Artificial intelligence-based power distribution panel with abnormal sign detection, fire, and failure prediction capabilities.

Abstract

An artificial intelligence-based power distribution panel device according to one embodiment of the present invention may include: a sensor unit installed inside the power distribution panel device that senses the internal environmental state and power state in a time series and generates sensing data; a data collection unit that collects sensing data from the sensor unit and performs preprocessing on the sensing data; an analysis processing unit that predicts a first abnormality sign and fire or failure risk of the power distribution panel device through a first predictive AI model based on the preprocessed sensing data; a control unit that controls the power distribution panel device based on the predicted abnormality sign and fire or failure risk; a communication unit that transmits the sensing data or the predicted abnormality sign and fire or failure risk to an external server or mobile device; and an display unit that displays or notifies the abnormality sign and fire or failure risk.

Description

Technical Field

[0001] The present invention relates to an AI-based power distribution board device having abnormal symptom, fire, and failure prediction functions, and an AI-based monitoring and prediction system including the same.

Background Art

[0002] A power distribution board is the center of a power system. It houses main circuit breaker devices and monitoring and control devices in a closed outer box that can protect the human body, and is configured to enable power system monitoring, protection, and operation according to the requirements of users. The types and sizes of products vary depending on the ratings of the housed devices, safety protection levels, and applicable specifications. Usually, it includes air-insulated switchgear, gas-insulated switchgear, and solid-insulated switchgear. Recently, while various functions such as reliability, convenience of use, and safety are required for power distribution boards, the demand for power distribution boards that can be miniaturized for space saving is increasing. Therefore, while collective switchboards that utilize space and gas switchboards that utilize SF6 gas have emerged, digitalization has been rapidly progressing to realize convenient functions of power distribution boards.

[0003] On the other hand, because power distribution panels primarily contain equipment for supplying or interrupting power, it is difficult to detect abnormal signs, fires, or malfunctions. Furthermore, generally, power travels along high-voltage lines complexly connected according to power system diagrams, and the power supplied to industrial equipment passes through the power distribution panel, where the AC voltage is reduced by transformers inside the panel and then supplied as various power demand elements via busbars. In this process, if heat is generated from within the power distribution panel, problems such as increased electrical contact resistance due to overcurrent and loosening of fixing bolts at the busbar connection points occur. During prolonged use of power equipment, heat is also generated due to the progression of insulation deterioration. If such abnormal signs are left unchecked, they can lead to leakage current accidents, ground faults, or short circuits, resulting in fires. Therefore, existing power distribution panels have emerged that incorporate temperature sensing devices such as temperature tapes and PT100 temperature sensors to detect fires and malfunctions, provide alarms, or implement control functions accordingly.

[0004] In this regard, Korean Patent Publication No. 10-0984679 describes a system in which a thermal imaging camera is installed inside a power distribution panel to constantly monitor the panel and power equipment, capturing images of the temperature, degree of deterioration, overload conditions, and contact problems of energized parts. The image data is transmitted to a server for diagnosis, and if a problem is determined with the product, the system is configured to either control the equipment or generate an abnormal alarm signal in the power distribution panel. Furthermore, Korean Patent Publication No. 10-2303585 describes a system that utilizes images captured by cameras installed inside or outside the power distribution panel, along with public data via an OPEN API, including fire occurrences by region and time, temperature and humidity, and earthquake occurrences, to detect and predict signs of abnormalities in the power distribution panel and notify the inspector located closest to the power distribution panel requiring inspection.

[0005] However, such sensing and alarm systems for power distribution panels typically analyze data from temperature sensors and thermal imaging cameras, and if deterioration of the panel is detected, an alarm is transmitted to the inspector. There is a lack of systems that provide more proactive response options, such as determining the abnormal condition of components within the panel, predicting the risk of fire or failure, and determining when power equipment should be replaced. Typically, when a power distribution panel malfunctions or breaks down, it is necessary to call in a specialist to inspect it, which takes a considerable amount of time. When signs of abnormality are detected in a power distribution panel, industrial equipment operators often take emergency measures, but because it is difficult to understand the internal condition of the panel without a specialist, there is a problem in that when signs of abnormality are detected in a power distribution panel, an A / S technician must be called in and wait for a more precise assessment. [Prior art documents] [Patent Documents]

[0006] [Patent Document 1] Korean Patent Publication No. 10-0984679 [Patent Document 2] Korean Patent Publication No. 10-2303585 [Overview of the project] [Problems that the invention aims to solve]

[0007] The objective of the present invention to solve the aforementioned problems is to provide an artificial intelligence-based power distribution panel device and a monitoring and prediction system for the power distribution panel device that can be miniaturized and improve reliability, ease of use, and safety by automatically making predictions in real time, such as determining signs of abnormality in the power distribution panel, and predicting the risk and / or location of fire or failure, based on environmental data and power data sensed over time to the power distribution panel.

[0008] Another object of the present invention is to provide an artificial intelligence-based power distribution panel device and a monitoring and prediction system for the power distribution panel device that facilitates proactive preventive measures against fire or failure in the power distribution panel by automatically making real-time predictions such as the determination of abnormal signs in the power distribution panel, the degree of risk of fire or failure, and / or location, and by allowing the timing of replacement of internal power equipment in the power distribution panel to be determined at the appropriate time, thereby extending the service life of internal equipment and maximizing its maximum performance.

