Orientation-Independent Predictive Maintenance IoT System Using Recursive DC-Blocking Filter and Self-Adaptive Statistical Thresholding

Abstract

This invention relates to an orientation-independent predictive maintenance IoT system using a recursive DC-blocking filter and self-adaptive statistical thresholding, which can be used in industrial production facilities, predictive maintenance systems, condition monitoring systems, smart factory applications, Industry 4.0 applications, IoT-based monitoring systems, rotary machine monitoring systems, electric motor monitoring systems, pump monitoring systems, compressor monitoring systems, fan monitoring systems, generator monitoring systems, conveyor systems, CNC machines, production lines, automation systems, power generation plants, maintenance management systems, and remote machine monitoring systems.The machine consists of a microcontroller (1) that enables the processing of mechanical vibration data generated on the machine, an MPU6050 sensor (2) that detects mechanical vibrations on the machine, an axial accelerometer (3) located within the said MPU6050 sensor (2) that produces acceleration data in three axes, an I2C bus connection (4) that provides digital data transfer between the MPU6050 sensor (2) and the microcontroller (1), and a power supply (5) that provides energy for the operation of the microcontroller (1) and the MPU6050 sensor (2).

Description

1 TARIFF RECURRENT DC-BLOCKING FILTER AND SELF-ADAPTIVE ORIENTATION-INDEPENDENT USING STATISTICAL THRESHOLDING PREDICTIVE MAINTENANCE IOT SYSTEM 5 Technological Field: This invention is suitable for industrial production facilities, predictive maintenance systems, and condition monitoring. systems, smart factory applications, Industry 4.0 applications, IoT-based monitoring 10 systems, rotary machine monitoring systems, electric motor monitoring systems, pumps monitoring systems, compressor monitoring systems, fan monitoring systems, generator monitoring systems, conveyor systems, CNC machines, production lines, automation systems, in energy production facilities, maintenance management systems and remote machine monitoring systems Available recursive DC-blocking filter and self-adaptive statistical thresholding 15 This relates to orientation-independent predictive maintenance IoT systems. State of the Art: Monitoring the operating conditions of rotating equipment used in industrial facilities and 20 Various predictive methods are used to identify maintenance needs before a malfunction occurs. Maintenance systems are used. These systems are generally based on vibration analysis. and is based on mechanical vibration data obtained from the equipment. The aim is to identify signs of malfunction by means of evaluation. Known In these techniques, high-cost industrial vibration analyzers are used for this purpose, while low-cost methods are available. Costly MEMS-based accelerometer systems are used. Piezoelectric sensors are commonly used in high-cost industrial systems. The data obtained is used and stored on central servers or cloud infrastructures. These systems are being analyzed. They are capable of performing highly accurate measurements. 30 together, complex installation processes, high hardware costs, and advanced data 2 They need processing infrastructure. This is especially true for small and medium-sized enterprises (SMEs). This makes widespread adoption of these systems difficult for businesses. In low-cost MEMS-based solutions, vibrations obtained from accelerometers are used. Anomaly detection is performed using the data. However, in these systems, sensor 5 the gravity-induced constant component within the data effectively The inability to separate them creates significant technical problems. Measuring the effect of gravity. Because the data becomes dominant, the sensors are positioned on specific axes and with precision. It needs to be positioned at the mounting angles. From the ideal position of the sensor In case of deviation, measurement accuracy decreases and erroneous evaluations occur. 10 It is possible to come. Another problem encountered in known techniques is that common problems occur in anomaly detection procedures. These are static thresholding methods used. In these methods, the determined threshold... The values ​​are kept constant throughout the system's lifespan. However, the machine 15 wear and tear on components over time, load changes, operation due to variations in conditions and environmental influences, the initially determined threshold Their values ​​may become outdated. As a result, systems are highly... This can generate false alarms or prevent real malfunctions from being detected in a timely manner. This can lead to problems. 20 In addition, a significant portion of existing systems perform analysis processes. This requires storing or centralizing large amounts of raw data. This requires transferring data to the systems. This approach involves high memory usage, increasing network traffic, additional data storage requirements and central processing costs 25 This creates various technical and economic disadvantages, such as rising prices. In particular... In embedded systems and IoT devices with limited resources, these system requirements... This limits its performance and applicability. Description of the invention: 30 3 The invention relates to gravity-induced constants during the processing of mechanical vibration data. It allows for the software-based separation of components. This enables the sensor to be... The need for positioning on specific axes or with precise mounting angles is eliminated. It is being removed, and the system can reliably operate in different orientations. Thus Installation processes are simplified and measurement errors caused by assembly are reduced. 5 is provided. Thanks to the recursive DC-blocking filter used in the invention, vibration is reduced. The effect of gravity within the signal can be filtered in real time. Thus, data obtained from mechanical vibrations can be more accurately interpreted. 10 It is possible to evaluate the sensor's performance and assembly. It is being made independent of its orientation. Another advantage of the invention is that statistical analysis processes are performed using the Welford algorithm. This is achieved using this approach. Thanks to this approach, the mean and variance are 15. calculations without needing to store the entire dataset This can be achieved even with limited processor and memory resources. It becomes possible to apply advanced analysis processes on microcontrollers. It is coming. The self-adaptive statistical thresholding mechanism used in the invention controls the operation of the system. It can dynamically adapt to changes occurring in its conditions. Z- The current operating behavior of the machine is assessed through a score-based evaluation approach. It is constantly being analyzed, and the decision-making mechanism is based on this data. It is being updated. This helps to reduce false alarm rates and anomaly detection. 25 It contributes to increasing its accuracy. The invention enables complex data processing operations to be performed on low-cost microcontrollers such as the ESP32. By being able to implement this, it contributes to reducing system costs. This eliminates the need for expensive industrial analyzers. It is possible to expand the use of predictive maintenance applications. is happening. 4 Instead of transferring all raw vibration data to central systems, the invention analyzes it. Communicating the results is preferred. This approach reduces network bandwidth usage. reducing data transmission load and cloud-based storage needs This minimizes the system's energy efficiency and communication performance. is being increased. 