Big Data and artificial intelligence in activity of SO UES. Interview of TAdviser with Gleb Ligachev

Gleb Ligachev: Informatization of power industry is an adaptation of all most effective and successful solutions existing in the adjacent areas, selection of the steadiest checked, having scaling potential. As for our company - the System operator of the Integrated power grid, it is really specific – first, the we – most computerized, most "tsifrovizovanny" company of power industry having the most saturated history of application of IT in the industry. It is no secret that in supervisory control, along with nuclear sector and meteorology, in 50-60 years of the last century the concept of "industrial informatization" - every possible wide use of IT in productive activity was born the electrical power modes.

In terms of architecture of IT processes, we are the distributed company, we have about 60 dispatching centers for the whole country. At the same time each of them works as a part of a whole, power industry "think-tank", but at the same time has capability to exercise supervisory control in the region, executing a full range of functionality in the area of responsibility, and has such degree of autonomy that it is potentially capable to continue work even in completely standalone mode, in a lead over other dispatching centers.

It means that at loss of communication channels by any dispatching center with the others – for any reason, - such dispatching center should have a technological capability to continue to manage power supply system of the region. And it imposes certain restrictions: for example, on a centralized basis we cannot use cloud computing in the dispatching systems. Each dispatching center is forced to have functionally full range and the equipment, and the managing systems.

There are also additional difficulties in creation of the systems of the System operator. Either the distributed system or centralized is usually created, and we have it uniform and distributed is after all a rarity.

Gleb Ligachev: Our feature is requirements for reliability of our work. They, naturally, unprecedented. When, for example, at failures in a banking system clients half a day or day are not able to perform operations, nothing catastrophic occurs. And as a result of failure in our work, both the whole city, and effects of it can "go out" there will be absolutely others. We cannot relax for a second.

So it is impossible to compare our attitude towards information security to other industries in categories "more strictly/more freely". It at us in effect another. We build the solutions not on assessment of gravity of effects of incidents, acceptable damage, acceptable unavailable time of critical infrastructures, and we recognize that incidents should not be. Zero. At the same time it is not imposed to us by someone or something external. We have an internal conviction that it is important area, and we strongly invest in it.

About how we ensure information security, in detail I cannot tell for clear reasons. And the result is obvious - from all known attacks, we for all this time had no waves of cybernetic attacks of any "adventure". Also the fact that our company wins prizes for the best solutions in the field of information security, and a number of the companies which come to SO UES to get acquainted with our solutions is indicative begin to create the solutions cybersecurity on our scenario.

Gleb Ligachev: In the accepted IT strategy till 2018 all projects connected with functioning of crucial services, plus those which gave economy of future expenses are executed. For example, the project of virtualization of servers allowed to reduce to 2 times the park of the equipment and, respectively, expenses on its renovation in the future.

Auxiliary and service projects faded into the background that is connected with sharp change in the exchange rate of dollar at the very beginning of action of the document. The budget at our company ruble therefore concerning the import equipment our consumer capability significantly decreased.

By this moment we held an open competition and selected the contractor on development of new IT strategy. We are going it to develop for the following half a year.

Gleb Ligachev: As well as last time, we look at 4-5-year perspective. Of course, we clearly see how way of our IT structure and in much longer term – for 15-20 years should change. But planning in 4-5 summer horizon allows to make plans even in the conditions of a large amount of the uncertainty connected with financial and technology features of development of the country in global space.

Gleb Ligachev: Part of these technologies now in the world gains popularity, but for us as I already told, is not new. For example, Big Data. The system operator works with Big Data many decades. The management of power supply system requires processing of a huge number of the arriving indicators of telemetry and teleinformation from the whole country. About 700,000 distance measurements of about tens of thousands of power generation facilities continuously come to the dispatching centers of the System operator. The systems of the dispatching centers make processing of this array of information, including in real time.

As for artificial intelligence, I think, for the present this technology is not at that maturity stage that it could entrust human lives. So far she matures and trains on tasks of entertainments and the analysis of business risks. Consider, to find who can issue the credit some patterns in clients of CRM - such use of artificial intelligence gives the chance something to improve, but at the same time does not bear for itself any serious effects. Let's assume, you offer not those clients not that product. They either will take offense, or will ignore the offer, and nothing terrible, just the company will gather additionally profit.

In turn, in supervisory control the error price at best – millions, tens of millions of losses for the enterprises which were left without power supply. At worst are lives of people which appeared in difficult situations and to which it became impossible to come to the rescue. For example, the frozen city. So to issue the right decision making to artificial intelligence in management of an integrated power grid of Russia is premature today. There will be no clear understanding of the mechanism of its work, audit of the decisions made by it yet, it, probably, in responsible areas will not be able to replace the trained person.

And, certainly, the control object – the Integrated power grid of Russia – needs to change strongly to become capable to management from AI. And on it more resources will be required immeasurably.