[0009] The technical problems that this invention aims to solve are not limited to those described above, and any other technical problems not mentioned will be clearly understood by a person with ordinary skill in the art to which this invention belongs from the following description. [Means for solving the problem]

[0010] An artificial intelligence-based power distribution panel device according to one embodiment of the present invention for achieving the above objective may include: a sensor unit installed in the power distribution panel device that senses the internal environmental state and power state in a time series and generates sensing data; a data collection unit that collects the sensing data from the sensor unit and performs preprocessing on the sensing data; an analysis processing unit that predicts a first abnormality sign and fire or failure risk of the power distribution panel device through a first predictive AI model based on the preprocessed sensing data; a control unit that controls the power distribution panel device based on the predicted abnormality sign and fire or failure risk; a communication unit that transmits the sensing data or the predicted abnormality sign and fire or failure risk to an external server or mobile device; and an display unit that displays or notifies the abnormality sign and fire or failure risk.

[0011] According to one embodiment of the present invention, the sensing data for the internal environmental state includes at least one of the following: gas content data, flame sensing data, temperature data, humidity data, gradient data, vibration data, acceleration data, and open sensing data, and the sensing data for the power state may include at least one of the following: current data, voltage data, active / reactive power data, power factor data, operating time data, frequency / phase angle data, harmonic data, and ultrasonic data.

[0012] The aforementioned preprocessing may include at least one of the following: missing values ​​removal and noise removal.

[0013] The data acquisition unit may further include a virtual sensor module that, based on the accumulated sensing data from the data acquisition unit, analyzes the recently received sensing data to determine whether there are any data errors due to abnormalities in the power distribution panel device, and, if it is determined that there are data errors, generates virtual sensing data at the time the data error occurred through a virtual sensor AI model based on the pre-processed sensing data sensed up to the time the data error occurred, and transmits it to the analysis processing unit.

[0014] The analysis processing unit may further include a time-series analysis module that, based on the pre-processed sensing data, analyzes the time-series trend when time-series anomaly signal data or numerical data outside a pre-set range occurs, and transmits an anomaly indication or alarm to the display unit.

[0015] The first predictive AI model is at least one of the following: a pre-trained model, a demand / load adaptive model, and an iterative optimization model.

[0016] Another embodiment of the present invention may include an artificial intelligence-based monitoring and prediction system for a power distribution panel, and a power control and monitoring server that receives the preprocessed sensing data or the first abnormality indicator and fire or failure risk from the power distribution panel, predicts the second abnormality indicator and fire or failure risk of the power distribution panel through a second predictive AI model, and transmits it to the power distribution panel.

[0017] The system may further include a mobile device that receives and displays a second abnormality sign and fire or failure risk of the power distribution panel, receives control information based thereon from the user, and transmits the input control information to the power distribution panel.

[0018] The power control and monitoring server may include a big data server unit that receives and stores the pre-processed sensing data or the first abnormality sign and fire or failure risk received from the power distribution panel device; a database unit that stores machine-learning models that can be used as the first predictive AI model or the second predictive AI model; and an analysis server unit that predicts the second abnormality sign and fire or failure risk of the power distribution panel device through the second predictive AI model based on the pre-processed sensing data or the first abnormality sign and fire or failure risk received from the big data server unit.

[0019] The second predictive AI model comprises at least one of a single-variable predictive AI model and a multivariate predictive AI model based on two or more variables, or comprises multiple models by integrating these.

[0020] The analysis server unit then processes the pre-processed sensing data as follows: (i) Predicting the location of individual fires or faults through individual variable prediction AI models, (ii) Predicting the risk of fire or malfunction by integrating and applying the individual variable prediction AI model and a multivariate prediction AI model based on two or more variables, and (iii) Apply the multivariate prediction AI model to predict abnormal signs and the risk of fire or failure, and based on the predicted abnormal signs and the risk of fire or failure, input them into the predictive maintenance machine learning model to derive corresponding maintenance policies, and perform at least one of these actions.

[0021] The analysis server (a) Based on the preprocessed sensed data, learn or relearn at least one of the models stored in the database section. (b) Based on the preprocessed sensed data generated during the period when a preset situation occurs, relearn at least one of the models stored in the database section to generate a user / load adaptation model, and (c) Further perform at least one of the following actions: based on the preprocessed sensed data generated during a preset period in accordance with the operation or equipment input situation of the power distribution board device, repeatedly relearn at least one of the models stored in the data section to generate an iterative optimization model. [Advantages of the Invention]

[0022] According to an embodiment of the present invention, based on environmental data and power data sensed in a time series for a power distribution board, predictions such as the determination of abnormal signs of the power distribution board, the risk of fire or failure, and / or the location are automatically made in real time, thereby improving reliability, ease of use, and safety, and enabling miniaturization of the power distribution board device.

[0023] According to an embodiment of the present invention, predictions such as the determination of abnormal signs of the power distribution board, the risk of fire or failure, and / or the location are automatically made in real time, making it easy to take preventive measures against fire or failure of the power distribution board, and it is possible to confirm the replacement time of the internal power equipment of the power distribution board at an appropriate time, so that the service life of the internal equipment can be extended and the maximum performance can be maximized.

