System for predicting status and notifying of faults in mine ventilation fans

Abstract

Described is a smart system for predicting the condition of a mine ventilation fan and reliably notifying of operating conditions, both in real time and predictively, based on neural networks. At least one sensor per control parameter captures and sends information or data regarding the performance of the ventilation fan to a web application database in which same are stored, and to a microprocessor located inside a smart panel and in which the data are processed by a computer program based on a predictive artificial intelligence model. The processed data are subsequently sent to and stored in the web application database, which comprises data sent by the sensors and the processed information and data. The sensors are selected from one or more of temperature sensors, vibration sensors and electric power sensors. The data sent by the sensors, the data processed in the microprocessor and the previously gathered and stored data enable the predictive artificial intelligence model to be trained. The smart system also generates a fault alarm or alert, selected from one or more of: a visible alert, an audible alert and a text alert.

Description

[0001] SYSTEM FOR PREDICTING THE STATUS AND ALERTING OF FAILURES IN MINING FANS

[0002] FIELD OF INVENTION

[0003] The present invention relates to mining, and in particular, refers to an intelligent system for reliably predicting and altering the operating conditions, both in real time and predictively, of the vital state or operating condition of an underground mining fan, using deep learning (neural networks).

[0004] BACKGROUND

[0005] Technologies based on computational intelligence have experienced strong development and it is possible to find engineering companies that offer the development of predictive studies of diverse natures, however, they only reach a level of specificity that fails to address a complex problem such as that associated with the operation of mining fans or civil works.

[0006] It is well known that fans are critical equipment, and predicting the behavior of main fans provides significant savings for mining companies in the event of failures. These companies suffer millions in losses due to the shutdown of mining operations caused by unscheduled fan failures, which force the evacuation of all mining personnel. This translates into lower productivity, reduced output, and lower revenue for the companies. Knowing the future condition of these fans is crucial for providing better safety and confidence to workers, as well as preventing unscheduled shutdowns that cost enormous sums of money.

[0007] In the prior art, the following patent publications can be noted: CN117150414B (Guangdong Sohoo Technology Co ltd) refers to a fault diagnosis method, comprising the steps of data acquisition and preprocessing (S1), feature extraction and selection (S2), fault diagnosis model training and evaluation (S3), remote monitoring and diagnosis (S4), integrated hardware diagnostic function (S5), intelligent diagnostic tool (S6) and software application, integrated fault prediction function (S7), fault repair and maintenance (S8), data transmission (S9) to a mobile terminal, virtual power system simulation environment configuration (S10), and automatic fault repair (S11), where an automatic algorithm and artificial intelligence technology perform automatic fault diagnosis, so that faults are located quickly and accurately.and provide the corresponding solutions. In particular, a power supply system is monitored and diagnosed in real time via internet connection and remote access technology, so that maintenance personnel can remotely troubleshoot, reduce downtime, and obtain real-time fault information for subsequent monitoring, analysis, and prevention. The user has an interface that allows for easy operation and fault diagnosis using the intelligent diagnostic tool of the discussed invention. CN108803552B (Chengde Jianlong Special Steel Co Ltd) describes a system and method for monitoring equipment faults, comprising: a data acquisition node, a data processing and storage server, and a data processing and analysis module.where the data acquisition node is used to acquire information about the state of the industrial equipment, predicting the probability of failure and the time of failure for each piece of industrial equipment based on the equipment state information through a pre-established prediction model, and sending the probability of failure, the time of failure, and the equipment state information to the data processing and storage server; and the data processing and storage server is used to store the equipment state information, display the equipment state information, the probability of failure, and the time of failure in real time,The system corrects the prediction model based on the equipment's status information and sends the corrected model prediction to the data acquisition node. The fault detection system is applicable to various types of industrial equipment and can be expanded to include improved accuracy and real-time performance.

[0008] CN111356910B (Nanoplus Sci) discloses systems and methods for monitoring the condition of rotating equipment. A sensor near the rotating equipment detects its vibration. The sensor generates a digital signal corresponding to the vibration and transmits it over the communication network. The server receives the digital signal and preprocesses it using empirical joint mode decomposition (EEMD) techniques. The server processes the digital signals using a wavelet neural network (WNN) to detect a fault in the rotating device. Additionally, the server processes the digital signal using a wavelet neural network to predict the remaining useful life (RUL) of the rotating equipment.