5 The invention has implications for manufacturing industries, energy production systems, HVAC applications, logistics, and in different application areas such as storage systems and smart city infrastructures It offers a flexible architecture that can be used. In addition, computers and software 10 disciplines of engineering, electrical-electronics engineering and mechanical engineering It forms a versatile predictive maintenance solution located at the intersection. Explaining the Figures: The invention will be described by referring to the attached figures, so that the features of the invention are outlined in 15 It will be understood and appreciated more clearly, but the purpose of this invention is this obvious It is not about limiting it with regulations. On the contrary, the invention is defined by the accompanying claims. all alternatives, modifications, and options that could be included within the defined area The aim is to cover their equivalences. The details shown are only for the present invention. It was shown to illustrate the preferred arrangements and both methods 20 shaping, as well as the rules and conceptual features of the invention, in the most useful way. It should be understood that they are presented to provide a readily understandable definition. This in the drawings; Figure 1 shows a schematic view of the system. 25 Illustrations that will help understand this invention are shown in the attached image. They are numbered and their names are given below. References Explanation: 30 1. Microcontroller 2. MPU6050 Sensor 3-axis accelerometer 4. I2C Bus Connection 5. Power Supply 6. Data Collection 5 7. Recursive DC-Blocking Filter 8. Statistical Analysis 9. Anomaly Decision 10. Warning Transmission 11. Processor 10 12. Server Description of the Invention: The invention enables the processing of mechanical vibration data generated on a machine. microcontroller (1), MPU6050 which detects mechanical vibrations on the machine to the sensor (2), the said MPU6050 sensor (2) contains accelerometers in three axes. the axial accelerometer (3) that produces data, the MPU6050 sensor (2) and the microcontroller (1) an I2C bus connection that provides digital data transfer between them (4), A power supply 20 provides power for the operation of the microcontroller (1) and the MPU6050 sensor (2). It has the source (5). The invention has an axial accelerometer (3) that produces acceleration data in three axes. The invention has an anomaly It has a server (12) that receives its information. The invention relates to the raw vibration data from the MPU6050 sensor (2) through the data acquisition step (6). reading, raw data read using recursive DC-blocking filter (7) filtering, statistical analysis step (8) on the filtered data and mean and Updating the variance values ​​using the obtained statistical values. Performing deviation analysis with anomaly decision step (9) and anomaly detection 30 If this happens, the process of sending the alarm information is done with the warning transmission step (10). It includes the steps. 6 The invention is a recursive DC-blocking system that isolates the gravity component from vibration data. The invention includes the filter step (7). The invention updates the mean and variance values. It includes the statistical analysis step (8). The invention is a statistical method that calculates mean and variance values ​​using the Welford algorithm. It includes the analysis step (8). The invention calculates the statistical deviation using the Z-Score. This includes the anomaly decision step (9) which evaluates. The invention describes the alert transmission step (10) which sends the alarm information over a wireless network. The invention includes a data acquisition system that reads vibration data at a rate of 100 samples per second. It includes step (6). Detailed Description of the Invention: The invention enables the monitoring and analysis of vibration data occurring in machinery and equipment. a predictor for identifying abnormal operating conditions by means of Maintenance is related to the IoT system. Within the scope of the invention, the main control unit of the system is a microcontroller (1) 20 It consists of a microcontroller (1) managing the system components, sensors the collection, processing, and transmission of analysis results of data It performs the following: Detecting mechanical vibrations occurring on the machine. An MPU6050 sensor (2) is used for this purpose. The said MPU6050 sensor (2) a one-axis accelerometer (3) 25 that performs acceleration measurement in three axes The axial accelerometer (3) detects the vibrations that occur on the machine. It converts these into digital acceleration data. Data communication between the MPU6050 sensor (2) and the microcontroller (1) is via an I2C bus. The connection is made via (4). The I2C bus connection (4) is made via sensor 30 It enables the transfer of acceleration data generated by the microcontroller to the microcontroller (1). 7 The electrical energy required for the operation of the system components is a power supply (5) is provided by. Power supply (5), microcontroller (1) with MPU6050 This ensures that the sensor (2) operates without interruption. When the system is started, the microcontroller (1) initializes the MPU6050 sensor (2) and 5 It starts collecting data from the sensor. At this stage, the data collection step is (6) is performed and raw acceleration data produced by the MPU6050 sensor (2) is I2C The data is transferred to the microcontroller (1) via the bus connection (4). The raw vibration data reaching the microcontroller (1) is then recursively processed by DC-10 It is processed by the blocking filter (7). Recursive DC-blocking filter (7), By removing the stationary components from the sensor data, vibration-induced It enables the acquisition of variable components. The filtered data is transferred to the statistical analysis step (8). 15 In the statistical analysis step (8), mean and variance values ​​of vibration data It is constantly being updated. These calculations are performed while the data flow continues. The process is ongoing, and the statistical parameters obtained are kept up-to-date. The data obtained in the statistical analysis step (8) are then used in the anomaly decision step (9) 20 is transferred. In the anomaly decision step (9), the current vibration data is statistically transferred. its position within the distribution is evaluated and its normal working behavior is assessed. Cases exhibiting deviations are identified. If an abnormal operating condition is detected, the warning transmission step (10) is activated 25 It enters. Warning transmission step (10) transmits the generated alarm information to the microcontroller (1) It transfers it to the network environment via. The processor (11) located within the microcontroller (1), data acquisition step (6), recursive DC-blocking filter (7), statistical analysis step (8) and anomaly decision 30 It ensures the execution of the operations carried out within the scope of step (9). 8 Alarm information generated as a result of warning transmission step (10) is sent to a server (12) is sent. The server (12) receives the alarm logs transmitted by the system. and ensures its concealment. In this way, the system continuously monitors the vibration data generated on the machine, 5 analyzing the data obtained and identifying abnormal operating conditions In this case, it transmits the relevant alarm information to the server (12). 15 25