[0024] The effects of the present invention are not limited to those described above, but should be understood to include all effects that can be inferred from the detailed description of the present invention or the configuration of the invention as described in the claims. [Brief explanation of the drawing]

[0025] [Figure 1] This is a schematic diagram of an artificial intelligence-based power distribution panel device with abnormality detection and fire or failure prediction functions according to one embodiment of the present invention. [Figure 2] This is a flowchart illustrating how time-series sensing data is processed and analyzed in an artificial intelligence-based power distribution panel device according to one embodiment of the present invention. [Figure 3] This is a schematic diagram of an artificial intelligence-based monitoring and prediction system for a power distribution panel according to one embodiment of the present invention. [Figure 4] This is a schematic flowchart illustrating how individual variable prediction AI models and multivariate prediction AI models based on two or more variables, stored in a database, individually or in combination, predict abnormal signs, fires, and / or failures in an artificial intelligence-based monitoring and prediction system for power distribution panels according to one embodiment of the present invention. [Figure 5] This diagram shows an exemplary UI screen displayed to an operator on a mobile device in an artificial intelligence-based monitoring and prediction system for a power distribution panel according to one embodiment of the present invention. [Figure 6] This is a schematic flowchart of an artificial intelligence-based method for predicting abnormal signs, fires, or malfunctions in power distribution panels according to one embodiment of the present invention. [Figure 7] This is a schematic flowchart of an artificial intelligence-based monitoring and prediction system for a power distribution panel according to one embodiment of the present invention, which, after an abnormal sign, fire, or malfunction is predicted, displays this to the inspector or controls the power distribution panel accordingly. [Figure 8]This diagram shows an exemplary UI screen in an artificial intelligence-based monitoring and prediction system for a power distribution panel according to one embodiment of the present invention, in which fire / fault prediction and power monitoring are performed on a power control and monitoring platform. [Figure 9] This diagram shows an exemplary UI screen in an artificial intelligence-based monitoring and prediction system for a power distribution panel according to one embodiment of the present invention, in which abnormal signs are identified and fire / failure risk is predicted on the power control and monitoring platform. [Modes for carrying out the invention]

[0026] The present invention will be described below with reference to the attached drawings. However, the present invention can be embodied in various different forms and is therefore not limited to the embodiments described herein. In order to clearly illustrate the present invention with reference to the drawings, parts unrelated to the description have been omitted, and similar parts have been given the same reference numerals throughout the specification.

[0027] Throughout the specification, when a part is described as being "connected (linked, in contact with, or joined)" to another part, this includes not only cases where they are "directly connected" but also cases where they are "indirectly connected" through other components in between. Furthermore, when a part is described as "containing" a component, this means, unless otherwise stated, that it further comprises other components rather than excluding them.

[0028] The terms used herein are used solely to describe specific embodiments and are not intended to limit the invention. Singular expressions include plural expressions unless the context clearly indicates otherwise. In this specification, terms such as “includes” or “having” should be understood to indicate the presence of features, numbers, steps, operations, components, parts, or combinations thereof described in the specification, without prejudice to the presence or possibility of adding one or more other features, numbers, steps, operations, components, parts, or combinations thereof.

[0029] The artificial intelligence-based power distribution panel device having abnormality sign detection and fire or failure prediction functions of the present invention may include the following two types. First, an independent AI power distribution panel device that has the capability to detect abnormality signs, analyze and predict for fire or failure within the power distribution panel device, and includes a module for composite fire / failure monitoring and a fire / failure alarm control module for the power distribution panel device, and does not require a separate external analysis big data server. Second, an artificial intelligence-based monitoring and prediction system configured such that primary abnormality sign detection and fire or failure prediction / analysis are performed within the AI ​​power distribution panel device, and secondary detailed analysis / prediction is performed on an external big data server (e.g., an artificial intelligence-based power control and monitoring server) using the data sensed from the power distribution panel device and / or the primary analysis / prediction results.

[0030] However, the power distribution panel may consist of a low-cost power distribution panel equipped only with a composite fire / fault monitoring module, and in the artificial intelligence-based monitoring and prediction system, the power distribution panel may consist of a low-cost power distribution panel, and the determination of abnormal signs and prediction / analysis of fire or fault may be performed only by an external artificial intelligence-based power control and monitoring server based on sensing data collected by the composite fire / fault monitoring module of the low-cost power distribution panel.

[0031] Embodiments of the present invention will be described below with reference to the drawings.

[0032] Figure 1 is a schematic diagram of an artificial intelligence-based power distribution panel device having abnormality indicator detection and fire or failure prediction functions according to one embodiment of the present invention, and Figure 2 is a flowchart showing how time-series sensing data is processed and analyzed in an artificial intelligence-based power distribution panel device according to one embodiment of the present invention.

[0033] Referring to Figure 1, an artificial intelligence-based power distribution panel device 100 according to one embodiment of the present invention includes one or more sensor units 110, a data acquisition unit 120, an analysis processing unit 130, a control unit 140, a communication unit 150, and an output unit 160, and may further include a model database unit 170, a combined fire / fault monitoring unit 180, or a fire / fault alarm and control unit 190.

[0034] The sensor unit 110 is installed inside the power distribution panel and can sense the internal environmental state and power state in a time series, generate sensing data, and transmit it to the data collection unit 120. Referring to Figure 2, the environmental data indicating the internal environmental state is data sensed in a time series by at least one of the individual environmental sensors or composite sensors 112 such as a gas sensor, flame sensor, temperature sensor, humidity sensor, vibration sensor, rotation sensor, and shock sensor, and includes at least one of the following: gas content data, flame sensing data, temperature data, humidity data, gradient / vibration / acceleration data, and open sensing data, but is not limited to these as long as it indicates the internal environmental state of the power distribution panel. The power data indicating the power state is data sensed in a time series by, for example, a current sensor and / or voltage sensor 114 such as a CT sensor, and includes at least one of the following: current data, voltage data, active / reactive power data, power factor data, drive time data, leakage current data, frequency / phase angle data, current unbalance data, voltage unbalance data, harmonic data, and ultrasonic data, but is not limited to these.

