BEHAVIORAL MODEL-BASED ADAPTIVE ANOMALY DETECTION IOT-SUPPORTED AUTOMATIC WATER SHUT-OFF SYSTEM AND METHOD

Abstract

The invention relates to an IoT-supported automatic water shut-off system and method that performs behavioral model-based adaptive anomaly detection by learning user-specific water consumption behavior in terms of time, flow rate, and usage probability via a flow sensor and motorized valve connected in series to the main water line of a residence or workplace, and detecting abnormal continuous flows using dynamic thresholds and automatically shutting off the water supply.

Description

1 TARIFF IoT-powered system for behavioral model-based adaptive anomaly detection. AUTOMATIC WATER SHUT-OFF SYSTEM AND METHOD Technical Area 5 The invention relates to an automatic water shut-off system and method. The invention specifically relates to a flow sensor connected in series to the main water line of a residential or commercial building, and Through the motorized valve, the user-specific water consumption behavior is determined by time, flow rate, and Dynamic thresholds of abnormal continuous flows are learned through the axis of probability of use. Behavior 10, which enables detection and automatic water shut-off using an IoT-supported automated water shut-off system that performs model-based adaptive anomaly detection. and is related to the method. State of the Art Nowadays, leaving the tap running in homes, plumbing malfunctions, or small leaks are common occurrences. The continuous water flows caused by this result in property damage, water waste, and safety risks. 15 This results in systems based on static time / flow rate / volume thresholds, on a household basis. because it cannot adapt to changing usage habits (a) normal use (b) unnecessary closures under the assumption of leakage and (b) delays in actual leakage situations This can lead to intervention. In the current technology, to reduce unwanted water flows in residential water installations, (i) 20 Water leak sensors located near the ground / device (alarm / shutoff on water contact), (ii) static time / volume that shuts down when a certain time or volume threshold is exceeded (iii) IOT valve that provides manual shut-off via mobile application based solutions commercial solutions that detect leaks using systems and (iv) pressure wave / flow analysis. is used. Fixed thresholds are generally used in such solutions. User 25 Dynamic adaptation is limited according to habits. Depending on whether the house is full or empty. personalized behavioral model aimed at differentiation or reducing false alarms It is not available. Other known approaches in current technology include placing it near the ground or device. Leak sensors that are positioned and give an alarm or cut off the water supply in case of water contact. 30 2 These sensors are located on the ground and react after the water actually reaches the surface. intervention is delayed because they operate independently of user habits. They are unable to detect small and slowly forming leaks. In more advanced systems... However, through flow sensors and motorized valves mounted on the main water line, a certain amount of water is obtained. Static time or volume-based 5 that cuts off the water when a time or volume threshold is exceeded. Solutions are used. In such systems, threshold values ​​are usually factory set. It is either provided as is or manually determined by the user. With this... Together, these static thresholds reflect the varying water consumption habits of households throughout the day. usage intensity varies depending on the time of day, or between weekdays and weekends. It fails to take consumption differences into account. This situation is a normal shower or dishwashing 10 a high-flow and long-lasting but entirely expected flow, such as that of a machine This event is classified as abnormal by the static threshold, leading to unnecessary water outages. This leads to a situation where, conversely, a low-flow but continuous leakage occurs. In this case, the time required to overcome the static thresholds is very long, which... This leads to delayed intervention in terms of material damage and water waste. In recent years, 15 The developed IoT-based valve systems allow the user to manually access the valves via a mobile application. It offers the option to turn off the power and provides real-time consumption information. However, these systems... Most of them leave the assessment of the flow anomaly and the decision for automatic intervention to the user. They either abandon it or still use simple static thresholds. Some commercially available ones Solutions that perform pressure wave or flow analysis determine the pressure throughout the entire plumbing system of the house. They try to detect leaks by monitoring changes, but these systems are high-performance. It is costly, involves installation complexity, and also requires dynamic user behavior. Adaptation is limited. All these existing systems share common technical characteristics. The problem is their inability to learn user-specific water consumption behavior, time, flow rate, and They are unable to generate dynamic thresholds that jointly assess the likelihood of use, and