Electrical Fire Prevention System and Method for Power Facilities based on Artificial Intelligence

Abstract

According to one aspect of the present invention, an artificial intelligence-based electrical fire prevention system for power facilities is provided, comprising: a monitoring unit that captures a thermal image and a visible light image of a power facility composed of one or more components; an edge computing unit that processes the thermal image and the visible light image to identify the one or more components and predicts the fire risk for the one or more components; and a power facility management computing unit that communicates with the edge computing unit via a network and transmits alarm information to a manager terminal according to the prediction result.

Description

Technology Field

[0001] The present invention relates to an artificial intelligence-based system and method for preventing electrical fires in power facilities. Background Technology

[0002] For general monitoring of power facilities, a monitoring / management worker visits each facility in person carrying a portable thermal imaging camera, opens the casing while the power is on, and inspects the interior.

[0004] Consequently, since real-time monitoring of power facilities is difficult, there are aspects where it is difficult to prevent fires caused by the deterioration of power facilities.

[0006] For this reason, there is an increasing demand for systems capable of real-time monitoring of power facilities and predicting fire occurrences. Prior art literature

[0007] Registered Patent 10-2503525 (Publication Date: February 24, 2023) The problem to be solved

[0008] The artificial intelligence-based electrical fire prevention system and method for power facilities according to an embodiment of the present invention is intended to provide real-time monitoring of power facilities and prediction of fire occurrence.

[0009] The problems of this application are not limited to those mentioned above, and other problems not mentioned will be clearly understood by a person skilled in the art from the description below. means of solving the problem

[0010] According to one aspect of the present invention, an artificial intelligence-based electrical fire prevention system for power facilities is provided, comprising: a monitoring unit that captures a thermal image and a visible light image of a power facility composed of one or more components; an edge computing unit that processes the thermal image and the visible light image to identify the one or more components and predicts the fire risk for the one or more components; and a power facility management computing unit that communicates with the edge computing unit via a network and transmits alarm information to a manager terminal according to the prediction result.

[0011] According to another aspect of the present invention, a method for preventing electrical fires in power facilities based on artificial intelligence is provided, comprising: a step in which a monitoring unit captures a thermal image and a visible light image of a power facility composed of one or more components; a step in which an edge computing unit processes the thermal image and the visible light image to identify the one or more components and predicts the fire risk for the one or more components; and a step in which a power facility management computing unit communicates with the edge computing unit via a network to transmit alarm information to a manager terminal according to the prediction result.

[0012] The edge computing unit can process the thermal image and the visible light image to perform object recognition and temperature derivation for one or more components.

[0013] The edge computing unit can predict the fire risk of the component by processing the temperature of the component for which object recognition has been performed through a component degradation model.

[0014] The above component degradation model can be constructed by machine learning a dataset on the degree of degradation according to temperature for each component.

[0015] The above monitoring unit further includes a sound sensing unit that senses sound generated from one or more of the above components, and the component degradation model can be constructed by machine learning a dataset on the degree of degradation according to temperature and sound for each component. Effects of the invention

[0016] An artificial intelligence-based electrical fire prevention system and method for power facilities according to an embodiment of the present invention can provide real-time monitoring of power facilities and prediction of fire occurrence based on a component degradation model constructed according to machine learning by mapping thermal images and visible light images.

[0017] The effects of the present application are not limited to those mentioned above, and other unmentioned effects will be clearly understood by a person skilled in the art from the description below. Brief explanation of the drawing

[0018] Figure 1 shows an artificial intelligence-based electrical fire prevention system for power facilities according to an embodiment of the present invention. Figure 2 shows an example of the structure of an edge computing unit. Figure 3 shows an example of the structure of the monitoring unit. Specific details for implementing the invention

[0019] Embodiments of the present invention will be described in detail below with reference to the attached drawings. However, it will be readily apparent to those skilled in the art that the attached drawings are provided merely to facilitate the disclosure of the content of the present invention, and that the scope of the present invention is not limited to the scope of the attached drawings.

[0020] Furthermore, the terms used in this application are used merely to describe specific embodiments and are not intended to limit the invention. The singular expression includes the plural expression unless the context clearly indicates otherwise.

