Intelligent diagnosis and early warning system and method for ultrasonic atomization dust removal

Abstract

The invention discloses an ultrasonic atomization dedusting intelligent diagnosis early warning system and method, and relates to the technical field of ultrasonic atomization intelligent monitoring and control. The system comprises a basic dust removal unit (an ultrasonic atomizer, a fan and a water tank) and an intelligent diagnosis early warning unit (a sensing layer, a data transmission layer, a control diagnosis layer, an execution layer and a data storage layer). The sensing layer is provided with multiple sensors to collect multi-dimensional parameters, data are transmitted to the control diagnosis layer through the industrial Ethernet, the edge calculation unit recognizes faults through an AI fault recognition model and calculates the overall risk degree in combination with the formula of fault degree * weight coefficient, the early warning threshold judgment module sets three levels of threshold values, the execution layer correspondingly performs linkage actions, and the control diagnosis layer performs control diagnosis. And two working condition threshold values and self-adaptive parameter regulation and control are supported. According to the invention, accurate fault diagnosis, graded early warning and on-demand dust removal are realized, energy consumption and fault expansion risks are reduced, and the method is suitable for industrial dust removal scenes.

Description

Technical Field

[0001] This invention relates to the field of intelligent monitoring and control technology for ultrasonic atomization, and in particular to an intelligent diagnostic and early warning system and method for ultrasonic atomization dust removal. Background Technology

[0002] The core components of existing ultrasonic atomization dust removal systems are ultrasonic atomizers, fans, and water tanks. Their operation has the following significant limitations:

[0003] 1. Fixed parameters: Relying on preset fixed operating parameters (such as atomization frequency and fan speed), it is impossible to dynamically adjust according to the real-time dust concentration in the dust removal area, which can easily lead to problems such as "excessive energy consumption" or "incomplete dust removal".

[0004] 2. Lack of monitoring: There is a lack of real-time monitoring capabilities for critical equipment status. It is impossible to keep track of core parameters such as ultrasonic atomizer current, water tank level, fan speed and vibration in real time, making it difficult to detect potential equipment failures in advance.

[0005] 3. Lack of early warning and control: A few systems equipped with dust concentration sensors can only display data. They do not have a hierarchical early warning mechanism or parameter linkage control logic. They cannot respond automatically when the dust concentration exceeds the standard or the equipment is abnormal. They rely entirely on manual intervention, which can easily lead to dust removal interruption, increased dust pollution, or expanded equipment failure.

[0006] In summary, existing systems, lacking dynamic adjustment, status monitoring, and intelligent early warning and control, are unable to meet the high-efficiency, stable, and low-maintenance dust removal requirements in industrial scenarios. Therefore, how to construct an intelligent ultrasonic atomization system with sensing, diagnosis, early warning, and adaptive control capabilities has become an urgent technical problem to be solved. Summary of the Invention

[0007] To solve the above-mentioned technical problems, the present invention is achieved through the following technical solution:

[0008] This invention provides an intelligent diagnostic and early warning system for ultrasonic atomization dust removal, including a basic dust removal unit and an intelligent diagnostic and early warning unit. The basic dust removal unit includes an ultrasonic atomizer, a fan, and a water tank.

[0009] The intelligent diagnostic and early warning unit includes a perception layer, a data transmission layer, a control and diagnostic layer, an execution layer, and a data storage layer for storing historical operating data and fault records. The data storage layer is a local server or a cloud database.

[0010] The data transmission layer connects to the sensing layer and the control and diagnostic layer. The sensing layer deploys current sensors, liquid level sensors, dust sensors, speed sensors, temperature sensors, and vibration sensors.

[0011] Among them, the current sensor is electrically connected to the ultrasonic atomizer, the liquid level sensor is embedded in the water tank, the dust sensor is deployed in the dust removal area, the speed sensor is installed on the fan shaft end, the temperature sensor is attached to the equipment housing, and the vibration sensor is fixed to the fan bearing seat.

[0012] The control and diagnostic layer is configured with an edge computing unit and a visual operation and maintenance platform. The edge computing unit includes an AI fault recognition model and an early warning threshold judgment module.

