Denis Kasimov, Clover Group: production workers began to understand benefit of forecasting of breakdowns of the equipment

Denis Kasimov: If to remember history, at the beginning of the 20th century repaired what already broke, then began to make preventive repair (in the Soviet terminology - the Party of Pensioners of Russia, scheduled preventive maintenance), and at last – repair on actual state, with forecasting and warning of breakdowns. It is necessary to understand that one, special, magic technology which would make only the emergence possible forecasting of breakdowns of industrial equipment does not exist. That it became possible, the whole pool of technologies, techniques and so on had to appear. And the first necessary condition - the sufficient power of ADP equipment which made possible processing and the analysis of large volumes of information. The second condition – the necessary level of development of mathematical algorithms. The third component – mental readiness of customers and society in general to apply technologies of predictive data analysis of industrial equipment. And, at last, the fourth necessary condition for implementation of the predictive analysis of a condition of equipment and its repair on a status – rather high level of industrial automation assuming use of the modern equipment equipped with telemetric sensors, and SCADA systems.

Denis Kasimov: The idea appeared at me for a long time - ten years ago when I still was a student of the Kazan power university. Along with study I actively worked in consulting, in the field of hydropower and metallurgy, implementing methodology of RCM (Reliability-Centered Management, management of reliability and security of the equipment). One of my tasks as consultant was – to come to production, finding out on site what breakdowns of the equipment happen more often how the equipment is diagnosed then to systematize the gained knowledge, to put them in a certain methodology. Then I was also convinced that diagnostics on production was most often performed at the intuitive level, literally – the experienced specialist puts a palm on the compressor, and to the touch defines how bearings as far as they are operational work. And then I realized that there is a big gap between production, and those who sit in the management company most of which often is in Moscow. It leads to untimely repair, the increased accident rate. And one of tragic confirmations to that is accident on Sayano-Shushenskaya HPP in 2009 when did not keep track of technical condition of this ill-fated hydraulic unit No. 2. Then, at the university, in parallel with work I studied a neural networks rate, and understood that neuronets - that technology which it is possible and need to be applied to forecasting of repair on a status. As a result I wrote the thesis on assessment of technical condition of the transformer on the basis of technology of neural networks. At that time I had no managerial experience yet, eight years saved it, understood two years ago that I am capable to create business.

Denis Kasimov: In the created type of a segment of a similar class in the Russian market not yet, it only arises today, and represents that is called - the "blue ocean" open for all players. So far only the few companies made products for the forecast of technical condition of the equipment, capable to this open ocean "come up". It is natural that such giants as SAP, IBM, Yandex and some other vendors in this direction develop, some of them already have in a portfolio corresponding specialized products.

Denis Kasimov: They are provided, but complete implementations, as far as I know, not yet. World IT market, by the way, on this direction not too outstripped Russian in the development.

Denis Kasimov: Formation of this segment of software, certainly, is connected with a current status of the Russian economy, but much more it is connected with a global trend of digitalization of economy. The trend on digitalization became obvious to all including for most the large Russian enterprises, there was a corresponding change of business models. Production workers realized an economic benefit of implementation and use of the systems of the class PMM (Predictive Maintenance & Monitoring) which reduce hardware maintenance cost, provide reliability and the continuity of production that is especially critical, for example, for the sphere of power. In general it is possible to note that equipment analysis systems using technology of neuronets – one of elements of the big digital world in which we should live.

Denis Kasimov: One of Clover platform modules – Clover PMM (Predictive Maintenance & Monitoring) which consists of three blocks. The lowermost of them - the ETL level (Extract Transform Load) where there is everything that is connected with unloading, storage, data processing, including distributed computing and the data warehouse. This level is written on Java and Scala. The following level of the module - mathematical models, logic, algorithms of calculations, a programming language - Python. And the third component - the visualization level at which information obtained in an analysis result of data is provided in a type, convenient and evident for the user, is written on JavaScript. As DBMS we use Apache Cassandra.

Denis Kasimov: The Clover platform is brought to the market as a commercial product. In our portfolio there are already both commercial implementations, and pilot projects. At the same time, the platform is able to do not all. The full-function version of the platform is in process of completion as a system is still not capable to solve equally well all problems, to analyze and predict repair of any kind of the equipment.