[0021] In this application, terms such as "comprising" or "having" are intended to specify the existence of the features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0023] FIG. 1 illustrates an artificial intelligence-based electrical fire prevention system for power facilities according to an embodiment of the present invention. As shown in FIG. 1, the artificial intelligence-based electrical fire prevention system for power facilities according to an embodiment of the present invention includes an edge computing unit (100), a monitoring unit (300), and a power facility management computing unit (500).

[0025] The monitoring unit (300) captures thermal images and visible light images of power equipment composed of one or more parts.

[0026] Power equipment may include at least one of a transformer, a switchboard, a fast charger, and a slow charger, but the present invention is not limited to such power equipment.

[0027] The components may be coils, capacitors, circuit breakers, switches, controllers, cooling fans, etc., but the present invention is not limited thereto.

[0028] As shown in FIG. 3, the monitoring unit (300) may include a thermal imaging camera (310) that captures a thermal image and a visible light camera (330) that senses visible light. The control unit (430) can control the thermal imaging camera (310) and the visible light camera (330), as well as the sound sensing unit (350) and the temperature and humidity sensing unit (370) which will be described later.

[0029] The monitoring unit (300) can communicate with the edge computing unit (100) via wired or wireless communication through the communication unit (390) or the input / output interface (410).

[0030] The monitored component of the power equipment may vary depending on the installation location of the monitoring unit (300). For example, if the monitoring unit (300) monitors the exterior of the power equipment, the monitored component may be located outside the power equipment. Additionally, if the monitoring unit (300) monitors the interior of the power equipment, the monitored component may be located inside the power equipment.

[0032] The edge computing unit processes thermal images and visible light images to identify one or more components and predicts the fire risk for one or more components.

[0033] FIG. 2 illustrates an example of the structure of an edge computing unit (100). As shown in FIG. 2, the edge computing unit (100) may include a processor (120), memory (130), a bus (110), an input / output interface (150), a display (160), and a communication interface (170). In some embodiments, the edge computing unit (100) may additionally include fewer or different components than the above components.

[0034] The bus (110) may include, for example, a circuit that connects the components to each other and transmits communication (e.g., control messages and / or data) between the components.

[0035] The processor (120) includes at least one of an MCU (Micro Controller Unit), a CPU (Central Processing Unit), and an AP (Application Processor), and may further include one or more of a CP (Communication Processor), a GPU (Graphic Processing Unit), an NPU (Neural Processing Unit), and a DPU (Data Processing Unit).

[0036] The processor (120) can perform operations or data processing regarding the control and / or communication of at least one other component of the edge computing unit (100), for example.

[0037] The memory (130) may include non-volatile memory. For example, the non-volatile memory may include at least one of ROM (Read Only Memory), HDD (Hard Disk Drive), ODD (Optical Disk Drive), SSD (Solid State Drive), and flash memory, but is not limited thereto.

[0038] The memory (130) may store at least one of commands, logic, files, and data related to at least one other component of the edge computing unit (100), for example. According to one embodiment, the memory (130) may store software and / or programs (140).

[0039] In addition to memory (130), the edge computing unit (100) may include volatile memory such as RAM.

[0040] In the drawing, the memory (130) is shown communicating with the processor (120) via the bus (110), but is not limited thereto. For example, the processor (120) may connect to the memory (130) located externally or remotely via a network (162), a communication interface (170), or an input / output interface (150) to receive and process at least one of commands, logic, and data. Additionally, the processor (120) may receive and execute software and / or programs (140) from the memory (130) located externally or remotely.

[0041] The program (140) may include, for example, a kernel (141), middleware (143), an application programming interface (API) (145), and / or an application program (or “application”) (147), etc. At least part of the kernel (141), middleware (143), or API (145) may be called an operating system (OS).

[0042] The kernel (141) can control or manage system resources (e.g., bus (110), processor (120), or memory (130), etc.) used to execute operations or functions implemented in other programs (e.g., middleware (143), API (145), or application program (147)). Additionally, the kernel (141) can provide an interface that controls or manages system resources by accessing individual components of the edge computing unit (100) from the middleware (143), API (145), or application program (147).

[0043] Middleware (143) can act as an intermediary to enable, for example, an API (145) or an application program (147) to communicate with the kernel (141) to exchange data. Additionally, middleware (143) can perform control (e.g., scheduling or load balancing) on ​​work requests received from the application program (147) by using methods such as assigning a priority to at least one application among the application programs (147) to use the system resources of the edge computing unit (100) (e.g., bus (110), processor (120), or memory (130), etc.).

