One of the important industrial applications as Internet IIoT and edge computing, predictive maintenance had high hopes about two years ago. Major companies and start-ups have been focusing on the layout, it seems to have convinced predictive maintenance will become one of the few IIoT “killer” applications. Such as Huawei seize market pain points, select from the “ladder networking” cut into the elevator operation and maintenance areas. ABB in Bangalore set up new digital remote service center for energy-saving inverter, year-round Remote Access is located within the drive end-user facility, to achieve predictive maintenance and condition monitoring. Honeywell launched predictive interconnected auxiliary power unit maintenance services GoDirect, Hainan Airlines to become the world’s first airline using the GoDirect. Airbus chose self-built edge computing and cloud platform capabilities, predictive maintenance system tailor-made for their own use. Given the number of stock equipment market is considerable, the vast majority have not adopted effective predictive maintenance program, while 50% of the total maintenance costs incurred over the life cycle cost of equipment. Who would not want a better means to prevent equipment malfunction? Authoritative institutions predict that predictive maintenance in 2022 before the market will maintain rapid growth in Control Engineering Copyright , a compound annual growth rate (CAGR) of 39%.
Two years later, how well the development of predictive maintenance? Bain recently released a report outlining a surprise Overall, the core idea makes sense: the development of predictive maintenance is less than expected. The development and progression anything, has its inevitable reasons. Typical applications for predictive maintenance industry as the Internet, analyze its development status for the entire field, with a good reference. Therefore, in this article, you will see: predictive maintenance of the status quo thoroughly. Why is the development of predictive maintenance is less than expected? What predictive maintenance capability of start-ups in practice? Maintenance and repair of industrial equipment thoroughly the status quo is broadly divided into three kinds of methods: corrective maintenance: maintenance after the event belongs to remedy the situation. Preventive Maintenance: Maintenance belong in advance, time, performance and other conditions of the equipment for regular maintenance based, more or empirically. Predictive Maintenance: Maintenance belongs advance, based on various sensors mounted on the device, operating state of the real-time monitoring devices, and more accurately determine when a fault occurs. If you find hidden faults automatically trigger an alarm or repaircommand. GE released a study two years ago fresh in my memory – survey results show that a large number of industrial enterprises are moving towards predictive maintenance “embrace.” Internally, predictive maintenance for the optimization of manufacturing operations, will bring 20-30% efficiency gain. From the outside, the device manufacturer if the introduction of predictive maintenance services, it is possible to reverse the current competition format. From a strategic point of assessment, predictive maintenance and industrial services represents the historical choice of the future business model transformation. In particular cloud platform edge computing and the development of artificial intelligence, opening up with the latest technology to change the predictive maintenance market structure doors of opportunity. Considering the above points, various companies have vied for predictive maintenance this party “fertile soil”, of course. Although the development of predictive maintenance remains strongly bullish, but many companies have realized the true predictive maintenance to be effective taking longer than expected. In early 2019, Bain executives of more than 600 European companies conducted a survey, many customers look for predictive maintenance, it has become more rational by the enthusiasm. Predictive maintenance solution implementation process more difficult than expected, and the difficulty of extracting valuable insights from the data is far beyond imagination. Bi Beien advice on research carried out twice, respectively in 2016 and 2018, the real implementation of preventive maintenance programs and plans to use the proportion of companies that have reduced, although everyone for predictive maintenance are convinced of the future, but compared to 2016 the resulting red dotted line of research, many companies are adjusting the tempo and slowed the advance predictive maintenance.
for predictive maintenance difficulties and risks faced by the program in advance, they tend to be more objective judgment. In addition to many companies concerned about security, return on investment analysis, IT and OT difficult to integrate than for lack of technical knowledge, portability, variable handover of data providers and program risks are re-evaluated. From the practical point of view, although the industry to enhance Internet security, accelerate the integration of IT and OT of each other, given the uncertainty of the return on investment analysis, the problem has always been the business concern. But two years later, which seems to still no progress expected in the. From the point of view of the implementation of the priority application, predictability in the position of the first camp did not change maintenance. However, quality control goes beyond predictive maintenance, industrial Internet applications become the most popular business. In addition, the remote deviceMonitoring, production site has become a popular asset tracking applications. Service providers and suppliers to promote the willingness of predictive maintenance, more intense than as a customer of industrial enterprises. Perhaps because compared to remote monitoring, predictive maintenance have greater earning potential, and therefore in a hot vendor quadrant. Many companies are also willing to try and maintenance of equipment related to augmented reality or virtual reality applications, but the number of suppliers and the ability to have significant deficiencies. From the domestic point of view, with foreign somewhat different, but for reference as a whole. Let’s look at the overall situation of the market. Predictive maintenance market, and even the entire Internet industry market, many companies are optimistic about continued.