therefore 25 Their inability to distinguish between normal operation and abnormal continuous flow with sufficient accuracy. This deficiencies, leaving the tap running in homes, plumbing malfunctions or small leaks The continuous flows caused by this situation are both noticed late and cause false alarms. leading to high rates of pollution, resulting in property damage, water waste, and safety risks. It creates. In the current technology, there is also a threshold of 30 depending on whether one is at home or not. changing sensitivity, updating user profile over time, or different An integrated system that performs adaptive anomaly detection using learning algorithms. It is not available. This invention aims to overcome all these limitations in technology by using the main water. The installation is built on a flow sensor and motorized valve connected in series to the line. Afterwards, it learns the user's typical water usage patterns, analyzing each flow event by time, flow rate, 35 3 Generating dynamic thresholds by evaluating them in terms of duration and usage probability, and abnormal an adaptive model-based system that automatically cuts off the water when it confirms a continuous flow. The system offers a system that allows for the comparison of normal usage with real anomalies. The distinction between them is made with high accuracy, false alarms are minimized, and Water waste significantly reduces the risk of material damage. 5 Patent document number CN119803598A describes a smart water meter and water leak detection system. The system is being discussed. The invention concerns a water consumption monitoring module and abnormal water consumption. Internet of Things and smart home, including warning module and water leak detection module. It offers a system in the field of technologies. Abnormal water consumption warning module, 10 While used to obtain abnormal parameters, the water leak detection module monitors water flow. It detects parameters and collects flow rate and volume data via a flow sensor. The system collects data and alerts the user if the specified threshold values ​​are exceeded. It provides and interrupts the water flow. In the present invention, however, unlike this system, There is no warning and cutoff mechanism based solely on static thresholds, but also provides the user with 15 Specific water consumption behavior in terms of time, flow rate, and usage probability. Dynamic thresholds created through learning are used, thus simplifying normal usage. The distinction between abnormal and sustained flow is made with much higher accuracy. Patent document number WO2026064878A1 describes automatic water leak detection in building units. The text refers to a smart water valve device designed for detection and management. The invention consists of a water inlet, a water outlet, a shut-off valve, a water flow sensor, and a pressure sensor. It includes a sensor, and all these components are operationally connected to a control unit. It is connected. The control unit monitors water usage and detects abnormal flow events. advanced algorithms, including AI-based leak detection, to do this 25 It uses pressure and automatically shuts off the water when a leak is detected. It offers more comprehensive monitoring by combining data from flow sensors. Patent document number AU2021105136A4 describes using the Internet of Things to develop water. The text describes an AI-based smart meter for usage and quality monitoring. 30 The invention provides details of water consumption using a water flow sensor, without any It detects leaks and monitors water quality via a turbidity sensor. The system aims to address the shortcomings of existing meters and to improve consumption. It provides its data to the user via a cloud platform. 35 4 Patent document number CN205721254U describes an Internet of Things application. The system being discussed involves a light controller, gas, etc., inside a dwelling. a controller, a system that integrates a smart water meter and a smart electricity meter It offers. The smart water meter has a built-in flow detection device and It also includes a pulse counter that allows users to view their water consumption data. 5 It is stated that the system is known to the public. The system protects lives and property when no one is home. It also includes features aimed at ensuring security. Patent document number CN204989953U describes an Internet of Things application. The system is being discussed. The invention includes an access control device, a video intercom, and a solar 10. energy-powered water heater controller, light controller, electrical appliance controller, automatic Anti-theft curtain, warning control device and wireless transmitter-receiver device It includes a smart water meter microprocessor that works in an integrated manner. Smart water The counter microprocessor is connected to a flow measuring device and a valve control system, It collects water consumption data and can control the valve under specific conditions. 