[0013] The edge computing unit independently acquires the single-parameter fault characteristics of any sensor, combines them to construct the correlation between the acquired multiple parameters, and assigns fault weight coefficients to each sensor parameter. Based on the fault degree of each sensor parameter and its corresponding fault coefficient weight, a composite analysis is performed to calculate the overall risk level of the ultrasonic atomization system's fault state. According to the overall risk level of the ultrasonic atomization system's fault state, the execution layer outputs corresponding linkage actions.

[0014] The execution layer includes an ultrasonic atomizer and its backup unit, an audible and visual alarm, a power cut-off module, and an automatic cleaning device for cleaning the ultrasonic atomizer.

[0015] As a preferred technical solution of the intelligent diagnostic and early warning system of the present invention: the AI ​​fault recognition model is trained based on historical fault data, wherein the historical fault data includes characteristic data such as current increase of not less than 20% when the atomizer is blocked and vibration frequency higher than 50Hz when the fan bearing is worn.

[0016] As a preferred technical solution of the intelligent diagnostic and early warning system of the present invention: the sensing layer adopts a low-power sensor, wherein the dust sensor adopts a low-power LoRa type sensor.

[0017] As a preferred technical solution of the intelligent diagnostic and early warning system of the present invention: the early warning threshold judgment module sets at least three risk thresholds according to the overall risk level, namely, early warning threshold, alarm threshold, and emergency threshold. The execution layer takes actions in conjunction with the early warning level: when an early warning occurs, the atomization volume of the ultrasonic nebulizer is increased; when an alarm occurs, the backup unit of the ultrasonic nebulizer is activated; and when an emergency occurs, the fan is stopped and the power cut-off module is triggered.

[0018] As a preferred technical solution of the intelligent diagnostic and early warning system of the present invention: the control and diagnostic layer is configured with an adaptive operation optimization function, and the early warning threshold judgment module is configured with at least two types of early warning thresholds for different operating conditions, including an early warning threshold for industrial production period scenarios and an early warning threshold for industrial shutdown period scenarios. The early warning threshold for industrial production period scenarios is set within a time range of N minutes from the start of industrial production to the end of industrial production, where N ≥ 10.

[0019] The control diagnostic layer acquires historical data that matches the current real-time operating conditions, predicts the dust concentration change within a fixed time t in the future, and linearly regulates the atomization frequency and fan power.

[0020] As a preferred technical solution for the intelligent diagnostic and early warning system of this invention: the data transmission layer adopts an industrial Ethernet module, and the transmission rate of the industrial Ethernet module in the data transmission layer is ≥100Mbps. The AI ​​fault recognition model is a simplified CNN model, and the data processing latency is ≤500ms.

[0021] As a preferred technical solution of the intelligent diagnostic and early warning system of the present invention: let the fault degree of each sensor parameter obtained by the control diagnostic layer be G1, G2, G3, ..., Gn, and the fault weight coefficients assigned by the control diagnostic layer to each sensor parameter be β1, β2, β3, ..., βn, respectively. Then the overall risk degree of the fault state of the ultrasonic atomization system is ξ=G1*β1+G2*β2+G3*β3+...+Gn*βn.

[0022] Wherein, when φx>φ0, the sensor parameter fault degree Gx=φx / φ0, Gx∈[G1, G2, G3, ..., Gn], φx is the parameter value detected by the sensor in real time, and φ0 is the ideal reference value preset by the system; when φx≤φ0, the sensor parameter fault degree Gx=0.

[0023] This invention provides an early warning method for an intelligent diagnostic and early warning system for ultrasonic atomization dust removal, the details of which are as follows:

[0024] Step 1: Data Acquisition: The sensing layer acquires data in real time by using a current sensor to collect the operating status of the ultrasonic atomizer, a liquid level sensor to collect the water volume in the water tank, a dust sensor to collect the dust concentration in the dust removal area, a speed sensor and a vibration sensor to collect the fan status in real time, and a temperature sensor to collect the equipment temperature rise data in real time.

[0025] Step 2, AI Diagnosis: The data collected by the perception layer is transmitted to the edge computing unit of the control and diagnosis layer via the data transmission layer. The data is analyzed by the AI ​​fault identification model to identify the type of equipment fault and determine whether the warning threshold has been triggered.