[0035] According to one embodiment of the present invention, the sensor of the sensor unit 110 may be an IoT sensor, and sensing data may be transmitted wirelessly to the data collection unit 120.

[0036] The data acquisition unit 120 can receive, collect, and store the time-series sensing data from the sensor unit 110 via wired or wireless connection. The data acquisition unit 120 may perform digital processing on the collected sensing data, such as preprocessing (filtering, exception handling, and sensing error removal) or virtual data augmentation generation.

[0037] Referring to Figure 2, in a power distribution panel device 100 according to one embodiment of the present invention, the data acquisition unit 120 includes an acquisition module 122 and a preprocessing module 124, and may further include a virtual sensor module 126. The acquisition module 122 can collect and store sensing data by communicating with at least one sensor 112, 114 of the sensor unit 110 via wired or wireless means.

[0038] The preprocessing module 124 performs preprocessing on the sensing data, and the preprocessing performed may include at least one of the following: missing values ​​removal and noise removal. Time series data generally has two inherent characteristics: time step, which is specified at regular time intervals from start to finish, and a time difference between observed values. The current observed value is represented by a lag value (Lag), and the observed values ​​exhibit autocorrelation or serial correlation. In the process of analyzing time series data, if the data fluctuates greatly or is not constant over time, it exhibits non-stationarity. If this data is trained as a machine learning model without preprocessing, problems such as simple backward prediction, performance degradation, and errors occur. Therefore, for accurate prediction through a machine learning model, time series data requires missing values ​​removal and / or noise removal before input. Furthermore, in the case of instructional learning models, for example, time series data must be preprocessed so that the data is processed in pairs of target variables to be predicted and input variables used during prediction. In order for machine learning / learning models to train stably, an additional preprocessing step is required to unify the scale of the data.

[0039] The data acquisition unit 120 may further include a virtual sensor module 126 for augmenting / generating virtual sensing data. In such a case, the preprocessing module 124 can additionally perform processing based on the accumulated and stored sensing data to analyze whether data errors have occurred in recently received sensing data due to abnormalities in the power distribution device or the external environment. If the preprocessing module 124 determines that there is a data error, the virtual sensor module 126 can generate virtual sensing data at the time the data error occurred, based on the preprocessed sensing data sensed up to the time the data error occurred, through a virtual sensor AI model, store it in the acquisition module 122, and transmit it to the analysis processing unit 130 in place of the data sensed at the time the data error occurred. This means that if a signal different from the actual signal is transmitted to the data acquisition unit 120 due to a sensor failure, sensor communication error, etc., and the data transmitted deviates significantly from the existing pattern that does not reflect the fire failure characteristics in the preprocessing module 124, then sensor and communication errors are assumed, and the sensor and communication parts are replaced with a virtual sensor module until the errors are corrected and repairs are made, thereby augmenting and generating the necessary data. The virtual sensor AI model is a machine learning or deep learning model, and is a model that augmentes and generates virtual data based on existing analyzed data.

[0040] The analysis processing unit 130 may include a prediction module 132 that predicts first abnormal signs and fire or failure risk of the power distribution panel equipment through a first prediction AI model based on the preprocessed sensing data. The prediction module 132 may further predict fire factors and / or failure factors. Fire factors are identified, for example, by analyzing time-series sensing data through a first prediction AI model to identify features such as arcs, discharges, and insulation degradation. Failure factors are identified by analyzing time-series sensing data through a first prediction AI model to identify signs such as vibrations, shocks, temperature, humidity, and gases. The prediction module 132 may also predict equipment abnormal signs by performing deep learning analysis through a first prediction AI model on time-series sensing data adapted to the load characteristics of the customer.

[0041] The first predictive AI model used in the prediction module 132 is a machine learning or deep learning model, and can be generated by training it using existing sensing data from a power distribution panel, using at least one of a variety of existing artificial intelligence (AI) models, such as existing guided learning models, unguided learning models, reinforcement learning models, and time-series machine learning models. According to one embodiment of the present invention, the first predictive AI model is at least one of a pre-trained model, a customer / load adaptive model, and an iterative optimization model. The customer / load adaptive model is generated by retraining at least one of the existing pre-trained predictive models based on the pre-processed sensing data generated during a period in which a predetermined situation occurs, and the iterative optimization model can be generated by repeatedly retraining at least one of the existing pre-trained predictive models based on the pre-processed sensing data generated during a predetermined period in accordance with the operation or equipment loading status of the power distribution panel device.

[0042] According to one embodiment of the present invention, the first predictive AI model may be stored in a model database unit 170 inside the power distribution unit 100 in the case of an independent power distribution unit, or, according to another embodiment of the present invention, a predictive AI model learned from an external server may be provided. The first predictive AI model may be learned and generated in the analysis processing unit 130 of the power distribution unit 100, or it may be learned and generated from an external server based on existing sensory learning data. According to one embodiment of the present invention, the first predictive AI model may consist of at least one of a predictive AI model for individual variables and a multivariate predictive AI model based on two or more variables, or it may consist of multiple models by integrating these.

[0043] The analysis processing unit 130 may further include a time series analysis module 134 that, based on the pre-processed sensing data, analyzes time series trends, for example through deep learning time series analysis and anomaly detection, and transmits anomalies or alarms to the display unit when time series anomalous signal data or numerical data outside of a previously set range is generated. The time series analysis module integrates time series transition analysis and various anomaly detection algorithms to create a model and make predictions that match the basic characteristics of the power distribution board, and can automatically distribute prediction models that match the customer / load characteristics through periodic model retraining / upgrading when more precise predictions are needed.