[0009] CN118423264A (Sinopec Shengli Petroleum Engineering Corp Southwest Branch) refers to a method and device for detecting faults in a mud pump, where the method comprises: collecting sound data from the mud pump in an operating state and transmitting the sound data (S1); receiving and storing sound data and establishing a database (S2); analyzing and processing sound data in the database by adopting an intelligent algorithm to judge whether the mud pump has faults or anomalies (S3); and when the fault of the mud pump is determined, the alarm is activated.By collecting sound data from the mud pump and combining it with an intelligent algorithm, sound faults can be automatically detected and diagnosed, helping field personnel find and resolve the problem more quickly. The operating status of the mud pump is monitored more accurately, fault problems are detected in time and an alarm is triggered, and the safety and reliability of the equipment are improved. CN218524197U (Xinjiang University) describes a device for monitoring the condition of an electric motor and diagnosing faults, comprising: a first fixed housing, a second fixed housing assembled on top of the first fixed housing, a display screen arranged inside the upper end of the second fixed housing, and an indicator light fixed on top of the second fixed housing.The positioning blocks are fixed to two sides of the lower portions of the two ends of the second mounting housing. These positioning blocks connect to the first mounting housing via a sliding mechanism, and arc-shaped positioning blocks are fixed to two sides of each positioning block. The interior of the first stationary housing houses a control module and a battery. The device can detect the motor's status during operation and alert the operating personnel via a warning light. The connection between the housing of the first assembly and the housing of the second assembly is secure due to the interlocking of the first housing assembly, the second housing assembly, and a positioning arc.

[0010] CN118200185A (Jiangsu Anjiang Equipment Co ltd) refers to a system and method for monitoring the safety status of mining equipment based on a wireless sensor network, which allows automatic monitoring of the operating status of mining equipment, thereby improving the efficiency of equipment fault processing, reducing maintenance costs and the risk of accidents.

[0011] US10983514B2 (Strong Force IoT Portfolio 2016 LLC) describes an apparatus, method, and system for data collection in a production environment.The system may include a data collector communicatively coupled to a plurality of input channels, wherein a first subset of the plurality of input channels is connected to a first set of sensors that measure operating parameters of a production component, a structured data storage for storing a plurality of collector paths and collected data, a structured data acquisition circuit for interpreting a plurality of detection values ​​from the data collected from the production component, and a structured data analysis circuit for analyzing the collected data and evaluating a first collection routine of the data collector based on the analyzed data, wherein based on the analyzed collected data the data collector performs a change of collection routine.

[0012] This intelligent system, a type of smart device or smart panel, can be flexibly and adaptively installed in a mining operation and used in mine control rooms, dispatch areas, or both. It features integrated alarms that can be easily and intuitively managed and provides predictive solutions for the vital status or operating condition of the fan through a captive portal, thus facilitating access to real-time predictions and fan behavior. Furthermore, this intelligent system includes a programmable logic controller (PLC) that can function as an actuator to connect with other equipment or devices at the mine site. This intelligent system is also highly reliable and durable, resistant to the demanding mining environments characterized by high humidity, heat, dust exposure, and other challenging conditions.

[0013] BRIEF DESCRIPTION OF THE INVENTION

[0014] The present invention relates to an intelligent system for reliably predicting and alerting, both in real time and predictively, the operating conditions of a mining or civil engineering fan, preferably an underground mining fan, using deep learning (neural networks). At least one sensor per control parameter captures and sends information or data related to the fan's performance to a web application database, where it is stored. A microprocessor located inside an intelligent device / panel processes this information or data using a computer program based on an artificial intelligence predictive model for the fan's performance. The processed information or data (prediction) is then sent to and stored in the web application database.This database comprises, in addition to the information or data sent by the sensors and the processed information and data, previously collected and stored information and data related to the operating status of the fans. The artificial intelligence predictive model varies depending on the type of fan and is calibrated based on the information or data sent by the sensors, the information or data processed by the microprocessor, and previously collected information or data related to the fan's performance, which is stored in a properly structured web application database located on the smart device / dashboard.