[0026] Section 3, Early Warning and Control: When the control and diagnostic layer triggers an early warning, it sends a notification via audible and visual alarms, SMS, and industrial platform, and simultaneously links the execution layer to adjust operating parameters or start backup equipment; in an emergency, it controls the equipment to stop and cut off the power supply via the power cut-off module.

[0027] Step 4: Adaptive Optimization: The execution layer dynamically adjusts the frequency of the ultrasonic atomizer and the speed of the fan based on the real-time dust concentration to achieve dust removal on demand.

[0028] Step 5: Data storage: The data storage layer stores the collected real-time data, fault diagnosis results, and early warning records for subsequent fault trend analysis and predictive maintenance.

[0029] Compared with existing technologies, the beneficial effects of this invention are:

[0030] 1. This invention relies on sensors such as current, liquid level, and dust to collect multi-dimensional data. Combined with an AI fault recognition model trained based on historical fault characteristics (such as atomizer blockage and fan bearing wear), it can identify single-parameter faults (such as abnormal current) and multi-parameter correlated faults (such as "increased current + decreased atomization volume + sudden increase in dust concentration" to identify compound faults). This avoids the general alarms of traditional monitoring and significantly improves the identification rate of compound faults. Furthermore, by analyzing and calculating the overall system risk using the formula "fault severity × weight coefficient," the influence weight of each parameter on the fault is clearly defined, making the diagnostic results more quantitatively based rather than relying on subjective judgment.

[0031] 2. This invention sets three levels of thresholds, namely "early warning - alarm - emergency", through an early warning threshold judgment module. The execution layer corresponds to different linkage actions: increasing the atomization volume when an early warning occurs, activating the atomizer backup unit when an alarm occurs, and shutting down and cutting off power when an emergency occurs. This avoids the escalation of the fault (such as increased dust levels or equipment damage) and solves the problem of traditional systems that "only display data and have no automatic response".

[0032] 3. In this invention, the control and diagnostic layer supports two operating condition thresholds: "industrial production period" and "industrial shutdown period" (the production period covers a certain duration after production ends). It combines historical data to predict dust concentration changes within a fixed time period in the future, and linearly adjusts the atomization frequency and fan power to avoid "excessive energy consumption" (no need for high power operation during non-production periods) or "incomplete dust removal" (efficiency is improved as needed during production periods). The execution layer dynamically adjusts equipment parameters according to real-time dust concentration to achieve "increased efficiency when dust concentration is high and reduced frequency when concentration is low". Compared with traditional fixed parameter operation, energy consumption is significantly reduced. Attached Figure Description

[0033] Figure 1 This is a schematic diagram of the logical structure of the ultrasonic atomization dust removal intelligent diagnostic and early warning system of the present invention.

[0034] Figure 2 This is a schematic diagram of the logical structure of the perception layer in this invention.

[0035] Figure 3 This is a schematic diagram of the logical structure of AI fault diagnosis and risk calculation in this invention.

[0036] Figure 4 This is a schematic diagram of the logical structure of the early warning threshold judgment module in this invention.

[0037] Figure 5 This is a schematic diagram of the logical structure of the execution layer in this invention.

[0038] Figure 6 This is a schematic diagram of the logical structure of the control and diagnostic layer and adaptive operation optimization in this invention.

[0039] Figure 7 This is a schematic diagram of the logical structure of the ultrasonic atomization dust removal intelligent diagnostic and early warning system method of the present invention. Detailed Implementation

[0040] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0041] Example 1: This invention designs an intelligent diagnostic and early warning system for ultrasonic atomization dust removal, with the following specific configuration:

[0042] (I) Overall System Architecture

[0043] This system adopts a two-layer architecture of "basic dust removal unit + intelligent diagnosis and early warning unit". The intelligent diagnosis and early warning unit follows a full-link design of "sensing-transmission-diagnosis-execution-storage", and the specific architecture is as follows: Figure 1 As shown.