[0044] The analysis processing unit 130 may further include a learning module 136 for training or retraining the first predictive AI model stored in the model database unit 170 or received from an external server. By retraining or augmenting (updating) the first predictive AI model in accordance with the load and demand characteristics, the prediction module 132 can perform more precise predictions.

[0045] The control unit 140 can control the power distribution panel 100 based on the predicted abnormal signs and the risk of fire or failure. The control of the power distribution panel 100 includes, for example, at least one of the following: shut-off control for specific power equipment inside or connected to the power distribution panel, door closing / opening control, fan operation control, positive pressure device control, fire extinguishing device operation control, constant temperature / humidity / dehumidification device operation control, and further includes, but is not limited to, control of warning devices such as LEDs, LCDs, sounds, lasers, or display units 160 that can be attached to the power distribution panel, or operation control of alarm sounds, and control for transmitting alarms to the mobile devices of personnel or to relevant organizations such as fire stations and police stations via the communication unit 150.

[0046] The communication unit 150 can transmit the sensing data or the predicted abnormal signs and fire or failure risk to an external server or the operator's mobile device. The communication unit 150 may transmit an alarm or fire or failure-related information to relevant organizations such as fire departments or police stations if abnormal signs are detected in the power distribution panel or if a fire or failure is predicted. The communication unit 150 may receive control signals or data necessary for control of the power distribution panel from an external server or the operator's mobile device.

[0047] The display unit 160 can display the abnormal signs and the risk of fire or malfunction on a display or notify through an alarm sound. The display unit 160 may also indicate an alarm through a change in the hue of the illumination, and the method of indication is not limited thereto.

[0048] The model database unit 170 may store at least one of the following: a predictive AI model for individual variables that have been learned and used in the first predictive AI model; a multivariate predictive AI model based on two or more variables; and / or an integrated model thereof.

[0049] Figure 3 is a schematic diagram of an artificial intelligence-based monitoring and prediction system 10 for a power distribution panel according to one embodiment of the present invention.

[0050] Referring to Figure 3, the monitoring and prediction system 10 includes a power control and monitoring server 200 connected by wire or wirelessly to a power distribution panel 100 according to one embodiment of the present invention disclosed in Figure 1, and may further include mobile devices 300 connected wirelessly to these. The power control and monitoring server 200 may be an independent physical server or a cloud server that constitutes part of the space within the cloud.

[0051] The power control and monitoring server 200 receives the pre-processed sensing data or first abnormality signs and fire or failure risk from the power distribution device, predicts the second abnormality signs and fire or failure risk of the power distribution device 100 through a second predictive AI model, and can transmit the predicted result data to the power distribution device 100 or a mobile device 300. The power control and monitoring server 200 according to one embodiment of the present invention includes a big data server unit 210, a database unit 220, an analysis server unit 230, and may further include a display unit 240. The power control and monitoring server 200 may further include management software for a power control and monitoring platform 250, through which it performs power control and monitoring of the power distribution device 100, which may be used to further facilitate this.

[0052] The big data server unit 210 can collect and store the pre-processed sensing data or the first abnormality signs and fire or failure risk received from the power distribution panel device 100.

[0053] The database unit 220 may store trained predictive models usable for a first predictive AI model used in the power distribution device or a second predictive AI model used in the power control and monitoring server 220's analysis server unit 230. The database unit 220 may also store cumulative data of sensed data from the power distribution device or other environmental or power distribution-related data supplied to an external server for training or retraining the stored predictive models. The database unit 220 may store at least one of trained individual variable predictive AI models used in the first or second predictive AI model, a multivariate predictive AI model based on two or more variables, and / or integrated models thereof.

[0054] The analysis server unit 230 can predict the second abnormality sign and fire or failure risk of the power distribution panel device 100 through a second predictive AI model based on the preprocessed sensing data from the big data server unit 220 or the first abnormality sign and fire or failure risk, and perform remote precision analysis thereon. The second predictive AI model can be generated by training it using existing sensing data of the power distribution panel, using at least one of a variety of existing machine learning models, such as an existing guided learning model, an unguided learning model, a reinforcement learning model, or a time-series machine learning model. According to one embodiment of the present invention, the second predictive AI model is at least one of a pre-trained model, a customer / load adaptive model, and an iterative optimization model.

[0055] The analysis server unit 230 may train the prediction models used for the first prediction AI model and / or the second machine learning model based on existing sensing datasets for the power distribution panel 100 or training data supplied from an external server, and may further retrain the models for optimization. Referring to Figure 4, the analysis processing unit 230 may retrain the first prediction AI model or the second prediction AI model based on the pre-processed sensing data generated during a period in which a predetermined situation occurs to generate the customer / load adaptive model, and may repeatedly retrain at least one of the pre-trained prediction models based on the pre-processed sensing data generated during a predetermined period in accordance with the operation or equipment input status of the power distribution panel 100 to generate an iterative optimization model for model optimization.

[0056] In the monitoring and prediction system 10 according to one embodiment of the present invention, the analysis processing unit 130 of the power distribution panel 100 performs a primary predictive analysis for abnormal signs, fire, and failure in the power distribution panel 100, and the power monitoring and control server 200 performs a secondary, detailed predictive analysis for abnormal signs, fire, and failure in the power distribution panel 100, in which case the secondary, detailed predictive analysis may be performed using a customer load-adaptive AI model. However, in the monitoring and prediction system 10 according to another embodiment of the present invention, the power distribution panel 100 performs simple sensing and simplified analysis, and based on such data, the power monitoring and control server 200 performs a detailed predictive analysis for abnormal signs, fire, and failure in the power distribution panel 100.