[0015] These sensors are selected from one or more temperature sensors, vibration sensors, electrical power sensors, and others. They can be located in different parts of the fan; preferably, the temperature sensor(s) are located in the center of the fan; the vibration sensor(s) are located at the base of the fan; and the electrical power sensor(s) are located at the top of the fan body. These sensors are preferably wireless and communicate remotely with the smart device / panel using the communication channels of the mining or civil engineering operation.

[0016] This intelligent system for reliably predicting and alerting on the operating conditions, both in real time and predictively, of the vital or operational status of a mining or civil engineering fan, also generates or issues: a normal status alert when the value of said information or processed data is less than the minimum threshold of a range of values ​​pre-established as the failure risk range; a failure alert when the value of said information or processed data is greater than the lower threshold of a range of values ​​pre-established as the failure risk range and is less than the upper threshold of said range of values ​​pre-established; and a critical failure alert when the value of the information or processed data is equal to or greater than the upper threshold of the range of values ​​pre-established.The preset values ​​comprise predefined values ​​for the control parameters, namely temperature, vibration, and electrical power. The range of preset values ​​varies as information or data related to the vital status of the fan is received from the sensors and processed by the device's / smart board's microprocessor.

[0017] Alerts are selected from a visual alarm, and optionally, an audible alarm, a text alarm, or both. A "No Fault" alert confirms the ventilator is functioning correctly and corresponds to a green light in the visual alert. A "Fault" alert warns of irregular ventilator performance and corresponds to a yellow light in the visual alert. A "Critical" alert warns of imminent ventilator failure and corresponds to a red light in the visual alert. See Fig. 4.

[0018] Furthermore, the present intelligent system has a programmable logic controller (PLC) that can function as an actuator to connect with other equipment or devices on the job site, and as an example, activate auxiliary lights in the mining tunnel or civil works synchronized with the lights of the intelligent device / panel associated with the monitored fan to warn in the environment of its vital status.

[0019] Brief Description of the Figures

[0020] Figure 1 shows the LSTM predictions for 500 neurons, considering temperature and time.

[0021] Figure 2 shows the LSTM predictions for 500 neurons, considering electrical power and time.

[0022] Figure 3 shows the LSTM predictions for 100 neurons, considering vibrations and time.

[0023] Figure 4 shows the LSTM predictions in a multivariable model, using a lag (number of steps that need to be known in order to predict the future step) of 45 hours.

[0024] Figure 5 shows the scheme of the alerts / alarms to be activated according to the vital status of the ventilator.

[0025] Figures 6A to 6D show the smart device / board from various views. Figure 6A shows the interior of the smart device / board and specific mountings for some components. Figure 6B shows the smart device / board partially opened and the internal board with all its components. Figure 6C shows the smart device / board with the display screen and visual alarm. Figure 6D shows the smart device / board from the side surface that has the power supply terminal.

[0026] Figures 7A to 7C show a possible location for the sensors on the smart device / board. Fig. 7A shows a possible location for the temperature sensor. Fig. 7B shows a possible location for the vibration sensor. Fig. 7C shows a possible location for the power sensor. Detailed Description of the Invention

[0027] This intelligent system for reliably predicting and alerting on the operating conditions of a mining fan, preferably an underground mining fan or civil engineering project, in both real-time and predictive time, comprises multiple sensors located at strategic points on the fan. These sensors allow the system to represent the operating conditions of this equipment based on control parameters, primarily temperature, vibration, and electrical power. The strategic points may vary in practice depending on the fan model and operating conditions. See Figs. 7A-7C. Additionally, it includes an intelligent device / panel of the appliance or accessory type, meaning it can begin operating upon connection to the fan control room server.This smart device / controller's main function is to process and analyze the information or data captured and sent by multiple sensors that indicate critical variables of the fan's vital status (temperature, electrical power, and vibrations). Based on this data, the device can provide predictions and display visualizations through its interface. Furthermore, the smart device / controller is designed to trigger an alarm, which can be a visual and audible alarm, when the predicted values ​​exceed the established limits for normal, alert, and critical states. See Figs. 6A-6D.

[0028] The smart device / tablet also includes an access point and captive portal where any user can connect from a mobile device such as cell phones or tablets, to make inquiries about the future status of the vital signs (critical variables or control parameters) of the ventilator, where the device delivers these predictions through tables and graphs that are displayed on the visualization screen.