15 Investigations revealed unwanted water leaks in residential and commercial water plumbing systems. Detecting and repairing continuous flows caused by leaks or taps left open. In existing systems aimed at preventing this, sensors that are in contact with the ground are generally used. Valve systems operating with static time or volume thresholds, or simple IoT-based 20 Manual closing mechanisms are observed to be used. These systems share common technical characteristics. The deficiency is their inability to learn user-specific water consumption behavior and their understanding of time, flow rate, and flow. Dynamic thresholds that consider multidimensional variables such as duration and time of day together. Their inability to produce. Static thresholds, depending on the varying usage habits of households. Because it cannot adapt, using a normal shower or washing machine becomes abnormal. 25 This can lead to unnecessary water interruptions or low but continuous flow rates by sensing the problem. Delayed intervention in the event of a leak can cause material damage. and causes water waste. In addition, some advanced systems use artificial intelligence to detect leaks. Although detection algorithms are used, these approaches generally rely on instantaneous pressure or flow. It is based on data and, over a long-term learning period, the user's 30 not profiling by extracting typical water usage patterns and being at home Since it is not present, it does not offer an adaptive structure that changes the threshold sensitivity. Existing systems also take into account water quality or other home automation elements. because the focus is on the early detection of abnormal continuous flows directly and solely. Sufficient technical solutions have not been developed for automated intervention. 35 In conclusion, due to the negative aspects described above and the current solutions being the subject of discussion... Due to its shortcomings, an improvement is needed in the relevant technical field. It has been made. Purpose of the Invention 5 The invention was created by drawing inspiration from existing situations and overcoming the aforementioned drawbacks. It aims to solve the problem. The main purpose of the invention is to resolve abnormal continuous water flow occurring in the main water line of a residential or commercial building. dynamic water flows through learning user-specific water consumption behavior The aim is to enable early and accurate detection using thresholds. In this way, 10 Normal usage commonly encountered in traditional systems based on static thresholds mistakenly classified as abnormal or late in actual leakage situations By addressing these issues and eliminating problems, the risk of water wastage and material damage is minimized. It is being downloaded. Another purpose of the invention is to connect a flow sensor and a motorized valve in series with the main water line. 15 The goal is to offer an IoT-supported automated water shut-off system built upon this foundation. This integrated system... thanks to the structure, it is mounted on the ground and waits for the water to actually reach the ground. Real-time leak detection throughout the entire installation, without the need for classic leak sensors. This enables monitoring and immediate intervention, thus supporting small and slow-growing businesses. Even smugglers can be detected and prevented. 20 Another objective of the invention is to generate a probabilistic anomaly score for each flow event, high flow and long-lasting but fully expected functions such as showers or dishwashers. abnormalities caused by usage, taps left open, or plumbing malfunctions. The goal is to enable highly accurate differentiation between continuous flows. Another aim of the invention is to provide 25 different operating modes, such as away from home or remote mode. According to this, it can change the sensitivity of the dynamic thresholds, so that no one in the house If it is not present, the water will automatically shut off even at the slightest abnormal flow. while ensuring that normal life continues, unnecessary water outages are prevented. and thus ensures increased system reliability and user satisfaction. The goal is to create systems and methods. 30 6 Another purpose of the invention is to regulate water flow via a motorized valve in the event of abnormal flow detection. automatic interruption and simultaneous notification via mobile application, and This ensures the user is immediately notified of the event, and can also be remotely re-contacted. By allowing activation or manual override, the user can disable the system except in emergencies. control and traceability are ensured by recording all events with timestamps. 