[0057] The display unit 240 can display the second abnormality signs and fire or failure risk for the power distribution unit 100 predicted by the analysis server unit 230 (see Figure 9 below). The display unit 240 may also display the results of monitoring the power status of the power distribution unit 100 (see Figure 10 below).

[0058] In addition, the power control and monitoring server 200 can visualize various data, power quality / energy monitoring, display fire / failure risk, inspection logs, diagnostic schedule setting, predictive maintenance policy setting corresponding to the predicted second abnormality signs and fire or failure risk for the power distribution panel device 100, and analyze statistics / real-time torrents, and display them through the display unit 240.

[0059] Figure 4 is a schematic flowchart showing how, in an artificial intelligence-based monitoring and prediction system for a power distribution panel according to one embodiment of the present invention, the individual variable prediction AI model 222 and the multivariate prediction AI model 224 based on two or more variables, stored in the database unit 220, individually or in combination, perform abnormality signs, fires, and / or failure predictions.

[0060] Referring to Figure 4, the database unit 220 may store at least one of the following: a first predictive AI model 222 for individual variables used as a first predictive AI model used in the analysis processing unit 130 or a second predictive AI model used in the analysis server unit 230; a multivariate predictive AI model 224 based on two or more variables; and / or an integrated model thereof.

[0061] According to one embodiment of the present invention, the predictive AI model 222 for individual variables includes, but is not limited to, at least one of the following: an individual anomaly detection model (which detects signs of equipment abnormality or signs of electrical quality abnormality), a series / parallel arc (fire cause) prediction model, a contact resistance increase (fire / failure cause) prediction model, an insulation degradation (fire / failure cause) prediction model, a partial discharge (fire cause) prediction model, a power quality prediction model, a gas / flame prediction model, an impact (failure cause) / earthquake prediction model, and a high temperature / overheating / water inundation prediction model. Furthermore, according to one embodiment of the present invention, the multivariate predictive AI model 224 includes, but is not limited to, at least one of the following: a composite fire prediction model, a composite failure prediction model, a composite power quality prediction model, an energy / load prediction model, and a predictive maintenance model.

[0062] Referring again to Figure 3, the monitoring and prediction system 10 may further include a mobile device 300 for a person responsible for inspecting or controlling the power distribution unit 100. The mobile device 300 displays a first abnormality sign and fire or failure risk predicted by the analysis processing unit 130 of the power distribution unit 100, or a second abnormality sign and fire or failure risk predicted by the analysis server unit 230 of the power control and monitoring server 200, and can receive control information from the user based on such prediction information and transmit it to the power distribution unit 100. The mobile device 300 may provide statistics / real-time trends for time-series sensed data of the power distribution unit 100, perform inspection logging, or perform AR-recognized digital twin.

[0063] Figure 5 shows an exemplary UI screen displayed on a mobile device 300 in an artificial intelligence-based monitoring and prediction system 10 for a power distribution panel according to one embodiment of the present invention.

[0064] Figure 6 is a schematic flowchart of an artificial intelligence-based method for predicting abnormal signs, fires, or failures in power distribution panels according to one embodiment of the present invention.

[0065] Referring to Figures 1 and 6, an embodiment of the present invention provides an artificial intelligence-based method for predicting abnormal signs, fire, or failure of a power distribution panel 100, which may include: step S610 generating sensing data that senses the internal environmental state and power state of the power distribution panel 100 in a time series using a sensor unit 110 attached to the power distribution panel; step S620 collecting the time series sensing data from the sensor unit 110; step S630 performing preprocessing on the collected sensing data; step S640 predicting first abnormal signs and fire or failure risk of the power distribution panel 100 through a first predictive AI model based on the preprocessed sensing data; step S650 displaying the abnormal signs and fire or failure risk on a display or notifying via an alarm sound; and step S660 controlling the power distribution panel 100 based on the predicted abnormal signs and fire or failure risk.

[0066] The environmental data indicating the internal environmental state is, for example, data sensed in time series by at least one of individual environmental sensors or composite sensors 112 such as a gas sensor, flame sensor, temperature sensor, humidity sensor, vibration sensor, rotation sensor, or shock sensor, and includes, but is not limited to, at least one of gas content data, flame sensing data, temperature data, humidity data, gradient / vibration / acceleration data, and open sensing data, as long as it indicates the internal environmental state of the power distribution panel. The power data indicating the power state is, for example, data sensed in time series by a current sensor and / or voltage sensor 114 such as a CT sensor, and includes, but is not limited to, at least one of current data, voltage data, active / reactive power data, power factor data, drive time data, leakage current data, frequency / phase angle data, current unbalance data, voltage unbalance data, harmonic data, and ultrasonic data. In the preprocessing step, the preprocessing may include at least one of missing values ​​removal and noise removal.

[0067] The method may further include the step of analyzing whether data errors have occurred due to abnormalities in the power distribution device 100 or the external environment in the recently received sensing data based on the sensing data accumulated in the data acquisition unit 120, and if it is determined that there is a data error, generating virtual sensing data at the time the data error occurred through a virtual sensor machine learning model based on the pre-processed sensing data sensed up to the time the data error occurred, and transmitting it to the analysis processing unit.

[0068] The method may further include, based on the preprocessed sensing data, analyzing the time-series trend if time-series anomaly data or numerical data outside a pre-defined range occurs, and transmitting an anomaly indication or alarm to the display unit.