[0029] The smart device / board comprises a microprocessor, preferably a SER microprocessor; a PLC, preferably an Industrial Shield PLC; a PLC charger; a Raspberry Pi; a router, preferably a D-Link router; a fan; a screen, preferably a touch LCD screen; a speaker; slotted channels; LED lights; and other conventional components. Each of these components is mounted on a metal board.

[0030] The functionality of each of the devices inside the dashboard, as well as their connections and links between them, is detailed below.

[0031] Raspberry Pi 4: Functions as an Access Point and Captive Portal, allowing the smart device / tablet to connect via Wi-Fi to the associated web platform. After entering their username and password, users can request predictions about the fan's vital status. Microprocessor: Processes data from sensors located on the fan and performs the predictive process. It also sends a signal to an Arduino (PLC) to trigger sound and light signals for visual and audible alarms, and sends the predictions back to the Raspberry Pi 4, which then forwards them to the user.

[0032] Arduino PLC: It acts as an actuator, emitting a light signal based on the predicted future state of the fan. If the predictions fall within the first threshold, they are considered normal and a green light is displayed. If, on the other hand, they exceed this first threshold (fault alert state), a yellow light is displayed. Finally, if they exceed the second threshold, a red critical state light is activated.

[0033] Router Creates a permanent communication channel between the Raspberry Pi and the microprocessor, as well as an artificial network of its own for the device / smart board that serves as a backup, since, if the mining network where the device / smart board is located fails or is interrupted, this network replaces it and thus maintains a continuous communication of data between the microprocessor, the Raspberry Pi and the users of the device / smart board.

[0034] Display screen: This is the primary communication channel between the user and the microprocessor. It allows the user to view and interact with the device's vital component (the microprocessor), enabling them to observe predictions, the current status of the ventilator's vital signs, and other important data for monitoring and controlling the ventilator. A touchscreen is preferred.

[0035] UPS: It is a backup power supply that prevents equipment from being damaged by voltage and current fluctuations in the mine. Sudden and unexpected voltage changes are common in underground mining, so this device ensures a constant voltage for the five devices connected to it: Microprocessor, Raspberry Pi 4, PLC, Router, and the display screen, preferably a touch screen.

[0036] The components described above are mounted on a board with custom-designed fasteners for each component, manufactured using a 3D printer and black filament. See Figs. 6A-6D.

[0037] The present intelligent system for reliably predicting and alerting on the operating conditions, both in real time and predictively, of the vital or operational status of a mining fan, preferably an underground mining fan, or civil works, comprising:

[0038] a) one or more analog or digital sensors with digital recording of their readings / measurements, for each control parameter of the fan, which are located and fixed along the body of the fan, and capture and send information or data related to the vital state or operating state of said fan, to a structured web application database for storage, and to a microprocessor located inside a smart device / board for processing, where said fan control parameters are selected from temperature, vibration and electrical power, and where said sensors are selected from at least one temperature sensor, at least one vibration sensor and at least one electrical power sensor, where said sensors communicate wirelessly or remotely, with said microprocessor of said smart device / board and with said structured database,using the communication channels specific to the mining operation or civil engineering project where the fan is being used,

[0039] b) said appliance-type smart device / board, comprising:

[0040] b.1) said microprocessor that processes said information or data sent by each of said sensors by means of a computer program based on a predictive artificial intelligence model of the vital state or operating state of the ventilator, and where said processed information or data is subsequently sent to and stored in said structured database,

[0041] where said computer program establishes: a) a vital state without failures, when the value of said information or processed data has a value less than the minimum threshold of a range of pre-established values ​​for each control parameter, in said database structured as a failure risk range; b) a failure alert state, when the value of said information or processed data is greater than the lower value / threshold of a range of pre-established values ​​in the database structured as a failure risk range and is less than the higher value / threshold of said range of pre-established values ​​for each control parameter, in said database structured as a failure risk range;(yc) a critical failure alert state, when the value of the processed information or data is equal to or greater than the highest value / upper threshold of the range of pre-established values ​​for each control parameter, in said database structured as a failure risk range;

[0042] The computer program is continuously calibrated according to the information or data sent by each of the sensors; the information or data processed by the microprocessor for each control parameter; and the information or data collected related to the vital state of the ventilator.