Denis Kasimov: The platform is capable to work in any industry, with any equipment, and in this plan it is universal. The universality is provided by special mathematical modeling under each type of production and the equipment. Existence of a preliminary stage of setup of a mathematical model – the compulsory provision of implementation of the platform, otherwise the technology will not earn. It not boxed solution.

Denis Kasimov: The first task – intellectual diagnostics of industrial equipment. Usually the operator at best traces an equipment parameter value, controlling that they were in the normal range designated for each parameter by the producer. However such approach does not consider internal interrelations between parameters. The intellectual analysis based on use of a special mathematical apparatus allows to reveal the hidden abnormal operation modes of the equipment. The second important task – to define what of nodes of the equipment will break, and with important detailing – when why at what operation modes, i.e. to predict its breakdown.

Denis Kasimov: By and large sources are ICs of two types: low-level SCADA system and accounting systems. From the first system we take indications of telemetric sensors of the equipment, from the second – data on the recorded defects which are carried out repair and so on. In the amount this information also allows to predict breakdowns. Problems in respect of integration with information systems do not arise, we have examples on integration and with same 1C, and with products of the western vendors, and with self-written solutions.

Denis Kasimov: Potential customers of our platform - the capital productive productions of any industry having the big park of the equipment which should be managed effectively. So representatives of an Enterprise segment if to follow business terminology.

Denis Kasimov: Potentially, I think, it is interesting, but at the current stage of market development, it will be still too expensive solution for the enterprises of a segment of SMB.

Denis Kasimov: The service enterprise guarantees to the customer that the repaired equipment will work during guaranty period. Respectively, it is correct to understand, the equipment how exactly works, to predict when it breaks, for such companies very important. Some service enterprises already became our clients. As for equipment manufacturers, a feedback about operation of the equipment is important for them to improve its parameters, to increase reliability.

Denis Kasimov: Last year we won the tender for the project implemented in the service company "Locomotive Technologies" (Lokotekh) servicing the park of locomotives of the Russian Railway. This project became one of our first big implementations. It started in December, 2016, and now is in very active stage. The Lokotekh company has 92 service depots in different regions of the country in which on service there are 11 series of locomotives. As of today in 20 depots, within an implementation project of our platform, on trial operation there are 4 series.

Denis Kasimov: At a stage of trial operation experts obtain information from sensors of the locomotive delivered in depot then load data into our system, and that shows to specialists where problems are possible and where exactly it is necessary to look especially attentively. In September the second part of the project when a system predicts, what node will break within the next few hundred hours will be implemented. Project completion is planned until the end of the current year, since the beginning of the next year – commercial operation. It is how possible to judge by publications in the press, it will be the Russia's first implementation project of the systems of a similar class in commercial operation.

Denis Kasimov: On the transport direction the project on electric trains, similar to about what I told soon starts. Besides, one of our clients - the foreign producer of compressors having the Russian representative office. Compressors are on their complete service maintenance, they guarantee almost continuous operation of the equipment and, respectively, the producer undertakes all risks connected with breakdown. Therefore the company is very interested in use of our solution. At a pilot stage of the project we predicted that one of compressors will be switched-off in three weeks because of jumps of temperature in the cooling system that helped vendor to prevent its abnormal stop. One more example – on one of state district power plant we analyzed operation of the burned-down generator in two years and saw that it was possible to notice malfunctions in its starter in one and a half years before the generator burned down. And by that to prevent its breakdown and also expenses on major repair, components, we will notice, about 150 million rubles.

Denis Kasimov: The headquarters of our company is in Kazan, there is a branch in Moscow, in the state about 60 people. A state core – mathematics (data scientist), the building mathematical models, and engineers - the industry experts who are very well understanding operation of the equipment, physics of processes. Further there are developers, analysts, project managers and specialists entering commands of projects.

Denis Kasimov: Our main objective for the next period – to fill the Clover platform with mathematical models became possible as fast as possible to start the analysis practically of any industrial equipment. The second important direction of development of the platform – reduction of the threshold of an input when developing new solutions based on our platform. At some point we would like to give implementation to partners – system integrators, and to be engaged exclusively in development of the platform. It is important for us as software vendor.

Denis Kasimov: The partner network as such is absent so far, but there are partner companies with which we together develop the market of the industrial Internet in our country. For example, the consulting companies are available in partners. It is important not just to provide to business value, but also the nobility as to apply it – for example, to change logistics of repair of the equipment according to new opportunities regarding exact forecasting of breakdowns. Partners – konsalter and system integrators can also help with it just.