Gleb Ligachev: There are specialized systems and algorithms are acquired. It is not artificial intelligence, but mathematical calculation. Telemetered informations come to a mathematical model of power supply system, update its status, and as a result we understand what will be her behavior in case of these or those events and how to solve a problem of deduction of parameters of its functioning in tolerable limits.

Gleb Ligachev: To some extent yes, but she has the feature - meanwhile it is prediktivno possible to count compensation of single faults, but not multiple. And so far even at single unusual failures and the more so at mass, the decision is made by managers who conduct power supply system. This correctness of adoption of solutions by them is built for years of training, trainings and the analysis of real behavior of power supply system of the operational zone. Cultivation of the main manager of the dispatching center takes about ten years. As a result he "on finger-tips" knows where what condition of equipment and as it will behave in this or that situation.

Gleb Ligachev: We practically do not conduct own software development. It is not our function. We – the specialized company managing the electrical power mode in scales of national power supply system. While the majority of our technology systems - very specific, are also used only by us, there is an opportunity to separate hundreds of business processes into clear, algoritmiziruyemy tasks and to order or take the best of available solutions from the market. Therefore we have the technologists setting tasks and using systems in work, and also group of the project managers supervising work of external contractors.

At the same time, we have a number of the small historical self-written systems used generally on communications with specific objects in certain regions. The last years by us it is headed for absorption of their functionality single solutions therefore self-written solutions are washed away rather intensively.

Gleb Ligachev: There is what does not manage to be replaced because of existence of local features. Power objects in regions developed differently, at different technology ways – in the 40th, in the 50th, in the 60th, in the 80th, in the 2000th years, and alike it is not possible to treat them. We conduct a lot of work on unification of processes and solutions, but it is impossible to reach 100%.

In unification of IT infrastructure of the dispatching centers we made significant progress, and in Wednesday, "external" on the relation with SO, we are encumbered with the existing equipment and systems directly on power facilities. It is small illustration of the EEC told earlier about unavailability to pass under the power of artificial intelligence. Nevertheless, the quantity of the self-written systems is significantly reduced every year: instead of hundreds in each branch there will be units.

Gleb Ligachev: Approach to the choice of suppliers strongly does not change as the key technology systems for us, as well as earlier, become the specialized companies. Pass from them to other solutions impracticablly in view of absence for such solutions of sales market even on a global scale.

As for technical solutions, here we try not to hesitate too strongly because support of the high level of readiness of the equipment requires knowledge and skills of the servicing specialists. Reliability of the systems of SO UES is defined, including, and regulating documents according to which all main systems are reserved with a necessary depth. Because of it at us very difficult hardware-software complexes and in order that all their components could work in the mode of duplication and reservation, it is necessary to analyze possibilities of this equipment and software almost to the level of the source code.

The slightest undocumented incompatibility is capable to lead to system failures. Respectively and the people servicing these hardware-software complexes should have high qualification that it is possible to achieve only unification of set of the operated equipment.

Concerning IT service providers, our requirements also do not change: it is the high level of responsibility, readiness, motivation. All requirements of responsibility and penalties for untimely execution register in tough SLA. It is difficult to work with us because it is process, very expensive on time and intellectual efforts, – to penetrate into our specifics. Nevertheless, all of us hold competitions on electronic marketplaces, and in the majority in the form of open competitions and auctions.

Gleb Ligachev: Here the question is rather not in penalties, and in effects of low-quality work. At essential incidents investigation by Rostekhnadzor is made, the prosecutor's office which begin to deal with responsible is connected. It turns out that, without having organized really high level of work, it is already impossible to take to give up performance of the contract – it is necessary to answer also to regulators. So to work with us there are those who are very self-assured and can really perform required functions.

Gleb Ligachev: All system operators of power supply systems of the world have main technological solutions - local, and platform solutions (the equipment and operating systems) – the best of available in global market. Therefore all of us are obliged and we are able to consider those risks that you listed. I will not sound methods of their leveling, but we have them.

This problem has no ready-made, simple, fast and comprehensive solution therefore the failsafe operating system and DBMS suitable for the solution of a wide range of questions, we perceive not as a statement of the problem, and as parameter which needs to be considered when ensuring functioning of our information systems.

Gleb Ligachev: It is impracticable to involve experience of colleagues in connection with uniqueness of the principles of creation of power supply systems. We, certainly, watch trends of development of management systems in System operators, and even annually we book official audit of the innovative development in comparison with others. We try to find benchmarking parameters by which we compare ourselves to foreign system operators including on IT. We can use separate solutions, like creation of monitoring systems of transient regimes, functioning of these or those specialized settlement complexes, but system loan is impossible.

Also we take active part in the International association integrating 19 largest system operators of the world – VLPGO (Very Large Power Grid Operators). There on a number of technology questions we take the leading positions and, it turns out that, on the contrary, other countries by us are guided, but not we by them.