[0044] Unit classification Includes modules / components Core Functions Basic dust removal unit Ultrasonic atomizer, fan, water tank Basic dust removal function: Water is supplied from the water tank to the ultrasonic atomizer, which generates atomized particles to suppress dust, and a fan assists in airflow circulation. Intelligent diagnostic and early warning unit Perception layer, data transmission layer, control and diagnostic layer, execution layer, data storage layer Achieve closed-loop intelligent management of "data acquisition - fault diagnosis - hierarchical early warning - adaptive control - data storage"

[0045] (II) Detailed technical parameters and functions of each core unit

[0046] 1. Basic dust removal unit

[0047] Ultrasonic atomizer: The core dust removal component, which atomizes water into micron-sized particles through high-frequency vibration to adsorb dust in the air; equipped with a backup unit for seamless switching in case of failure.

[0048] Fan: Provides airflow power to accelerate the diffusion of atomized particles and dust adsorption in the dust removal area; real-time monitoring of status parameters such as speed, vibration, and temperature is required.

[0049] Water tank: Provides a stable water source for the ultrasonic nebulizer. The liquid level needs to be monitored in real time to avoid water shortage causing the nebulizer to run dry and be damaged.

[0050] 2. Intelligent Diagnostic and Early Warning Unit

[0051] 2.1. Perception Layer: Multi-dimensional Data Acquisition Terminal

[0052] The perception layer acts as the system's "sensors," deploying six types of low-power sensors to comprehensively collect device status and environmental parameters. Specific configurations are as follows: Figure 2 As shown.

[0053] Sensor type Deployment location / connection object Monitoring parameters Technical characteristics core role Current sensor Electrical connection to ultrasonic atomizer Atomizer operating current High precision, low power consumption Identify atomizer blockage (current increase ≥20%) Liquid level sensor Embedded inside the water tank Water tank level / remaining water volume Corrosion resistance and continuous monitoring To prevent the atomizer from running dry, avoid it from running dry. Dust sensor Key locations in the dust removal area (such as next to the production line). Area dust concentration (mg / m³) Low-power LoRa type, long transmission distance Assess dust removal effectiveness and trigger adaptive adjustment. Speed ​​sensor Installed on the fan shaft end real-time speed of the fan Non-contact measurement, accuracy ±1 r / min Identify abnormal fan speeds (such as stalling or slowing down). Temperature sensor Attach to the device housing (atomizer / fan) Equipment casing temperature rise data Temperature measurement range: -20℃ to 120℃; accuracy: ±0.5℃ Monitoring equipment overheating fault Vibration sensor Fixed to the fan bearing housing Vibration frequency of fan bearing (Hz) High frequency response, range 0–500Hz Identifying wear on fan bearings (vibration frequency > 50Hz)

[0054] Key design features: All sensors employ low-power solutions, with the dust sensor utilizing LoRa (long-range radio) technology to reduce energy consumption and meet the long-term continuous operation requirements of industrial scenarios.

[0055] 2.2 Data Transmission Layer: High-Speed ​​Data Link

[0056] Hardware selection: Industrial Ethernet modules are used, which are suitable for complex industrial electromagnetic environments and have strong anti-interference capabilities.

[0057] Key parameters: Transmission rate ≥100Mbps, ensuring real-time transmission of multi-sensor data (such as vibration and high-frequency current data) to support the low-latency requirements of subsequent AI diagnostics.

[0058] Connection logic: One end connects to all sensors in the perception layer, and the other end connects to the edge computing unit of the control and diagnosis layer, realizing seamless data flow of "acquisition-transmission-diagnosis".

[0059] 2.3 Control and Diagnostic Layer: The System's "Brain," the Core Decision-Making End

[0060] The control and diagnostic layer is the core of intelligent diagnosis and control, comprising edge computing units and a visual operation and maintenance platform. Specific functions and technical details are as follows: Figure 3 , Figure 6 As shown.

[0061] (1) Edge computing unit: core algorithm carrier - AI fault identification model

[0062] a. Model Foundation - Based on training with historical fault data, the core characteristics of historical faults include:

[0063] Atomizer blockage: Current increase of at least 20%.

[0064] Fan bearing wear: vibration frequency higher than 50Hz.

[0065] Complex faults (such as "atomizer blockage + atomizing plate aging"): increased current + decreased atomization volume + sudden increase in dust concentration.