5 is to ensure. Another objective of the invention is to improve the flow sensor type, learning algorithm, and communication. By offering a modular architecture in terms of protocol, it accommodates different installation conditions and costs. The aim is to provide adaptability to their requirements. This allows for the use of hall-effect turbines or... ultrasonic sensors, statistical models or deep learning algorithms, Wi-Fi, 10 A flexible system that can work with different protocols such as BLE, Zigbee, GSM or NB-IoT. They are obtained in such a way that they are used for everything from residences to offices, hotels to dormitories and small-scale projects. It can be used in a wide range of applications, including industrial production lines. Another purpose of the invention is to monitor agricultural irrigation lines, water reservoirs, and summer houses remotely. Our aim is to offer a water management system that can be applied in areas such as these. In this way, we can continuously provide 15 occurrences in facilities that cannot be monitored or are visited periodically Water wastage and financial losses that may occur due to leaks or valves being left open. The risk of damage is minimized, water resources are protected, and sustainable water management is ensured. Contributions are made to its management. To achieve the objectives described above, the invention is based on a behavioral model. It is an IoT-supported automatic water shut-off system that performs adaptive anomaly detection. Accordingly system;  Connected in series to the main water supply line of a residence or workplace, and instantaneous fluctuations in the line a flow sensor that measures flow rate, volume and flow time data,  sampling data from the aforementioned flow sensor in real time, 25 a microcontroller that operates and executes control decisions,  running on the aforementioned microcontroller and determined after installation During the learning period, the user's specific water consumption profile is determined by time, flow rate, and stream. A behavior that extracts data based on duration, time of day, and day type attributes. modeling module, 30  each according to the profile extracted by the aforementioned behavioral modeling module a probabilistic analysis of the new flow event based on time, flow rate, and duration window 7 a dynamic threshold and decision system that generates an anomaly score and sets dynamic thresholds module,  Anomalous flow detected by the aforementioned dynamic threshold and decision module a motor that physically cuts off or reopens the water flow when verified. valve, 5  running on a user device, in case the user detects an anomaly receiving notifications and remote monitoring, reactivation, and manual override a mobile application that allows users to enter commands,  Anomalous flow detection by the aforementioned dynamic threshold and decision module When this happens, it sends a notification to the user via a mobile application, remotely 10 Receiving monitoring, reopening, and manual override commands. an IoT communication module that provides,  Anomalous flow events can be detected in time via the aforementioned IoT communication module. It records all alerts and user interventions with a stamp, logging them. an event log and notification module 15 It includes. The invention also includes an IoT-powered system that performs behavioral model-based adaptive anomaly detection. It also includes an automatic water shut-off method. Accordingly, the method is as follows:  Instantaneous flow rate, volume and flow duration data in the main water supply line flow rate 20 continuous measurement by the sensor and transmission to the microcontroller,  Real-time sampling of transmitted data by the microcontroller and processing,  Behavioral modeling during the learning period determined after installation Running the module, 25  User-specific water consumption profile by the behavioral modeling module Extraction based on time, flow rate, flow duration, time of day, and day type attributes. and its storage in local memory,  Instantaneous snapshots of each new flow event after the completion of the learning period The data, along with the extracted user-specific water consumption profile, includes dynamic thresholds and 30 forwarding to the decision module,  a probabilistic analysis based on time, flow rate, and duration window for each flow event the anomaly score is generated by the dynamic threshold and decision module and the generated The probabilistic anomaly score is dynamically determined based on the user's specific water consumption profile. Comparison with the threshold value determined as 35 8  Dynamically adjusting the threshold value of the probabilistic anomaly score over a specific period of time If it exceeds this limit, the flow phenomenon is classified as abnormal sustained flow.  Dynamic threshold for classifying the flow phenomenon as anomalous continuous flow and the decision module sends a closing command to the motorized valve and the water physically interrupting the flow via a motorized valve, 5  Transmission of abnormal flow detection information to the IoT communication module (6) and mobile Sending instant notifications to the user via the application,  Remote monitoring, turning the water back on or off via mobile application manual override commands are sent to the system via the IoT communication module. transmission, 10  incidents of abnormal flow events, user interventions, and system decisions recording and timestamping by the registration and notification module,  Dynamic threshold and decision module threshold depending on the system's operating status. changing sensitivity It includes the steps of the process. 15 The structural and characteristic features and all the advantages of the invention are given in the figures below. This becomes clearer thanks to the detailed explanation written with references to these figures. This will be understood as such, and therefore the evaluation will also take these forms and detailed explanations into account. This should be done taking this into consideration. 