Artificial Neural Network System for Emergency Detection Using Multiple Model Data Sources

Abstract

The invention relates to a deep neural network-based system that helps detect anomalies (emergencies, potential security risks, or missing persons). The system is structured to process Geographic Information Systems (GIS) map data, network infrastructure status information, and user location data together. Unusual changes or risky patterns (anomalies) between these different datasets are analyzed using trained artificial neural networks. When the system detects a potential crisis situation, it sends an alert notification, including location information, to the relevant security units or identified individuals.

Description

1 TARIFF Artificial Neural Network System for Emergency Detection Using Multiple Model Data Sources Technical Area 5 The invention enables the detection of anomalies (emergencies, potential safety risks, or missing persons). It involves a deep neural network-based system that provides assistance. State of the Art 10 Today, various systems are used for the detection and management of emergencies. However, Existing systems generally rely on a single data source or different data types. It falls short in terms of integration. This situation prevents a rapid and accurate response to emergencies. This makes it difficult to detect, leading to an increase in false alarms or delays in critical situations. This leads to it being overlooked. In particular, geographical risk factors, network status, and Analyzing user behavior together allows for more effective response to emergencies. It will ensure that it is managed. In conclusion, due to the negative aspects described above and the current solutions regarding topic 20 Due to its shortcomings, it has become necessary to make improvements in the relevant technical field. Purpose of the Invention The invention was created by drawing inspiration from existing situations and addressing the aforementioned drawbacks. 25 It aims to solve the problem. The main purpose of the invention is to detect anomalies (emergencies, potential safety risks, or missing persons). The goal is to provide a deep neural network-based system that helps with detection. Another purpose of the invention is to analyze Geographic Information Systems (GIS) map data and network infrastructure status. The goal is to provide a system that can process user information and location data together. Another aim of the invention is to use trained artificial neural networks to correlate information between different datasets. 35 A system capable of analyzing unusual changes or risky patterns (anomalies) The purpose is to notify the relevant security units or designated individuals when a potential crisis situation is perceived. 2 The goal is to develop a system that sends an alert notification including location information; at the center of the crisis. The fastest search and rescue via CBS that can calculate the base and the number of people affected. The goal is to provide a system that can autonomously generate its own routes; location, affected person estimation, Targeted recovery emergency package including network outage map and autonomous routes The goal is to provide a system that can instantly transmit information to its units. 5 The structural and characteristic features and all the advantages of the invention are given in the figures below. The detailed explanation, written with references to figures, makes it clearer. This will be understood. Therefore, the evaluation should also take these figures and detailed explanations into account. It must be done by taking 10. Figures that will help understand the invention. Figure 1 shows the general architecture of the system that is the subject of the invention. Description of Part References 15 1. CBS database 2. Network monitoring server 3. User mobile device 4. Data preprocessing and alignment module 5. Artificial neural network processing engine 20 6. Graph-based network analyzer 7. Temporal telemetry classifier 8. Spatial decision fusion network 9. Anomaly detection and decision threshold controller 10. Automatic emergency notification and dispatch coordinator 25 11. Emergency management and security units interface Detailed Description of the Invention In this detailed explanation, the preferred configurations of the system in question are only 30 This is explained to facilitate a better understanding of the subject. In this detailed description, the preferred configurations of the system in question are only those related to the subject matter. It is explained in order to facilitate a better understanding. 35 3 The system described in the invention detects anomalies (emergencies, potential security risks, or missing persons) It is a deep neural network-based system that helps in detection. The system uses Geographic Information. The systems (GIS) combine map data, network infrastructure status information, and user location data. It is structured to work in between. Using trained artificial neural networks, this different data... Unusual changes or risky patterns (anomalies) between the sets are analyzed. The system has 5 When it detects a potential crisis situation, it provides the location to the relevant security units or designated individuals. It sends a warning notification that also includes this information. In the preferred configurations of the system that is the subject of the invention; • Risk maps of geographical areas (fault lines, flood basins, etc.), building densities and 10 storing geometric data of safe assembly areas and providing spatial references to the system. CBS database (1), • Active user numbers and real-time traffic density of base stations in the cellular network, Network monitoring server that simultaneously monitors signal quality (SNR) and packet drop rates. (2), 15 • With the help of sensors such as a GPS receiver, accelerometer, barometer and gyroscope, it enables the user to... User mobile device that generates and transmits instantaneous speed, location and altitude data (3), • CBS (Geographic Information Systems) that arrive at different time intervals and in different formats (vector, numeric, time series), synchronizing network and telemetry data on a common geotemporal index data preprocessing and alignment module (4), 20 • Performs feature extraction from multimodal inputs and uses these features in deep learning. artificial neural network processing engine (5) that generates a holistic risk score by combining layers, • A base station that models the telecommunications infrastructure as a graph (nodes and edges). Graph Convolutional Networks detect collapses and unusual load transfers at neighboring stations. Graph-based network analyzer (6), which detects with (GCN), 25 • Sequential acceleration, elevation, and position data from mobile devices are processed by LSTM / GRU layers. by processing through anomalies such as "sudden fall", "uncontrolled escape" or "immobility" capturing temporal telemetry classifier (7), • The final layer of the artificial intelligence engine, feature vectors from GCN and LSTM. GIS combines the geographic risk factor with spatial decision 30 to convert it into a crisis probability score. fusion network (8), • By comparing the generated fusion risk score with a dynamically changing threshold value, emergency situations can be identified. Anomaly detection that makes status decisions and prevents false alarms by going beyond static limits. and decision threshold controller (9), 4 • As soon as an anomaly is detected, the CBS (Computer-Based System) calculates the epicenter of the crisis and the number of people affected. automated emergency system that autonomously creates the fastest search and rescue routes. notification and guidance coordinator (10), • Autonomous crisis alert from the coordinator, map-based location data, and evacuation Emergency management system 5 that relays routes to rescue teams (AFAD, Police, Fire Department, Ambulance, etc.) and security units interface (11) It is located. The working principle of the system in question is as follows: CBS database (1), network monitoring The heterogeneous raw data streams from the server (2) and user mobile devices (3) are simultaneously processed. It is listened to as follows. Raw data is purged of noise, missing location data is interpolated, and The data are aligned in a common space-time matrix. Synchronized grid topology, graphene anomalies between base stations are processed by the network analyzer (6) based on the anomaly between base stations patterns are extracted. Time series telemetry stream from mobile users, temporal Individual shock / acceleration anomaly score 15, analyzed with telemetry classifier (7). The resulting network and user anomaly features are extracted. The spatial decision fusion network (8) These are multiplied by CBS risk coefficients and combined to produce the fusion anomaly score. Anomaly detection and decision threshold controller (9) determines the historical baseline of the relevant location. According to the data, a specific dynamic tolerance limit is calculated for that moment. The calculated fusion anomaly. The score is then subjected to a mathematical comparison with the dynamic decision tolerance threshold. Comparison 20 If the score exceeds the threshold, an autonomous crisis alarm signal is generated and the crisis center is activated. The coordinates are determined. This is based on the geographical features of the crisis area (GIS data) and current road conditions. According to the system, the safest and fastest routes for rescue teams are calculated autonomously. Location, Emergency package including affected person estimate, network outage map and autonomous routes, The information is instantly transmitted to the targeted emergency management and security units interface (11). 25