[0043] where said computer program allows the capture of said information and data from said sensors, the training or calibration of the predictive model base of the computer program, the generation of predictions for the vital state of the ventilator, the creation of a captive web portal for the user, as well as the generation of codes, and program instructions that allow the flow of information between said sensors and said microprocessor, said microprocessor and said structured database, said sensors and said structured database, said microprocessor and visual alarms, among others,

[0044] b.2) a visual alarm that alerts in real time, of the predicted vital state for the fan and comprises: a green light that illuminates for the alert of a no-fault state; a yellow light that illuminates for the alert of a risk of failure; and a red light that illuminates for the alert of a critical failure;

[0045] b.3) a display screen for the processed information or data, preferably through graphics, for the user, allowing them to see and interact with said microprocessor, and observe the predictions, the current vital signs status of the ventilator, and important data to monitor and keep the ventilator under surveillance;

[0046] b.4) a Raspberry Pi 4 unit that functions as an Access Point and Captive Portal, allowing connection via WIFI to an associated web platform, which can be accessed after entering their username and password, to request predictions of the vital status of the fan, b.5) an Arduino (programmable logic controller, PLC) and its corresponding charger, which generates the light signal for said visual alarm, and optionally, generates the sound signal for the audible alarm, and also sends the predictions to said Raspberry Pi 4 for its display on the visual screen, and optionally, functions as an actuator to connect with other equipment or devices of the work, including activating auxiliary lights in the mining tunnel or civil works, alerting beyond the control room of the vital status of the fan,

[0047] b.6) a router that creates a permanent communication channel between the Raspberry Pi and the microprocessor, forming its own artificial network that maintains continuous data communication if the mining network where the device / smart board is located fails or the internet supply is interrupted, as well as between the microprocessor, Raspberry Pi, and the user; b.7) a UPS unit that provides backup power and protects the equipment from damage due to changes in voltage and current experienced in the mine, ensuring a constant voltage power supply to all other components of the device / smart board.

[0048] b.8) a fan and slotted channels that maintain the temperature inside the device, safeguarding its operation, a speaker, a protective metal casing, power supply terminals, power switches, among other conventional electronic components, including a computer board having an operating system that allows managing each of the conventional electronic components of the same, including peripherals and hardware resources, and

[0049] c) said structured database that manages the range of pre-established values, varying it as the information received continuously from said sensors increases, that is, information or data related to the performance of said fan, and the information received from said microprocessor, that is, processed information and data, and also adding new information or data collected from the performance of the fan.

[0050] Preferably, one or more temperature sensors are located in the center of the fan; one or more vibration sensors are located at the base of the fan; and one or more electrical power sensors are located at the top of the fan body. Preferably, these sensors are attached to the fan using fastening means selected from one or more magnetic fasteners, industrial adhesives, clamps, screw and bolt fastening, among others. Optionally, these sensors may be protected from the environment in which they will operate and may have protective housings selected from stainless steel, plastic, epoxy resin, among other materials.

[0051] Preferably, the smart device / panel includes, in addition to a visual alarm, an audible alarm, including a beacon or siren, activated by the PLC, which reinforces the critical alert. Optionally, in addition to the audible alarm, it can generate a text message, which is displayed on the smart device / panel screen.

[0052] Example

[0053] The structured database contained collected fan performance data, approximately 10GB of information comprised of measurements from temperature, vibration, and voltage sensors.

[0054] Laboratory tests were performed using operational variable data from 3 axial fans located in an underground mine, where these fans have the characteristics described in Table 1 below:

[0055] Table 1

[0056] Power Flow Rate Pressure Fan Model Series

[0057] (HP) (KCFM) (InWg) AXIVANE 66-30

[0058] TVI02 IDV1 300 200 7” S2000 FB

[0059] Howden Joy 66- TVI01 370004 O-04 200 150 4.5”

[0060] 30 / 2000 FB

[0061] INAV 7200-0416- TVI03 J12-304 125 120 4”

[0062] 1000

[0063]

[0064] This information, obtained from three fans, generates a data set for each fan, which is stored in a structured database in the cloud, and where each of the three data sets comprises the variables shown in Table 2 below:

[0065] Table 2

[0066] Variable to be measured Unit

[0067] Bearing Temperature °C

[0068] Winding Temperature °C

[0069] Vibrations mm / s

[0070] Frequencies Hz

[0071] Electrical Power kW

[0072] Electric Current A

[0073]

[0074] The information or data captured / detected by the sensors for each of these variables is processed in the artificial intelligence predictive model programmed into the microprocessor of the smart device / board. This model, designed to analyze fan performance, learns the relationships, patterns, and mathematical behavior of each variable in order to predict future behavior. Programming is preferably done in Python in conjunction with MATLAB, where the code containing all the statements and instructions necessary for the predictive model to operate is designed.