Claims

9 REQUESTS 1- The invention relates to a recursive DC-blocking filter and self-adaptive statistical thresholding. This relates to an orientation-independent predictive maintenance IoT system, and its feature is:  a microcontroller that enables the processing of mechanical vibration data (1), 5  an MPU6050 sensor (2) that detects mechanical vibrations on the machine,  The MPU6050 sensor (2) contains acceleration data in three axes. accelerometer generating axis (3),  Digital data transfer between MPU6050 sensor (2) and microcontroller (1) an I2C bus connection (4) and 10  a microcontroller (1) and MPU6050 sensor (2) that provide power for operation It consists of a power supply (5). 2- Recursive DC-blocking filter and self-adaptive statistical mentioned in Claim 1. It is an orientation-independent predictive maintenance IoT system that uses thresholding, and its feature is; three 15 It is characterized by having an axial accelerometer (3) that produces acceleration data on the axis. It is done. 3- The recursive dc-blocking filter and self-adaptive statistical mentioned in Claim 1. It is an orientation-independent predictive maintenance IoT system that uses thresholding, and its feature is; 20 It is characterized by having a server (12) that receives anomaly information. 4- The invention relates to a recursive dc-blocking filter and self-adaptive statistical thresholding. It is a method for orientation-independent predictive maintenance IoT systems that uses, Feature; 25  Raw vibration data from MPU6050 sensor (2) with data acquisition step (6) reading,  using a recursive DC-blocking filter (7) on the raw data read filtering,  Statistical analysis step (8) on filtered data with mean and variance 30 updating the values,  Using the obtained statistical values, the deviation with the anomaly decision step (9) conducting the analysis and  If an anomaly is detected, the alarm information is transmitted via the warning transmission step (10). It includes the steps involved in sending the message. 5- The method mentioned in Claim 4, its characteristic is that it extracts the gravity component from vibration data. It is characterized by having a recursive DC-blocking filter step (7) that removes it. It is done. 6- The method mentioned in claim 4, its characteristic is that it takes the mean and variance values ​​as 10. It is characterized by having an updating statistical analysis step (8). 7- The method mentioned in claim 6 is characterized by its use of Welford's method for mean and variance values. It is characterized by having a statistical analysis step (8) which calculates with its algorithm. It is done. 15 8- The method mentioned in Claim 4, its characteristic is that it calculates statistical deviation using the Z-Score. It is characterized by having an anomaly decision step (9) that evaluates. 9- The method mentioned in claim 4, its feature is; transmitting alarm information via wireless network 20 It is characterized by having a sender alert transmission step (10). 10- The method mentioned in Claim 4, characterized by its ability to collect vibration data at 100 samples per second. It is characterized by having a data collection step (6) that reads at speed. 30

Images