[0044] The API (145) is, for example, an interface for an application (147) to control functions provided by the kernel (141) or middleware (143), and may include at least one interface or function (e.g., command) for, for example, file control, window control, image processing, or character control.

[0045] The input / output interface (150) can serve as an interface capable of transmitting commands or data input from other external devices to other component(s) of the edge computing unit (100). Additionally, the input / output interface (150) can output commands or data received from other component(s) of the edge computing unit (100) to a user or another external device. For example, the input / output interface (150) may be a serial port, a parallel port, PS / 2, ADB (Apple Desktop Bus), SCSI (Small Computer System Interface), USB, HDMI, DVI-I, or Thunderbolt, but is not limited thereto.

[0046] The display (160) may include at least one of, for example, a liquid crystal display (LCD), a light-emitting diode (LED) display, an organic light-emitting diode (OLED) display, a microelectromechanical systems (MEMS) display, an electronic paper display, or a beam projector. The display (160) may display various content (e.g., text, images, videos, icons, or symbols, etc.) to a user. The display (160) may include a touch screen and may receive touch, gesture, proximity, or hovering input using, for example, an electronic pen or a part of the user's body.

[0047] The communication interface (170) can establish communication between, for example, the edge computing unit (100) and an external device. For example, the communication interface (170) can communicate with an external device by being connected to a network (162) via wireless communication or wired communication.

[0048] Wireless communication or wired communication may be at least one of, for example, Bluetooth communication, NFC communication, Zigbee communication, Wi-Fi communication, cellular communication and wired LAN, but is not limited thereto. Cellular communication may use at least one of, for example, LTE, LTE-A, CDMA, WCDMA, UMTS, WiBro, GSM or IMT-2020, but is not limited thereto.

[0049] The network (162) may include at least one of a LAN, WAN, internet, intranet, telephone network, mobile communication network and blockchain, but is not limited thereto.

[0050] The edge computing unit (100) can perform server-client communication with the power facility management computing unit (500).

[0052] As previously described, the edge computing unit can identify one or more of the above components by processing thermal images and visible light images.

[0053] To this end, the edge computing unit (100) can perform mapping between a thermal image and a visible light image to identify parts. The resolution of the thermal image (e.g., 160x120) may be smaller than the resolution of the visible light image (e.g., 640x360 to 3840x2160).

[0054] Accordingly, the edge computing unit (100) can perform data labeling that increases the resolution of the thermal image through pixel interpolation of the thermal image. The edge computing unit (100) can map the thermal image with increased resolution to a visible light image.

[0055] The edge computing unit (100) can detect the Region of Interest (ROI) from each image to calculate relative position coordinates and perform mapping through position correction for mapping between the thermal image with increased resolution and the visible light image.

[0056] The edge computing unit (100) can identify information about a part by comparing the coordinate data of the mapped image with the coordinate data where the actual object is located. The information about the part may be identification information of the part (e.g., name, unique code, etc.), installation date, etc., but the present invention is not limited thereto.

[0058] Next, the edge computing unit (100) explains how to predict the fire risk for one or more components.

[0059] The edge computing unit (100) can process thermal images and visible light images to perform object recognition and temperature derivation for one or more parts. Object recognition can be performed through a part recognition model built through machine learning on feature points of the part. The edge computing unit (100) can derive the temperature of the object-recognized part through the mapped thermal images.

[0060] To build a part recognition model, an edge computing unit (100) or a separate computing device can machine learn a dataset of feature points of a part. The built part recognition model can be installed in the edge computing unit (100).

[0061] An edge computing unit (100) or a separate computing device may perform preprocessing of the data set for the feature points. Data preprocessing may include processing of missing values, removal of outliers, and data scaling, but the present invention is not limited thereto.

[0063] The edge computing unit (100) can predict the fire risk of a component by processing the temperature of the component for which object recognition has been performed through a component degradation model. The component degradation model can be constructed by machine learning a dataset regarding the degree of degradation according to the temperature of each component. The machine learning of the dataset regarding the degree of degradation can be performed by the edge computing unit (100) or by a separate computing device.

