You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

140 lines
7.2 KiB

This file contains ambiguous Unicode characters!

This file contains ambiguous Unicode characters that may be confused with others in your current locale. If your use case is intentional and legitimate, you can safely ignore this warning. Use the Escape button to highlight these characters.

<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>Spatial Systems</title>
<link rel="stylesheet" href="/fullpage/fullpage.min.css">
<script src="/fullpage/fullpage.js"></script>
<link rel="stylesheet" href="/style.css">
<link rel="apple-touch-icon" sizes="180x180" href="/favicon/apple-touch-icon.png">
<link rel="icon" type="image/png" sizes="32x32" href="/favicon/favicon-32x32.png">
<link rel="icon" type="image/png" sizes="16x16" href="/favicon/favicon-16x16.png">
<link rel="manifest" href="/site.webmanifest">
</head>
<body>
<div class="container nav-container">
<div id="wrap">
<div id="row">
<div id="logo">
<a id="logo-container" href="/en/">
Spatial Systems
</a>
</div>
</div>
<div id="navigation">
<a id="navigation-link" href="/en/projects.html">Projects</a>
<a id="navigation-link" href="/en/contacts.html">Contacts</a>
<a id="navigation-link" href="/emergency.html">РУ</a>
</div>
</div>
</div>
<div class="fullpage">
<div class="section gradient-text">
<p>Heat Emergency ————— a geoinformation ML system devoted to modeling of emergencies at heat supply
facilities.
</p>
<p>Every day the city accumulates data on the emergencies that have occurred at heat supply facilities -
residential buildings and heating points. A huge array of such data, combined with data on the
characteristics of heat supply facilities and weather conditions, allows you to create new emergency
management tools.</p>
<p>
HeatEmergency is an application for the city that helps not only improve control to prevent
emergencies,
but also build the best response scenario to an accident that has already occurred
</p>
</div>
<div class="section">
<figure>
<img src="/images/emergency/cover.jpg" />
<figcaption>System interface</figcaption>
</figure>
</div>
<div class="section">
<p>Before using, it is necessary to import all the necessary data into the system — locations of residential
buildings and infrastructure facilities, communication of heat supply facilities with each other,
emergency situations of different types that have occurred in residential buildings, indicators of
general house metering devices</p>
<p>In case of industrial use of the system, all data should be imported in advance, and data requiring
frequent updates (emergency situations and indicators of general house metering devices) should be
transmitted to the system automatically through integrations with the corresponding information systems
of housing and communal services.</p>
</div>
<div class="section">
<figure>
<img src="/images/emergency/import.jpg" />
<figcaption>Import of source data</figcaption>
</figure>
</div>
<div class="section">
<p>After importing the emergency situations data into the system, a training set is formed, consisting of
the emergencies of a certain type in certain residential buildings on specific dates and the data of the
absence of such accidents.
</p>
<p>Besides, many signs are formed for the model, including the type of walls, year of construction, number
of storeys, data on the overhaul of various parts of the water supply system of residential buildings,
etc. (about 50 signs in total), weather conditions.</p>
<p>When building a model, the problem of binary classification is solved — the model determines the
propensity of each residential building to different types of accidents (for example, leakage in a pipe
in an apartment or entrance, lack of heating, etc.).</p>
<p>The result of the predicted propensity of residential buildings is displayed on the map with the
possibility of using many filters.</p>
</div>
<div class="section">
<p>After importing indicators of general house metering devices, the metrics on the pressure and water
temperature in the pipe at the entrance to the house are compared with the standard ones. In case of
deviations from the standards, the fact of an accident on a source — heating point is registered.
Further, similarly to emergencies in residential buildings, an ML model is built, which determines the
propensity of the heating point to an emergency.
</p>
<p>In forecasting, we use complex models that allow us to interpret the propensity for emergencies in a way
that is clear to the user — the forecast can be decomposed into contribution level of all factors
involved in the model and better understand which factors are most significant.</p>
</div>
<div class="section">
<figure>
<img src="/images/emergency/shap.jpg" />
<figcaption>Propensity prediction interpretation — visualization of the contribution of various factors
to the prediction of the propensity of a residential building the occurrence of a leak in the
heating system
</figcaption>
</figure>
</div>
<div class="section">
<p>In the system, it is possible to create a simulation of an accident that has already occurred at a
heating point to assess the number of consumers (residential buildings and socio-economic facilities)
who find themselves in the accident zone and the rate of their cooling down, in order to form an optimal
sequence of response and levelling of the emergency situation.
</p>
</div>
<div class="section">
<figure>
<img src="/images/emergency/sim.jpg" />
<figcaption>Simulation of an emergency at a heating point
</figcaption>
</figure>
</div>
<div class="section">
<p class="gradient-text">The functionality of the HeatEmergency system based on data analysis and machine
learning methods allows you to improve control over the city heat supply system, improve understanding
of patterns in the occurrence of emergency situations and prevent them.
</p>
<a href="/en/contacts.html" class="contacts-link">Contact us if you are interested in a system like
this<span class="arrow-diagonal">
</span></a>
</div>
</div>
<script>
new fullpage('.fullpage', {
navigation: true,
});
</script>
</body>
</html>