change the sleep curve according to their own preferences and name it with their favorite names. What’s more humanized is that “sleep curve store” is also provided in the app, and users can choose more functional modules according to their own needs. The intelligent sleep curve mode completely solves the problem that the air conditioner does not match human body temperature during sleep and frees consumers from adjusting temperature with a remote control all the time.

After deeply exploring the sleep curve of Chinese people, Haier Tianzun air conditioner provides diversified sleep modes according to different seasons, user groups, and genders. The summer curve and winter curve classified by season also provide a variety of choices according to different time periods. Users can choose the 12:00–14:00 sleep curve in the nap period accordingly, and they can also choose the 23:00–08:00 sleep curve when night comes. For different users, such as the elderly and children with low immunity, Tianzun air conditioner will automatically select a soft sleep curve with low temperature difference to control the indoor temperature. For adults who are greedy for cool or warm environment, quickly reaching the consumer’s favorite temperature would be a highlight function of Tianzun intelligent sleep curve to enhance user’s comfort. In addition, the sleep curve can be remotely controlled by using smart phones, pads, and other terminal devices. Just download the app of “Haier intelligent air conditioner,” and you can choose to turn on the sleep curve. Tianzun air conditioner will adjust its operating temperature and state according to your settings in the app.

UC3: Machine Visual Auto-inspection of Refrigerator Appearance

In the traditional production process of refrigerators, the appearance detection remains a bottleneck; high labor intensity of employees, complicated defect detection procedures, and various manual detection standards lead to the low detection rate of defective product to be only 92%, causing customer’s complaints from the market.

By statistically analyzing on-site problems, Haier gathered first-class resources to develop a visual automatic inspection system (Fig. 26.3) for the detection of refrigerator appearance. The system conducts online quality inspection for refrigerator appearance by intelligent learning of qualified products. Inspection scope includes printed matter, door body and appearance, etc.

Fig. 26.3
figure 3

Automatic visual inspection of appearance

Full size image

By using the appearance visual automatic detection technology instead of manual work, the automatic detection efficiency is improved by more than 50%, and detection rate reaches more than 99.5%. In addition, the detection information is visualized, and unqualified products will be kept on production line and not be sent to the market. The application of the system not only improves the production quality and efficiency but also increases the user satisfaction.

UC4: Voice Recognition Combined with In-depth Analysis to Provide Product Online Noise and Abnormal Sound Detection Solution.

In a typical detecting environment, the background noise is lower than 30 dB; in a traditional way of manual detection of the decibel value, the error is quite large. For each detection, the abnormal sound is tested through earphone, which contains the problem of voice leakage, as well as the difficulty to record the written statistics and poor controllability of the maintenance state, making it even more difficult to get an accurate result. Regarding this, a more advanced detecting method is becoming more and more people’s prospect.

The COSMOPlat cloud platform collects the users’ needs from Microblog, WeChat and search engine, etc., finds out various needs of users for all brands of air conditioning, and takes the air conditioning sound as the main concern through data analysis.

Air conditioning sound mainly includes noise and abnormal sound, and their level can be distinguished by decibels. There are tens of millions of different sounds. Relying on big data and artificial intelligence technology, a platform for online noise and abnormal sound detection (Fig. 26.4) is built. The platform can independently learn to identify abnormal sound and conduct automatic control and reduce the miscalculation of artificial detection, so as to improve the accuracy of identification and reliability of the product.

Fig. 26.4
figure 4

Online noise and abnormal sound detection

Full size image

While focusing on noise problems, the production process is traceable. By studying the big data of production process, the causes of abnormal sound (including poor installation of air conditioning fan and motor or the main frame) are analyzed, and key improvement measures for further prevention in advance are then summarized.

The independent learning to identify abnormal sound and control not only builds a modern detection way consisting of automatic isolation and transfer, real-time information interconnection, automatic storage, traceability, automatic process traceability, and problem closed loop, which largely improves the accuracy of identification, but also brings the benefit of labor cost saving of 5 million; more importantly, user experience is greatly improved during this process.

3.2 Value Chain of Use Cases

In the value chain of Haier’s personalized customization solutions (Fig. 26.5), there are three parties: industrial Internet platform enterprises, home appliance manufacturers, and home appliance buyers. For home appliance buyers, the solution solves the user’s demand for personalized customization and improves the user’s experience of purchasing and using home appliances. For home appliance manufacturing enterprises, the pursuit of “improving quality, reducing cost, and increasing efficiency” has been realized. Last but not the least, industrial Internet platform enterprises can continue to generate revenue in the process of providing services for users and manufacturing enterprises.

Fig. 26.5
figure 5

Value chain of user cases

Full size image

At present, MCM has built a transformation service solution for whole industry users to get whole process participation globally, covering 15 industry ecologies including clothing, emergency supplies, agriculture, etc., established 7 centers nationwide, covered 12 regions, and has been replicated and promoted in 20 countries.

With the promotion of MCM, COSMOPlat has got comprehensive leadership in the following three dimensions:

  • Standards leading. COSMOPlat is the only industrial Internet platform authorized by ISO, IEEE, and IEC to develop MCM standards.

  • Model leading. At the end of 2019, COSMOPlat ranked No. 1 Industrial IoT Software Platform of Q4 2019 by Forrester, the world’s leading market research and consulting organization.

  • Brand leading. As an ecological brand, COSMOPlat is the only industrial Internet platform in the top 100 Chinese brand value list in 2019.

From the leader of China’s “lighthouse factory” to the winner of German industry 4.0 award, MCM has fundamentally solved the problem of SMEs having customers but no users, empowering SMEs for digital transformation and upgrade while promoting integration and innovation of traditional industries with industrial Internet. Based on the ecological system of multilateral interaction, co-creation, and sharing, it gradually completes the transformation from “lighthouse factory” to “lighthouse base”, embracing an overall industry upgrade which leads to a high-quality development for enterprises as well as the entire economy.