Claims

REQUESTS 1. It is a deep neural network-based system that helps detect anomalies; its feature is: • Risk maps of geographical areas, building densities and safe assembly areas GIS database 5 that stores geometric data and provides spatial reference to the system (1), • Active user numbers and real-time traffic of base stations in the cellular network. A network that simultaneously monitors density, signal quality, and packet drop rates. monitoring server (2), • With the help of sensors such as a GPS receiver, accelerometer, barometer and gyroscope, it can detect 10 User mobile device that generates and transmits the user's real-time speed, location, and altitude data. (3), • CBS, network, and telemetry data arriving at different time intervals and in different formats. Data preprocessing that synchronizes data on a common geo-temporal index. alignment module (4), 15 • Performs feature extraction from multimodal inputs and deep-analyzes these features. artificial neural network processing that generates a holistic risk score by combining learning layers engine (5), • Modeling telecommunications infrastructure as a graph, including base station failures. and unusual load transfers at neighboring stations using graph convolutional networks 20 graph-based network analyzer (6), • Sequential acceleration, altitude, and position data from mobile devices are processed by LSTM / GRU by processing through its layers "sudden fall", "uncontrolled escape" or "immobility" Temporal telemetry classifier (7) that detects anomalies such as • The final layer of the AI ​​engine, feature 25, derived from GCN and LSTM. converting vectors into a crisis probability score by combining them with the CBS geographic risk coefficient. spatial decision fusion network (8), • By comparing the generated fusion risk score with a dynamically changing threshold value, emergency situations can be identified. anomaly that makes a decision about the situation, preventing false alarms by going beyond static limits Detection and decision threshold controller (9), 30 • As soon as an anomaly is detected, it calculates the epicenter of the crisis and the number of people affected. The automated system creates the fastest search and rescue routes autonomously via CBS. Emergency notification and guidance coordinator (10), • Autonomous crisis alert from the coordinator, map-based location data, and evacuation Emergency management and security units interface that transmits routes to rescue teams 35 (11) is included.

Images