[0064] The component degradation model can detect signs of electrical anomalies through the analysis of temperature distributions in mapped thermal images. To this end, the component degradation model detects Regions of Interest (ROIs) in the thermal images and extracts feature data through difference image and Gaussian analysis.

[0065] The edge computing unit (100) may be run on a Linux-based dedicated OS and may include a CPU, GPU, NPU and VPU (Visual Processing Unit) for running AI deep learning and artificial intelligence algorithms, but the present invention is not limited thereto.

[0066] Machine learning for object recognition or anomaly detection regarding parts may utilize decision trees, Bayesian networks, support vector machines (SVM), artificial neural networks (ANN), or convolutional neural networks (CNN), but the present invention is not limited thereto.

[0067] Machine learning can be performed through machine learning programs such as Keras, Apache MXNnet, Deeplearning4j, Tensorflow, Microsoft Cognitive Toolkit, or Theano, but the present invention is not limited thereto.

[0068] Prior to machine learning on the data set regarding the degree of degradation, preprocessing of the said data set may be performed. Data preprocessing may include processing of missing values, removal of outliers, and data scaling, but the present invention is not limited thereto.

[0070] The power facility management computing unit (500) communicates with the edge computing unit (100) through a network and transmits alarm information to the manager terminal (700) according to the prediction result.

[0072] For general power facility management, a monitoring / management worker directly carries a portable thermal imaging camera (310), visits each facility, opens the casing while the power is on, and checks the interior.

[0073] In contrast, the present invention can detect abnormal signs in advance by continuously deploying a thermal imaging camera (310) in the power facility to monitor the temperature of the power facility components in real time and analyzing the deterioration status of the power facility in real time through an edge computing unit (100) to which artificial intelligence-based object recognition and analysis technology is applied.

[0074] In the present invention, since object recognition of a component and prediction of deterioration based on the temperature of the component are performed in the edge computing unit (100), the burden on the power equipment management computing unit (500) is reduced, allowing for real-time monitoring of the power equipment and increasing the reliability of the monitoring.

[0076] The monitoring unit (300) may further include a sound sensing unit (350) that senses sound generated from one or more components. The component degradation model may be constructed by machine learning a dataset regarding the degree of degradation according to temperature and sound for each component. The sound sensing unit (350) may include a microphone, a filtering unit, and an ADC (Analog Digital Converter). Depending on the degree of degradation, abnormal sound may occur due to abnormal arcs, etc., in the components of the power equipment.

[0077] The edge computing unit (100) can increase the reliability of the prediction regarding the fire risk of power equipment by considering not only the temperature of the component but also the acoustics. The edge computing unit (100) can extract frequency characteristic data through Fast Fourier Transform (FFT) after performing frequency band sampling among the acoustic data. It can detect the formant of the acoustic sign of electrical abnormality through formant analysis of the extracted frequency characteristic data and generate metadata for the detected pattern.

[0078] The monitoring unit (300) may further include a temperature and humidity sensing unit (370). Accordingly, the component degradation model can be constructed by machine learning a dataset regarding the degree of degradation according to the temperature and acoustics of each component and the temperature and humidity of the atmosphere surrounding the component.

[0080] Next, an artificial intelligence-based method for preventing electrical fires in power facilities according to an embodiment of the present invention will be described. For the convenience of explanation, the method for preventing electrical fires in power facilities based on artificial intelligence according to an embodiment of the present invention described below has been explained through the previously described system for preventing electrical fires in power facilities based on artificial intelligence; however, the method for preventing electrical fires in power facilities based on artificial intelligence according to an embodiment of the present invention is not limited thereto.

[0082] An artificial intelligence-based method for preventing electrical fires in power facilities according to an embodiment of the present invention includes the steps of: a monitoring unit (300) capturing a thermal image and a visible light image of a power facility composed of one or more components; an edge computing unit (100) processing the thermal image and the visible light image to identify one or more components and predicting the fire risk for one or more components; and a power facility management computing unit (500) communicating with the edge computing unit via a network to transmit alarm information to a manager terminal according to the prediction result.

[0084] At this time, the edge computing unit (100) can process thermal images and visible light images to perform object recognition and temperature derivation for one or more parts.

[0086] The edge computing unit (100) can predict the fire risk of a component by processing the temperature of the component for which object recognition has been performed through a component degradation model.

[0088] A component degradation model can be constructed by machine learning on a dataset regarding the degree of degradation according to temperature for each component.