[0066] b. Model Architecture - A simplified version of CNN (Convolutional Neural Network) is used, with some scenarios optimized to "CNN + multi-head attention mechanism":

[0067] CNN is responsible for extracting single-parameter fault features (such as abnormal current or excessive vibration).

[0068] The attention mechanism focuses on capturing the correlation between multiple parameters (such as the linkage feature of "increased current + increased dust concentration"), which improves the recognition rate of complex faults to 98%.

[0069] c. Performance indicators: Data processing latency ≤500ms, meeting the needs of real-time industrial diagnostics.

[0070] d. Calculation logic for overall risk level:

[0071] The system uses a weighted formula of "failure severity × weighting coefficient" to quantitatively assess the overall failure risk of the equipment, avoiding subjective judgment.

[0072] Fault severity of each sensor parameter: G1, G2, ..., Gn (n is the number of sensors);

[0073] Fault coefficient weights for each parameter: β1, β2, ..., βn (key parameters such as dust concentration and atomizer current have higher weights);

[0074] Sensor real-time parameter value: φx, system preset ideal reference value: φ0.

[0075] Fault severity calculation: When φx > φ0 (parameter exceeds the limit), Gx = φx / φ0 (the more severe the exceedance, the larger Gx); when φx ≤ φ0 (parameter is normal): Gx = 0. Where Gx ∈ [G1, G2, G3, ..., Gn].

[0076] The formula for the overall risk level is: ξ = G1×β1 + G2×β2 + ... + Gn×βn.

[0077] e. Warning threshold judgment module:

[0078] The module sets three risk thresholds based on the overall risk level ξ and adapts to two different operating condition thresholds to achieve differentiated early warning and control, such as... Figure 4 As shown.

[0079] Level 3 risk threshold and corresponding actions:

[0080] Warning Level Triggering conditions Execution layer linkage action Warning ξ < warning threshold Increase the atomization volume of the ultrasonic atomizer to enhance the dust removal effect and prevent the dust concentration from rising further. Alarm Warning threshold ≤ ξ < alarm threshold The backup unit of the ultrasonic nebulizer is activated, and the audible and visual alarm is triggered, sending a notification to maintenance personnel. urgent ξ≥alarm threshold The system controls the fan to stop, triggering the power cut-off module to disconnect the main power supply to the equipment, preventing the fault from escalating (such as equipment burnout or dust explosion risks).

[0081] f. Thresholds for two operating conditions:

[0082] Industrial production period threshold: covering "the start time of industrial production to N minutes after the end of production" (N≥10). During this period, the amount of dust generated is large, so the threshold setting is more stringent.

[0083] Industrial downtime threshold: After N minutes of production completion, dust generation drops sharply, so the threshold can be appropriately relaxed to avoid excessive energy consumption.

[0084] (2) Visualized Operation and Maintenance Platform

[0085] Deployment method: Supports local touch screen or remote PC access.

[0086] Functions: Real-time display of sensor data, equipment operating status, fault diagnosis results, and early warning levels; supports historical data query and fault trend analysis, providing visual support for predictive maintenance.

[0087] 2.4. Execution Layer:

[0088] The execution layer is the system's "hands and feet," executing specific actions based on instructions from the control and diagnostic layer. Component configurations and functions are as follows: Figure 5 As shown.

[0089] Execution Component Core Functions Related warning levels Ultrasonic nebulizer (including spare unit) Basic atomization dust suppression; the backup unit activates upon alarm to ensure uninterrupted dust removal. All levels (enhance efficiency during early warning, switch to backup during alarm). Audible and visual alarm Emitting audible and visual signals to alert on-site maintenance personnel of equipment malfunctions. Alarms, Emergency Level Power cut-off module In an emergency, the main power supply to the equipment is cut off, forcing a shutdown. Emergency Level Automatic cleaning device Clean the ultrasonic atomizer nozzle regularly or after a malfunction to prevent / alleviate atomizer clogging. Warning level (activated when there are early signs of congestion) Fan Receive control commands to adjust the rotation speed, and work with the atomizer to optimize airflow. Full range (adaptive speed adjustment)

[0090] 2.5 Data Storage Layer: Data Support End

[0091] Storage medium: Supports local servers or cloud databases, which can be selected according to the data security requirements of industrial scenarios.