20 Figures that will help understand the invention. Figure 1 shows a schematic representation of the system that is the subject of the invention. Explanation of Part References 1. Flow sensor 2. Microcontroller 25 3. Behavioral modeling module 4. Dynamic threshold and decision module 5. Motorized valve 6. IoT communication module 9 7. Event logging and notification module 8. Mobile application Detailed Description of the Invention This detailed explanation describes the preferred system and method for the invention. Their structures are explained solely for the purpose of better understanding the subject. 5 The invention is an IoT-supported device that performs adaptive anomaly detection based on a behavioral model. An automatic water shut-off system occurs when a problem arises on the main water supply line of a residential or commercial property. early and accurate detection of abnormal, continuous water flows and automatic It is based on the principle of intervention. The invention is an IoT-powered automated water detection system that performs adaptive anomaly detection based on a behavioral model. It is a cutting system. Figure 1 shows a schematic representation of the system that is the subject of the invention. The system is provided accordingly; it is connected in series to the main water supply line of the residence or workplace. a flow sensor (1) that is connected and measures instantaneous flow rate, volume and flow time data in the line, sampling data from the mentioned flow sensor (1) in real time, a microcontroller that operates and executes control decisions (2), 15 running on the mentioned microcontroller (2) and determined after installation During the learning period, the user's specific water consumption profile is determined by time, flow rate, and flow duration. a behavior modeling module that extracts based on the time of day and day type attributes (3), According to the profile generated by the mentioned behavior modeling module (3), each new A probabilistic anomaly score of 20 is calculated based on time, flow rate, and duration window for the flow event. a dynamic threshold and decision module (4) that generates and determines dynamic thresholds when abnormal flow is confirmed by the mentioned dynamic threshold and decision module (4) a motorized valve (5) that physically cuts off or reopens the water flow, a user running on the device, allowing the user to receive notifications in case of anomaly detection and 25 mobile application (8), by the aforementioned dynamic threshold and decision module (4) When abnormal flow is detected, the user is notified via a mobile application (8) sender, remote monitoring, reopening and manual override commands an IoT communication module (6) that enables the receiving of the mentioned IoT communication all 30 which record abnormal flow events with timestamps via module (6) an event log and notification module that logs warnings and user interventions (7) It includes. The system works by connecting the flow sensor (1) in series to the main water supply line. is started; the aforementioned flow sensor (1) collects instantaneous flow, volume and flow time data in the line. It continuously measures and transmits this data in real time to the microcontroller (2). It transmits. In alternative applications, mechanical or 5 methods are used to obtain stream data. Instead of an electronic flow sensor, an ultrasonic flow meter can be used, or water Pressure and pressure changes in the line are measured via a pressure sensor, thus monitoring the flow. Status and flow-related values ​​can be derived. The ultrasonic flowmeter is applied to the pipeline. Measurements can be taken without intervention or by connecting to the pipe. It can provide. Data received from the pressure sensor is then transmitted to the microcontroller (2) 10 the initiation, termination, continuity or unusual occurrence of the flow by processing The change can be determined. The microcontroller mentioned (2) takes full control of the system by sampling and processing the incoming data. and executes decision-making mechanisms, followed by a learning process determined after installation. It runs the behavior modeling module (3) during the period. 15 The mentioned behavioral modeling module (3) creates a user-specific water consumption profile. It extracts data based on attributes such as time, flow rate, flow duration, time of day, and day type. The profile is stored in local memory. In the behavior modeling module (3), water consumption LSTM or Transformer is used to learn the time-dependent patterns of data. A time series model based on a baseline can be used. Alternatively, normal water consumption is 20 learning data that represents behavior and re-identifying deviations from that behavior an autoencoder-based anomaly for identification via generation error A detection model can be used. In another application, a user-specific normal model can be used. Extraction of consumption distributions and new flow events within these distributions Statistical density modeling for calculating probability 25 It is applicable. After the learning period is complete, the microcontroller will perform a new flow event. Instantaneous data exemplified by (2) the mentioned behavior modeling module The profile generated by (3) is transmitted to the dynamic threshold and decision module (4). The mentioned dynamic threshold and decision module (4) determines the time, flow rate and duration for each flow event. It generates a probabilistic anomaly score based on the window and applies this score to the learned data. It compares the data using threshold values ​​that are dynamically determined according to the profile. 