Comprehensive Gartner, IDC, Machina Research, Cisco, and other multi-Bain analysis data, the Internet industry in the entire area of IoT will inevitably account for a large share of the overall market size in 2021 is expected to double to $ 200 billion . To take advantage of future growth opportunities, many industrial equipment manufacturers and operators are investing heavily and layout. According to statistics S & P Capital IQ S & P Capital database, Siemens, Schneider, ABB and other industrial automation giants are to expand their circle of competence, continue to increase for the cloud platform, edge analysis, M & A and investment integration and other aspects of the software functions and systems. Amazon’s AWS IoT Greengrass and Microsoft Azure also continued to increase its penetration of the industry. In terms of support developer, PTC, Microsoft, IBM, GE and Amazon have the clear advantage. Analysis Although prospects for the future, but the current predictive maintenance market is less than expected but it is an indisputable fact. The reason, primarily 3:00, is a common problem in the field of the whole of things. ROI is difficult to calculate change the business model, have to change the thinking foundation is not solid, not enough data for us one by one. 1, ROI is difficult to calculate the return on investment ROI if calculate exactly, it means slow effect, the effect is difficult to assess, promote the willingness of industrial enterprises will naturally not increase. Industrial components and include scenes, people, machines, materials, methods, ring. The main predictive maintenance and “machine” hooks, which attributes different industries, different types of companies, many types of machinery, predictive maintenance value generated a big difference. From the whole industry chainOn view, it can be associated with the “machine” chain simple and crude analogy with the automotive industry to be. Automotive value chain including suppliers at all levels of the owners, 4S shop, depot, auto parts. “Machine” of the enterprise value chain, including the final application (end-user), equipment service providers (dealers, integrators), equipment manufacturers, various types of industrial automation vendors. Predictive maintenance business around the things enterprises in traditional industries in the chain need to find their own entry point. Predictive maintenance through to end users reflect the value, corresponding to the number of very large companies. Conservative estimates, the domestic implementation of the ERP or supply chain management system, the factory has the capacity of basic information, there are nearly 3 million. If you cut from large enterprises , these companies often choose self-built predictive maintenance capability, even chose to partner with New Enterprise things, due to the predictive maintenance itself involves both software and hardware, networking enterprises are likely to face a high degree of customization, the project is difficult to standardize, not widely copied dilemma. If you cut from the SMEs, it may be even worse. Faced with such a large number of enterprises, how to effectively reach and value of predictive maintenance to provide liquidity, there are many pit there. Industrial system after years of development is already quite mature, a lot of maintenance and repair machinery, the profit space itself is not high. New things companies may find toss all these years, even with a certain size, not be able to really make money. So the face of small and medium enterprises of things in addition to using predictive maintenance, the service links from “passive” to “active”, but also need to be able to provide more depth of services can be based. And from the point of view on the device type, engineering machinery, injection molding machines, CNC machine tools, air compressor … different industry concentration, device manufacturers the ability to provide services is also different. Therefore, the living space left to offer predictive service companies are different things. High-value equipment, or important type of equipment, their maintenance and repair, is more complete by the end users themselves, things rarely outsourced to a service-oriented enterprises. Some type of non-essential equipment, for a long time equipment failure does not occur, or after a failure has time to repair elasticity of equipment, will be outsourced to service providers to provide equipment maintenance and repair services. At this time, things business as a technology provider, between the end-user equipment manufacturers and the two ends, and may be accompanied by dataAsset ownership dispute devices. In addition, the completion of repairs to the transition from traditional maintenance predictive maintenance, but also need a good attitude, as well as time and money. In some cases, the end user does not want to “pocketed” the risk of downtime. They want the device to service providers in the cooperation agreement, to ensure the normal operation of the equipment , if the cut-off loss, equipment service providers need to bear some liability. This naturally filter out some of the company can only provide predictive maintenance on the PPT. At the same time, from the user’s point of view, how much money it is appropriate to pay predictive maintenance services? The benefits of predictive maintenance, if translated into financial indicators, need to go through a complete analysis and periodic verification. Only be regarded as a clear economic accounts, long-term end-users will be willing to pay the value of predictive maintenance. Because predictive maintenance to reduce downtime risks associated with the difficult economic returns linked to a single device is difficult to be linked to overall sales, the value of predictive maintenance is not immediate, need to go through six months or even a year verification cycle, and sometimes need to refine particles per unit of sales. If you encounter more difficult sales affected by environmental changes, fluctuations in end-user accounting. Therefore, predictive maintenance in desperate financial estimates earnings for a long time, the end user can not mobilize investment enthusiasm, but things heat of a business cycle. 