Anomaly detection
Introduction
Anomaly detection is the process of finding problems in street lighting or equipment. The process can be:
- fully automatic: the system generates an incident when an anomaly is detected;
- semi-automatic: the system performs aggregation and preprocessing of raw data. The resulting list contains items that directly or indirectly indicate anomalies. The user makes the final decision;
- manually: the user view raw data and makes a decision.
In most cases, identifying anomalies consists of two steps:
- Automatic preparation of datasets showing anomaly..
- Visual control of the data set by a specialist, rechecking of found anomaly based on raw data. Decision making: is it an anomaly or expected behavior.
To automatically detect anomalies, in most cases it is required to collect and process a lot of data. Therefore, the task of detecting anomalies is not a real-time task. Anomalies detected automatically can be sent to email on a schedule.
Anomalies detected semi-automatically
- Select Anomalies from the main menu.
- The names of the supported types of anomalies will be displayed on the tabs.
«Toxic» luminaries - Nh
Toxic luminaires are luminaires requiring specialist attention. The state of the metrics indicates potential problems. The Reason column displays the reason for including the luminaire in this list.
-
Each line of the table - one mounted on a mounting, controlled luminaire with abnormal performance.
- MAC - unique address of the luminaire head in the radio network;
- Read At - date and time of the last received metric;
- Since - the length of time since the last metric was received;
- Model - luminary model;
- Cabinet - OLCC from which the luminaire is powered. Link to the cabinet page;
- Mounting - the mounting on which the luminaire is mounted. Link to the mounting page;
- Google - opening the point of the mounting location on the Google map (for the possibility of building a route);
- Sockets - a socket in which a toxic luminaire is installed / the total number of sockets on the mounting;
- Radio Network - the code of the radio network that serves this luminaire. Link to the radio network page;
- Reason - the reason for inclusion in the list.
-
The structure of the columns may differ from project to project.
-
When unloaded to Excel, links to object pages are saved.
-
Reasons for inclusion in the list:
- Lost connection: no metrics have been received from the luminaire during the last N hours. It is considered that communication with him has been lost..
- Incorrect power - from the luminaire, in the two metrics in a row received, the value of the power consumption in W at 100% dimming level differs from the nameplate for this luminaire model in either direction (more or less than the nameplate) by N%.
-
The presence of a luminaire in the list does not guarantee its 100% inoperability. The luminaire may continue to function normally and there may be a problem between the luminaire and the central system if there is no communication or with the power measurement sensor.
Blinking radio gateways
A temporary loss of communication with the radio gateway is not a critical problem and may not even be [noticed by the dispatcher](/docs/instructions/dispatcher /real-time-monitoring #radio-networks-monitoring). But the fact of a temporary loss of communication must be investigated.
Radio gateways and luminaires served by a radio gateway can be connected to different phases or even different SLCC. Power supply to the radio gateway later than power supply to the luminaires that it is programmed to service leads to desynchronization between the real state of the luminaires with their statuses in the System.
Long-term lack of communication with the radio gateway is the reason for the inclusion of all luminaires served by the radio gateway in the list of “Toxic luminaires”.
Each line of the table is one hour in the past N hours.
- Hour
- Qty - the number of radio gateways online
- Blinks - the number of on / off events within an hour
Feedback
Was this page helpful?
Glad to hear it! Please tell us how we can improve.
Sorry to hear that. Please tell us how we can improve.