[0069] Another embodiment of the present invention provides an artificial intelligence-based method for predicting abnormal signs, fires, or failures in a power distribution panel 100, which may include, in addition to the steps of the method shown in Figure 6, the steps of: transmitting the sensing data or the predicted abnormal signs and fire or failure risk to a power control and monitoring server 200 or a mobile device 300 of an employee; collecting the pre-processed sensing data or the first abnormal signs and fire or failure risk received from the power distribution panel 100 at the power control and monitoring server 200; receiving the pre-processed sensing data or the first abnormal signs and fire or failure risk from the power distribution panel 100 at the power control and monitoring server 200 and predicting the second abnormal signs and fire or failure risk of the power distribution panel 100 through a second predictive AI model; and transmitting the predicted second abnormal signs and fire or failure risk to the power distribution panel 100 and / or a mobile device 300 of an employee. The method may further include the step of displaying the second abnormality indicator and the fire or failure risk for the power distribution panel 100 predicted by the analysis server unit 230 on the display unit 240.

[0070] The second predictive AI model can be generated by training it using existing sensing data from a power distribution panel, using at least one of a variety of existing machine learning models, such as existing instructional learning models, uninstructed learning models, reinforcement learning models, and time-series machine learning models. According to one embodiment of the present invention, the second predictive AI model is at least one of a pre-trained model, a customer / load adaptive model, and an iterative optimization model.

[0071] A method for predicting abnormal signs, fires, or failures of a power distribution panel 100 based on artificial intelligence according to yet another embodiment of the present invention may further include, after the power control and monitoring server 200 transmits the predicted second abnormal sign and fire or failure risk to the power distribution panel 100 and / or the mobile device 300 of the person in charge, the mobile device 300 transmits control information to the power distribution panel 100 based on information regarding the first abnormal sign and fire or failure risk, or the second abnormal sign and fire or failure risk, predicted by the analysis processing unit 130 of the power distribution panel 100.

[0072] This method enables the establishment of effective inspection strategies and optimal inspection schedules through abnormal signs, fire, or fault prediction analysis, as well as worker safety and inspection log analysis through door detection sensors / charging sensors. In addition to fault detection, it can also detect earthquakes or shocks through environmental sensing, allowing for rapid response and minimizing power outages, for example, in the case of power facilities.

[0073] Figure 7 is a schematic flowchart showing how an artificial intelligence-based monitoring and prediction system 10 for a power distribution panel, according to one embodiment of the present invention, predicts abnormal signs, fires, or malfunctions, and then displays these to an inspector or controls the power distribution panel accordingly.

[0074] Referring to Figure 7, the analysis processing unit 130 of the power distribution panel 100 according to one embodiment of the present invention predicts abnormal signs, fire, or failure risk-related prediction information, which is transmitted to the control unit 140 inside the power distribution panel 100, which controls the environmental conditions or power of the power distribution panel 100, or displays or alarms on the display unit 160. In this case, the display unit 160 can display at least one of the following information based on the prediction information: fire risk, failure risk, maintenance / diagnosis schedule, power quality information, and energy monitoring information.

[0075] The predictive information may be further transmitted to a fire / fault alarm and control unit 190, which is also included in the power distribution panel 100. Based on the transmitted predictive information, the fire / fault alarm and control unit 190 can generate shutoff control signals, door control signals, fan control signals, positive pressure device control signals, fire extinguishing device control signals, display / sound alarm control signals, and alarms to relevant organizations such as fire stations or police stations, and can be controlled so that even relatively unskilled inspectors can take appropriate preventive measures in response to the predictive information.

[0076] Furthermore, the prediction information generated by the analysis processing unit 130 may be transmitted via the communication unit 150 to an external power control and monitoring server 200 or to the mobile device 300 of the person in charge of the work.

[0077] Figure 8 shows an exemplary UI screen in an artificial intelligence-based monitoring and prediction system for a power distribution panel according to one embodiment of the present invention, in which fire / fault prediction and power monitoring are performed on the power control and monitoring platform. Figure 9 shows an exemplary User Interface (UI) screen in an artificial intelligence-based monitoring and prediction system for a power distribution panel according to one embodiment of the present invention, in which abnormality signs are judged and fire / fault risk is predicted on the power control and monitoring platform.

[0078] Figure 9 shows an example screen displayed when the "AI Prediction" button is clicked among the top items on the UI screen for a specific power distribution panel 100 on a power control and monitoring platform according to one embodiment of the present invention, and Figure 10 shows an example screen displayed when the "Monitoring" item is clicked among the top items. However, the displayed UI screen is not limited to these forms.

[0079] The technology applied to the artificial intelligence-based power distribution panel 100 in Figures 1 and 2, or the monitoring and prediction system 10 in Figure 3, can be applied not only to the power distribution panel 100, but also to artificial intelligence-based solar power connection panel devices or artificial intelligence-based electric vehicle charging systems for detecting abnormal signs, and for monitoring and predicting fires or malfunctions.

[0080] The monitoring and prediction systems or methods for power distribution panels described above may be embodied by hardware components, software components, and / or combinations of hardware and software components, or by including a cloud server. For example, the systems, devices, methods, and components described in the embodiments may be embodied using one or more general-purpose or special-purpose computers, such as a processor, controller, central processing unit (CPU), graphics processing unit (GPU), arithmetic logic unit (ALU), digital signal processor, microcomputer, field programmable gate array (FPGA), programmable logic unit (PLU), microprocessor, application-specific integrated circuits (ASICS), server, or other computer devices that execute and respond to instructions.