11 Anomaly score depending on the learning model; time series prediction error, autoencoder reconstruction error, statistical probability or density value, or These can be produced by considering them together. If the anomaly score exceeds the dynamic threshold for a certain period of time, The mentioned dynamic threshold and decision module (4) condition is anomalous continuous flow 5 It classifies and immediately sends a closing command to the motorized valve (5). The aforementioned motorized valve (5) physically controls the flow of water in accordance with the command it receives. It cuts the water supply, thus preventing potential material damage and water waste. Motorized The valve (5) can be implemented as a motorized ball valve, or alternatively as a solenoid valve or a valve already present in the water line that is retrofitted to a 10 It can also be implemented in the form of a retrofit valve driver. Simultaneously with the shutdown process, the aforementioned dynamic threshold and decision module (4) transmits the anomaly detection information to the IoT communication module (6). The mentioned IoT A communication module (6) provides the user with an instant notification via the mobile application (8). sending and allowing the user to remotely monitor, turn the water back on or manually override 15 It enables the system to transmit commands. IoT communication module (6), can perform Wi-Fi based communication as well as Zigbee, Matter, BLE, LoRaWAN, GSM, or NB-IoT communication technologies, at least one of them. It can use Zigbee, Matter, or BLE. If Zigbee, Matter, or BLE is used, the system uses a local network. 20 via mobile application (8) through gateway or compatible control centre communication is possible; and if LoRaWAN, GSM or NB-IoT are used, the scope is wider. Data and commands can be transmitted remotely over the area network. Throughout this process, abnormalities were detected via the aforementioned IoT communication module (6). Flow events, user interventions, and system decisions event logging and notification module. (7) is recorded and logged with a timestamp. Not at home or 25 Remote working mode is selected by the user via the mobile application (8) and the command regarding the selection is sent to the microcontroller (2) via the IoT communication module (6) is being transmitted. By activating this mode, the dynamic threshold and decision module... (4) threshold sensitivity is increased; thus, normal in the absence of the user. 30 abnormal flow rate, volume or flow duration depending on the operating mode The level of protection is ensured by automatically shutting off the water even during flows. is being increased.

Claims

12 REQUESTS 1. IoT-powered automated water system with behavioral model-based adaptive anomaly detection. It is a cutting system; its feature is:  Connected in series to the main water supply line of a residence or workplace, and instantaneous fluctuations in the line a flow sensor (1) that measures flow rate, volume and flow time data, 5  data from the mentioned flow sensor (1) in real time a microcontroller that samples, processes and executes control decisions (2),  running on the mentioned microcontroller (2) and determined after installation A user-specific water consumption profile during a learning period, including time, flow rate, and stream. A behavior that extracts data based on duration, time of day, and day type attributes. 10 modeling module (3),  each according to the profile extracted by the mentioned behavior modeling module (3) a probabilistic analysis of the new flow event based on time, flow rate, and duration window A dynamic threshold and decision system that generates an anomaly score and sets dynamic thresholds. module (4), 15  Anomalous flow by the mentioned dynamic threshold and decision module (4) a motor that physically cuts off or reopens the water flow when verified. valve (5),  running on a user device, in case the user detects an anomaly Receive notifications and remotely monitor, reopen, and manually override 20 mobile application that enables it to enter commands (8),  Anomalous flow detection by the mentioned dynamic threshold and decision module (4) when this happens, sends a notification to the user via a mobile application (8), remote monitoring, reopening and manual override commands an IoT communication module (6) that enables the reception, 25  abnormal flow events through the mentioned IoT communication module (6) It records all alerts and user interventions with a timestamp. an event log and notification module (7) It includes.