[0090] The monitoring unit (300) further includes a sound sensing unit that senses sound generated from one or more parts, and the part degradation model can be constructed by machine learning a dataset on the degree of degradation according to temperature and sound for each part.

[0092] As described above, embodiments according to the present invention have been examined. It is obvious to those skilled in the art that, in addition to the embodiments described above, the present invention may be embodied in other specific forms without departing from its spirit or scope. Therefore, the embodiments described above should be regarded as illustrative rather than restrictive, and accordingly, the present invention is not limited to the description above but may be modified within the scope of the appended claims and their equivalents. Explanation of the symbols

[0093] Edge computing unit (100) Monitoring unit (300) Thermal imaging camera (310) Visible light camera (330) Sound sensing unit (350) Temperature and humidity sensing unit (370) Communications Department (390) Input / Output Interface (410) Control unit (430) Power facility management computing unit (500) Administrator terminal (700)

Claims

Claim 1 A monitoring unit that captures thermal images and visible light images of power equipment composed of multiple components; an edge computing unit that processes the thermal images and the visible light images to identify the multiple components and predict the fire risk for the multiple components;It includes a power facility management computing unit that communicates with the edge computing unit via a network and transmits alarm information to a manager terminal according to the prediction result; the edge computing unit processes the thermal image and the visible light image to identify the plurality of components, performs mapping between the thermal image and the visible light image to identify the plurality of components, maps the thermal image with increased resolution through pixel interpolation to the visible light image, detects a Region of Interest (ROI) from the thermal image and the visible light image to map the thermal image with increased resolution and performs mapping through position correction via the calculation of relative position coordinates, compares the coordinate data of the mapped thermal image and the visible light image with the coordinate data where the plurality of components are located to determine the identification information and installation time of the plurality of components, processes the thermal image and the visible light image to perform object recognition and temperature derivation for the plurality of components, and object recognition is performed through a component recognition model constructed through machine learning on the feature points of the components, and the edge computing unit [retrieves] the object recognized An artificial intelligence-based electrical fire prevention system for power facilities, characterized by deriving the temperature of a component through the mapped thermal image and processing the temperature of the component for which object recognition has been performed through a component degradation model to predict the fire risk of the component, wherein the component degradation model is constructed by machine learning a dataset regarding the degree of degradation according to the temperature of each component, wherein the monitoring unit further includes a sound sensing unit that senses sound generated from the plurality of components, and wherein the component degradation model is constructed by machine learning a dataset regarding the degree of degradation according to the temperature and sound of each component. Claim 2 delete Claim 3 delete Claim 4 delete Claim 5 delete Claim 6 A step in which a monitoring unit captures a thermal image and a visible light image of a power facility composed of multiple components; a step in which an edge computing unit processes the thermal image and the visible light image to identify the multiple components and predict the fire risk for the multiple components; and a step in which a power facility management computing unit communicates with the edge computing unit via a network to transmit alarm information to an administrator terminal according to the prediction result.The edge computing unit includes, wherein the edge computing unit processes the thermal image and the visible light image to identify the plurality of components, performs mapping between the thermal image and the visible light image to identify the plurality of components, maps the thermal image with increased resolution through pixel interpolation to the visible light image, detects a Region of Interest (ROI) from the thermal image and the visible light image to perform mapping through position correction via the calculation of relative position coordinates, identifies identification information and installation time of the plurality of components by comparing the coordinate data of the mapped thermal image and the visible light image with the coordinate data where the plurality of components are located, processes the thermal image and the visible light image to perform object recognition and temperature derivation for the plurality of components, and object recognition is performed through a component recognition model constructed through machine learning on the feature points of the components, the edge computing unit derives the temperature of the object-recognized component through the mapped thermal image, and for the object-recognized component An artificial intelligence-based method for preventing electrical fires in power facilities, characterized by processing temperature through a component degradation model to predict the fire risk of said component, wherein said component degradation model is constructed by machine learning a dataset regarding the degree of degradation according to temperature for each said component, and said monitoring unit further includes a sound sensing unit that senses sound generated from said plurality of components, and said component degradation model is constructed by machine learning a dataset regarding the degree of degradation according to temperature and sound for each said component. Claim 7 delete Claim 8 delete Claim 9 delete Claim 10 delete

Images