2, change the business model, have to change the thinking, if only count the predictive maintenance downtime risk reduction, only count the economic accounts to save money, is not enough. Good business model, not necessarily the end-user help and more money, but to help equipment service providers or equipment manufacturers to make more money. In this process, networking companies to complete their business logic loop. Here to do a simple and crude analogy. Although widely criticized shared bicycle, but also has merit, after all, thanks to Mount changed their travel patterns to share. If we had created friction worship time, do not choose their own bike, but choose only smart lock, put us all useless bike unified management, providing shared travel services to each participant a bicycle idle resources are making money . So it goes, Morocco thanks to the formation of whether today’s development momentum it? the answer is negative. Morocco thanks to the success of bike sharing does not help users save money, but to stimulate the use of bicycles in demand, do not spend money originally walked the road, into riding the bike tour. If there is no uniqueBicycle design, uniform quality service, good travel experience, it is difficult to stimulate a large number of end-user needs for shared bicycles. In the process, thanks to Morocco to become the new “equipment service providers,” to take up the role of the bicycle’s resources remodeling. At the same time, thanks also to accelerate the development of Mount bike “Device manufacturer”, creating an unprecedented large number of devices. Mount worship analogy too, we say back to the industry. Current situation is highly fragmented industrial system, every equipment dealer or integrator, are faced with the dilemma of getting lower and lower profit margins, horizontal integration is likely to increase. Moreover, not all equipment manufacturers, have the ability to provide predictive maintenance. Therefore, the value is not necessarily predictive maintenance is to make every small service providers, equipment manufacturers have to make money, but to have the ability to provide good service, unified experience for equipment service providers and equipment manufacturers to make more money, equipment service from the “passive waiting” mode into “proactive” mode, equipment sales “pay-per-use” from the “one-shot deal” to the model in order to advance the lateral polymerization occurs. But now when everyone talking about predictive maintenance, often still in the money story. By extension, the Internet of things in the end is to save money, or to make money? This is a controversial topic. American Computer Industry Association CompTIA address this issue devoted to a survey. Guess What is the conclusion?
a third of people think that things should save money, one-third of people think we should make money, and one third of people think that to save money and make money are needed. We can see that for the benefit of the application of things there is no consensus. 3, the foundation is not solid, industrial equipment, insufficient amount of data predictive maintenance, are faced with common problems in a slurred, insufficient quantity sensor device itself, a lot of data is not yet effective long-term accumulation. Therefore, predictive maintenance is the most common stories aircraft engines, because the sensor is more than enough to monitor long enough. Why not before equipment manufacturer factory equipment, the installation of more sensors it? Because when things idea has not been popular, the installation of the sensor not only increases the cost, no reason to increase the complexity of the device, there is no clear sense of the application. The device itself is complicated enough, who do not want to mess add some seemingly useless sensor. Sensors are not enough, even if the latest artificial intelligence technology have to spend, not necessarily accurate prediction. AdvanceUncertainty, predictive maintenance on little value. Coupled with the accumulation and iterative device model takes a long time, so in the present application, the “hardware + software + services” for the mainstream charging mode, the embodiment also project-based system. From the fission trigger, too early. Many things companies are looking for ways to solve this problem. Uptake by way of mergers and acquisitions last year, gained control of the company and APT ASL database, is a clear signal. APT Company (Asset Performance Technologies) strategic asset database ASL, arguably the world’s most comprehensive database of industrial equipment failure. Although APT was founded in 2004, but ASL library had began to accumulate, the use of more than 20 years, collected information on nearly 800 kinds of important equipment power, mining, steel and other industries, can provide failure mode effect analysis FMEA and maintenance strategy recommendations. Uptake company was founded in Chicago in 2014, provides industry of artificial intelligence software development and services to help enterprise customers digital transformation. Uptake is currently the most successful cases comes as predictive maintenance and Kaitebile jointly developed platform, to improve the efficiency of its production operations. About Uptake company, it will be mentioned later. There are some things companies are trying to use hardware and software integration, low-cost wireless sensor, from where the original data is not up to the data acquisition, completed from the break 0-1. There are fundamental reasons to achieve this breakthrough 4:00: ubiquitous wireless connection, and the connection costs continue to reduce the size of a large number of low-cost sensors available enterprise cloud computing and edge started to accept the idea of using artificial intelligence collaboration platform to monitor the timing sensor Finally, the data becomes feasible innovation is the ability to see what predictive maintenance startups happen in practice? Many start-ups have realized that the real opportunity is the use of predictive maintenance to create new business. Now decide which companies will break loose too early, we might look at some predictive maintenance status of business survival. Here was giving names to foreign enterprises. 