[0081] The monitoring and prediction method for a power distribution panel according to an embodiment of the present invention is embodied in the form of program instructions delivered through various computer means and can be recorded on a non-temporary computer-readable medium. The non-temporary computer-readable medium may include program instructions, data files, data structures, etc., individually or in combination. The program instructions recorded on the medium may be specifically designed and configured for the embodiment, or they may be publicly known and available to those skilled in the computer software art. The hardware device is configured to operate as one or more software modules to perform the operation of the embodiment, and vice versa.

[0082] As described above, even if embodiments are described by limited drawings, a person with ordinary skill in the art can make various modifications and variations from the above description. For example, even if the described technique is performed in a different order than described, and / or the components of the described system, structure, apparatus, circuit, etc. are combined or assembled in a different manner than described, or substituted or replaced by other components or equivalents, suitable results can still be achieved.

[0083] The scope of the present invention is defined by the claims described below, and all modifications or alterations derived from the meaning and scope of the claims and the concept of equivalents thereof should be interpreted as being included within the scope of the present invention.

Claims

1. An artificial intelligence-based power distribution panel device, A sensor unit is installed inside the aforementioned power distribution panel and senses the internal environmental state and power state in a time series to generate sensing data. A data collection unit collects the sensing data from the sensor unit and performs preprocessing on the sensing data, An analysis processing unit predicts the first abnormal signs and fire or failure risk of the power distribution panel device through a first predictive AI model based on the pre-processed sensing data, Based on the predicted abnormal signs and the risk of fire or malfunction, a control unit controls the power distribution panel, A communication unit that transmits the aforementioned sensing data or the predicted abnormal signs and the risk of fire or failure to an external server or mobile device, A device including a display unit that displays or notifies the aforementioned abnormal signs and the risk of fire or malfunction.

2. The sensing data for the internal environmental state includes at least one of the following: gas content data, flame sensing data, temperature data, humidity data, gradient data, vibration data, acceleration data, and openness sensing data. The apparatus according to claim 1, wherein the sensing data for the power state includes at least one of the following: current data, voltage data, active / reactive power data, power factor data, operating time data, frequency / phase angle data, harmonic data, and ultrasonic data.

3. The apparatus according to claim 1, wherein the preprocessing includes at least one of a missing values ​​removal process and a noise removal process.

4. The aforementioned data acquisition unit, The apparatus according to claim 1, further comprising a virtual sensor module that, based on the accumulated sensing data from the data collection unit, analyzes the recently received sensing data for any data errors due to an abnormality in the power distribution panel device, and, if it is determined that there is a data error, generates virtual sensing data at the time the data error occurred through a virtual sensor AI model based on the pre-processed sensing data sensed up to the time the data error occurred, and transmits it to the analysis processing unit.

5. The aforementioned analytical processing unit The apparatus according to claim 1, further comprising a time-series analysis module that analyzes the time-series trend and transmits an abnormality indication or alarm to the display unit when time-series unusual signal data or numerical data outside a previously set range is generated based on the pre-processed sensing data.

6. The apparatus according to claim 1, wherein the first predictive machine AI model is at least one of a pre-trained model, a demand / load adaptive model, and an iterative optimization model.

7. A monitoring and prediction system for power distribution panels, The artificial intelligence-based power distribution panel device according to claim 1, A system comprising: a power control and monitoring server that receives the pre-processed sensing data or the first abnormality sign and fire or failure risk from the power distribution panel, and predicts the second abnormality sign and fire or failure risk of the power distribution panel through a second predictive AI model and transmits it to the power distribution panel.

8. The system according to claim 7, further comprising a mobile device that receives and displays a second abnormality indication and fire or failure risk of the power distribution panel, receives control information based thereon from a user, and transmits the inputted control information to the power distribution panel.

9. The aforementioned power control and monitoring server is A big data server unit receives and stores the pre-processed sensing data or the first abnormality sign and fire or failure risk received from the power distribution panel device, A database unit in which machine learning models usable as the first or second predictive AI model are stored, The system according to claim 7, further comprising: an analysis server unit that predicts a second abnormal sign and fire or failure risk of the power distribution panel device through a second predictive AI model based on the preprocessed sensing data from the big data server unit or the first abnormal sign and fire or failure risk.

10. The system according to claim 7, wherein the second predictive AI model comprises at least one of an individual variable predictive AI model and a multivariate predictive AI model based on two or more variables, or comprises a plurality of models by integrating these.

11. The analysis server unit then processes the pre-processed sensing data as follows: (i) Predicting the location of individual fires or faults through individual variable prediction AI models, (ii) Integrating and applying the individual variable prediction AI model and the multivariate prediction AI model based on two or more variables to predict the risk of fire or malfunction, and (iii) The apparatus according to claim 9, which performs at least one of the following: applying the multivariate prediction AI model to predict abnormal signs and fire or failure risk, and inputting the predicted abnormal signs and fire or failure risk into a predictive maintenance model to derive a corresponding maintenance policy.

12. The aforementioned analysis server, (a) Based on the preprocessed sensing data, train or retrain at least one of the models stored in the database unit. (b) Based on the pre-processed sensing data generated during the period in which the previously defined conditions occur, at least one of the models stored in the database unit is retrained to generate a consumer / load adaptive model, and (c) The apparatus according to claim 9, further comprising at least one of the following: repeatedly retraining at least one of the models stored in the data unit based on the preprocessed sensing data generated during a predetermined period in accordance with the operation status of the power distribution panel or the equipment activation status, thereby generating an iteratively optimized model.

Images