2. The system is compliant with Request-1 and its feature is that the flow sensor (1) detects the flow in the water line. an ultrasonic flowmeter or water meter that determines data via ultrasonic measurement deriving the flow state from the pressure and pressure changes in the line It has a pressure sensor that provides this. 13 3. The system complies with Request-1 and its features include: IoT communication module (6), Wi-Fi, Zigbee, Matter, BLE, LoRaWAN, GSM or NB-IoT communication It is an IoT communication module that uses one of these technologies.

4. The system is compliant with Request-1 and its feature is that the motorized valve (5) is a motorized ball valve. opening and closing of a valve, solenoid valve, or a valve present in the water line. a device subsequently mounted on the valve to perform the closing action This is in the form of a retrofit valve driver.

5. IoT-powered automated water system with behavioral model-based adaptive anomaly detection. It is a cutting method, its characteristic is;  Instantaneous flow rate, volume and flow duration data in the main water supply line Continuous measurement by the sensor (1) and transmission to the microcontroller (2),  The transmitted data is processed in real time by the microcontroller (2) 15 sampling and processing,  Behavioral modeling during the learning period determined after installation Running module (3),  user-specific water consumption profile by the behavior modeling module (3) Extraction based on time, flow rate, flow duration, time of day, and day type attributes 20 and its storage in local memory,  Instantaneous snapshots of each new flow event after the completion of the learning period The data, along with the extracted user-specific water consumption profile, includes dynamic thresholds and to be sent to the decision module (4),  A probabilistic 25 based on time, flow rate and duration window for each flow event anomaly score is generated by dynamic threshold and decision module (4) and the generated probabilistic anomaly score is based on the user's specific water consumption profile Comparison with a dynamically determined threshold value,  Dynamically adjusting the threshold value of the probabilistic anomaly score over a specific period of time If it exceeds this limit, the flow event is classified as abnormal sustained flow, 30  Dynamic threshold for classifying the flow phenomenon as anomalous continuous flow and the closing command to the motorized valve (5) by the decision module (4) sending and physically directing the water flow through the motorized valve (5) cutting, 14  Transmission of abnormal flow detection information to the IoT communication module (6) and mobile Sending instant notifications to the user via the application (8),  Remote monitoring accessed via mobile application (8), turning water back on or manual override commands via IoT communication module (6) 5.  incidents of abnormal flow events, user interventions, and system decisions recording and notification by the module (7) with a timestamp,  Dynamic threshold and decision module (4) depending on the operating status of the system Changing threshold sensitivity It includes the steps of the process. 10 6. This method complies with Request-5 and is characterized by being away from home or working remotely. The mode is selected by the user via the mobile application (8), selected The command regarding the operating mode is sent via the IoT communication module (6). 15 Transmission to the microcontroller (2) activates the said operating mode Increasing the threshold sensitivity of the dynamic threshold and decision module (4) and lower flow rate, volume or flow duration compared to normal operating mode Automatic closing of the motorized valve (5) in case of abnormal flow events It includes the steps of the process.

7. It is a method that complies with Request-5, and its feature is that the behavior modeling module (3), 20 creating a user-specific water consumption profile and from that profile LSTM or Transformer-based timekeeping to determine deviations serial model, autoencoder-based anomaly detection model or statistical The process step must use at least one of the density modeling methods. It includes. 25 8. The method is compliant with claim 7 and its feature is the dynamic threshold and decision module (4), Probabilistic anomaly score varies over time depending on the behavioral model used. serial prediction error, autoencoder reconstruction error, or statistical error. It is generated based on at least one of the probability or density values.

Images