[0075] This predictive artificial intelligence model can be designed using libraries such as Tensorflow, Pytorch, scikit-learn and using pandas and numpy, among others, for data cleaning, employing techniques such as statistical analysis, identification and replacement of outliers or atypical values ​​through machine learning tools such as Isolation Forest and KNN and analysis of variables by cumulative histograms and graphical visualization.

[0076] The construction of the predictive model is based on a univariate model, which learns and processes only one variable to predict the behavior of that variable, and multivariate models, in which multiple variables are entered simultaneously to predict all variables at the same time.In particular, the model is based on univariate LSTM (Long Short Term Memory) and multivariate LSTM models, which offer advantages such as working with neurons that are capable of learning the patterns and behavior of the variable. It is a densely connected recurrent neural network, which has a higher level of customization and adjustment since it has many hyperparameters that are adjusted according to the characteristics of each fan. These hyperparameters are specific values ​​that the programmer and designer of the neural network must adjust so that this network has the greatest accuracy in its predictions. In the case of the LSTM model, these hyperparameters are shown in Table 3.

[0077] Main Hyperparameters of the Model i LSTM Function

[0078] i Number of neurons for long-term memory and Number of neurons per layer i short-term i The LSTM model is composed of Number of hidden layers i multiple hidden layers, where each layer i contains a defined number of neurons. i Corresponds to the number of steps past Size of the predictive window i that the model will need to make the i predictions in the future.

[0079]

[0080] I is the type of mathematical algorithm for the Optimizer Type I

[0081] I search for the descending gradient.

[0082] i Number of neurons per hidden layer that Percentage or level of Dropout I

[0083] And they are deactivated during training

[0084]

[0085] i is the number of observations from the training set i that will make up each Batch Size or batch size i of the multiple batches through which the weights and biases of each neuronal junction will be updated.

[0086] It is the number of times the training set will go through the model until the training process is complete.

[0087]

[0088] Figures 1 and 2 illustrate the predictions with an hourly frequency, considering one week (7 days) as the predictive horizon.

[0089] It is also possible to enhance this LSTM model by making it multivariable, meaning the neural network learns multiple variables simultaneously. Its code is built in Matlab, and it offers the following predictions, using a lag of 45 hours. The lag is the number of steps required to predict the next future step. See Fig. 3. Optionally, the predictive model can use Transformers, which are deep learning tools used for natural language processing, such as those used in search engines like ChatGPT or Bing, thus increasing the model's reliability.

[0090] The key element in these LSTM models is the hyperparameters. These are specific values ​​that the designer and creator of the neural network must set, corresponding to the values ​​and algorithms that perform the neural network learning process. The best selection of hyperparameters ensures the model's reliable predictions. A standard application model is generated, which can be expanded to various models through retraining.