1, Uptake previously mentioned, Uptake was founded in 2014 start-up company, it can be said Uptake large start-up companies in IIoTWinner. It was reported Control Engineering Copyright , Uptake has deployed its “monitoring platform” wind farm in Berkshire Hathaway, a subsidiary of energy company. In the first week of deployment platform, we found the main gearbox bearing failure may occur, and which may cause the operation of a turbine tower is not working. A few hours of downtime will cause the wind farm loss of $ 5,000, and if the turbine is completely collapsed, lost as much as $ 250,000. But recently, General Electric GE to court. 6 Uptake GE executives litigation, are GE’s old unit, primarily from GE Power Generation Group and GE Digital Group. GE’s cause of action is: Uptake premeditated poaching, theft of trade secrets and unfair competition. GE believes that this position caused due six Jiujiang familiar with GE’s commercial secrets Control Engineering Copyright , also participated in the Uptake of “poaching” plan: not just to poach employees from GE, also poached GE customers, trying to maximize GE will be expanded to injury. GE said the poached client list is listed with the Uptake official website. This means that wind power equipment Berkshire Hathaway subsidiary of Energy and Caterpillar locomotive business. Uptake naturally dissatisfied. This year February 22, Uptake counterclaim GE violation of the software license agreement and abuse of trade secrets. Uptake says, GE misuse of company APT in software protocol developed by the Uptake of mergers and acquisitions. GE and Uptake of the lawsuit continues. More seriously, Uptake of experience seems to prove that the opposite is not wise to cut from large enterprises. November 2017, Caterpillar announced that no longer Uptake investment, but retain the Uptake of customer identification. According to the US business magazine “Crain’s Chicago Business,” the report, Caterpillar realized that if continued investment Uptake, would weaken their competitiveness. Caterpillar sales while gradually recover to the control of the monitoring software agents. Caterpillar end users use the software, you can run the testing equipment, implement predictive maintenance. Obviously, Caterpillar had hoped to keep the ability to retract their own predictive maintenance “in vivo.” 2, Predikto Predikto unique advantages that his technology platform capable of handling hundreds of differentIndustrial property datasets, and automate 80% of the analysis process, to quickly draw conclusions from the data analysis, providing practical advice, allow enterprises to take measures in advance to avoid the failure to extend the asset uptime. Predikto was acquired by United Technologies Corporation. 3, KONUX KONUX focused on the railway sector, they combine smart sensors and artificial intelligence-based analysis, maintenance planning and optimization through the use of predictive railways, achieve higher train punctuality and network capacity. The company in February 2019 just completed B round of financing, investors including Alibaba. KONUX The next step is to conduct business in China. 4, Mtell Mtell products mainly to solve the problem of unplanned downtime. He can help enterprises to improve asset utilization, and by predicting equipment failure occurs when accurate, failure to understand why and how to avoid failures provisions, in order to avoid unplanned downtime. Mtell is innovative, a new method of analysis: prescription-type maintenance. Simply put, prescription-type maintenance is like you go to the hospital to see a doctor, first symptom analysis, not only to analyze the current situation, but to look back at history to use, how faults have occurred in the past, and when troubleshooting, according to the the information for final diagnosis. Mtell done by these machines work, not only by the information provided by the sensors, Mtell also view the maintenance history of the past, the past failure, the failure how it happened, why it happened, this information to decide how to resolve the future, to do so a diagnosis of a failure. Currently Mtell AspenTech has been acquired. 5, etasense Petasense also founded in 2014, by monitoring the health of critical rotating machines, providing predictive maintenance solutions. In electricity, petroleum, chemical, metallurgical and other process industries, process control equipment data and vibration data are stored in two separate systems, they are not integrated with each other. Predictive maintenance needs precisely vibration data and operational data automatically close feedback and iteration. In order to solve this problem, Petasense chose to partner with OSIsoft, jointly launched predictive maintenance solution that is being used by power companies in Silicon Valley. It is worth mentioning that, OSIsoft established in 1980, is aHome real-time data management software maker, a subsidiary of PI real-time database system has been widely used in process industries. In 2017, the company was Softbank invested billions of dollars. We like to overestimate five years, but always underestimate decade. The phrase used in predictive maintenance here, it seems very appropriate. In order to promote predictive maintenance floor, in front of business the way things are no shortcuts. Things organizations need insight on industry-specific applications, and provides a more complete end to end solutions for the Internet of Things. Of course, this is not easy.