Claims

CLAIMS 1. An intelligent system for reliably predicting and alerting on the operating conditions, both in real time and predictively, the vital state or operating status of a mining fan, preferably an underground mining fan, or civil works, comprising: a) one or more analog or digital sensors with digital recording of their readings / measurements, for each control parameter of the fan, which are located and fixed along the body of the fan, and capture and send information or data related to the vital state or operating state of said fan, to a database for web applications, structured to be stored, and to a microprocessor located inside a smart device / board to be processed, where said fan control parameters are selected from temperature, vibration and electrical power, and where said sensors are selected from at least one temperature sensor, at least one vibration sensor and at least one electrical power sensor, where said sensors communicate wirelessly or remotely, with said microprocessor of said smart device / board and with said structured database, using the communication channels of the mining operation or civil works operation where said fan is being used, b) said appliance-type smart device / board, comprising: b.1) said microprocessor that processes said information or data sent by each of said sensors by means of a computer program based on a predictive artificial intelligence model of the vital state or operating state of the ventilator, and where said processed information or data is subsequently sent to and stored in said structured database, where said computer program establishes: a) a vital state without failures, when the value of said information or processed data has a value less than the minimum threshold of a range of pre-established values ​​for each control parameter, in said database structured as a failure risk range; b) a failure alert state, when the value of said information or processed data is greater than the lower value / threshold of a range of pre-established values ​​in the database structured as a failure risk range and is less than the higher value / threshold of said range of pre-established values ​​for each control parameter, in said database structured as a failure risk range;(c) a critical failure alert state, when the value of the processed information or data is equal to or greater than the highest value / upper threshold of the pre-established range of values ​​for each control parameter, in said database structured as a failure risk range, said computer program is continuously calibrated as stored in the structured database, said information or data sent by each of said sensors; said information or data processed, for each control parameter, by the microprocessor; and said information or data collected related to the vital state of the ventilator; where said computer program allows the capture of said information and data from said sensors, the training or calibration of the predictive model base of the computer program, the generation of predictions for the vital state of the ventilator, the creation of a captive web portal for the user, as well as the generation of codes, and program instructions that allow the flow of information between said sensors and said microprocessor, said microprocessor and said structured database, said sensors and said structured database, said microprocessor and visual alarms, among others, b.2) a visual alarm that alerts in real time, of the predicted vital state for the fan and comprises: a green light that illuminates for the alert of a no-fault state; a yellow light that illuminates for the alert of a risk of failure; and a red light that illuminates for the alert of a critical failure; b.3) a display screen for the processed information or data, preferably through graphics, for the user, allowing them to see and interact with said microprocessor, and observe the predictions, the current vital signs status of the ventilator, and important data to monitor and keep the ventilator under surveillance; b.4) a Raspberry Pi 4 unit that functions as an Access Point and Captive Portal, allowing connection via WIFI to an associated web application, which can be accessed after entering its username and password, to request predictions of the vital status of the fan, b.5) an Arduino (programmable logic controller, PLC) and its corresponding charger, which generates the light signal for said visual alarm, and optionally, generates the sound signal for the audible alarm, and also sends the predictions to said Raspberry Pi 4 for its display on the visual screen, and optionally, functions as an actuator to connect with other equipment or devices of the work, including activating auxiliary lights in the mining tunnel or civil works, alerting beyond the control room of the vital status of the fan, b.6) a router that creates a permanent communication channel between the Raspberry Pi and the microprocessor, forming its own artificial network that maintains continuous data communication if the mining network where the device / smart board is located fails or the internet supply is interrupted, as well as between the microprocessor, Raspberry Pi, and the user; b.7) a UPS unit that provides backup power and protects the equipment from damage due to voltage and current fluctuations experienced in the mine, ensuring a constant voltage power supply to all other components of the device / smart board; and b.8) a fan and slotted channels that maintain the temperature inside the device, safeguarding its operation, a speaker, a protective metal casing, power supply terminals, power switches, among other conventional electronic components, including a computer board having an operating system that allows managing each of the conventional electronic components of the same, including peripherals and hardware resources, and c) said structured web application database manages the range of pre-established values, varying it as the information received continuously from said sensors increases, that is, information or data related to the performance of said fan, and the information received from said microprocessor, that is, processed information and data, and also adding new information or data collected from the performance of the fan, 2. The intelligent system of claim 1 wherein said one or more temperature sensors are located in the center of the fan.

3. The intelligent system of claim 1 wherein said one or more vibration sensors are located at the base of the fan.

4. The intelligent system of claim 1 wherein said one or more electrical power sensors are located on the top of the fan body.

5. The intelligent system of claim 1 wherein said one or more temperature sensors are located in the center of the fan, said one or more vibration sensors are located at the base of the fan, and said one or more electrical power sensors are located at the top of the fan body.

6. The intelligent system of claim 1 wherein said sensors are attached to the fan by means of fastening selected from one or more of magnetic fasteners, industrial adhesives, clamps, screw and bolt fastening.

7. The intelligent system of claim 1 wherein said sensors have a protective housing selected from a housing of stainless steel, plastic, epoxy resin.

8. The intelligent system of claim 1 further comprising an audible alarm, including a beacon, to reinforce the critical alert.

9. The intelligent system of claim 1 further comprising a visual alarm corresponding to a telematic text message, which is displayed on the screen of the smart device / tablet

Images