[0092] Storage content:

[0093] Real-time data acquisition: Real-time parameters of each sensor (current, liquid level, dust concentration, etc.). Fault-related data: Fault type, trigger time, fault severity, and handling result.

[0094] Operation logs include: device start / stop records, parameter adjustment records, and warning records.

[0095] Core function: To provide data support for the iterative training of AI fault identification models, fault trend analysis, and predictive maintenance (such as predicting the life of wind turbine bearings based on historical data).

[0096] Example 2: This invention designs an early warning method for an intelligent diagnostic and early warning system for ultrasonic atomization dust removal. It achieves a closed loop of "collection-diagnosis-early warning-optimization-storage" through five core steps, as detailed below. Figure 7 As shown:

[0097] Step 1: Data Acquisition (Core Action of the Perception Layer). The six types of sensors in the perception layer simultaneously acquire data:

[0098] Current sensor: Real-time acquisition of the ultrasonic atomizer's operating current.

[0099] Liquid level sensor: Collects the remaining water level in the water tank.

[0100] Dust sensor: Collects real-time dust concentration in the dust removal area.

[0101] Speed ​​sensor and vibration sensor: collect fan speed and bearing vibration frequency.

[0102] Temperature sensor: Collects temperature rise data of atomizer / fan housing.

[0103] Step Two: AI Diagnosis (Controlling the core actions of the diagnostic layer)

[0104] 1. Data transmission: The raw data collected by the sensing layer is transmitted to the edge computing unit of the control and diagnostic layer via the industrial Ethernet module (data transmission layer).

[0105] 2. Model Analysis: The AI ​​fault identification model processes the data, extracts single-parameter / multi-parameter fault features, and identifies fault types (such as atomizer blockage and fan bearing wear).

[0106] 3. Threshold judgment: Based on the overall risk level ξ, determine whether the three-level threshold of "early warning-alarm-emergency" is triggered.

[0107] Step 3: Early Warning and Control (Linkage between Control Diagnosis Layer and Execution Layer)

[0108] If an alert is triggered: the execution layer increases the atomization volume of the ultrasonic atomizer, and the visualization platform displays the alert information.

[0109] If an alarm is triggered: the audible and visual alarm is activated, and an alarm notification is pushed to the operation and maintenance personnel via SMS or industrial platform (such as MES system). The execution layer then activates the backup unit of the ultrasonic atomizer.

[0110] If an emergency is triggered: the execution layer controls the fan to stop, triggering the power cut-off module to cut off power and prevent the fault from escalating.

[0111] No warning: The execution layer maintains the current running parameters and enters the adaptive optimization phase.

[0112] Step 4: Adaptive Optimization (Linkage between Control and Diagnostic Layers and Execution Layers)

[0113] 1. Operating Condition Identification: The control and diagnostic layer determines whether the current period is "industrial production period" or "downtime period" and matches the corresponding operating condition threshold.

[0114] 2. Data support: Retrieves historical data from the data storage layer that matches the current operating conditions.

[0115] 3. Predictive analysis: Based on historical data, predict the trend of dust concentration changes within a fixed time t (e.g., 10 min) in the future.

[0116] 4. Parameter control: Linear adjustment of ultrasonic atomizer frequency and fan power (increase efficiency when dust concentration is high, and reduce frequency when concentration is low) to achieve "dust removal on demand" and reduce energy consumption.

[0117] Step 5: Data Storage (Core Operation of the Data Storage Layer). The data storage layer persistently stores the following data: real-time data collected by the perception layer, AI diagnostic results (fault type, risk level), early warning records (early warning level, trigger time, and handling actions), and adaptive optimization parameters (adjusted atomization frequency and fan speed). Stored data is used for subsequent model training, fault review, and predictive maintenance.

[0118] In summary, the core technological advantages of the ultrasonic atomization dust removal intelligent diagnostic and early warning system of this invention are as follows:

[0119] Precise diagnosis: Through multi-sensor + AI model, accurate identification of single-parameter faults and compound faults is achieved, and the diagnostic results are quantified (based on ξ value) to avoid general alarms.

[0120] Early warning tiered system: Three threshold levels correspond to differentiated actions, solving the problem of "only displaying but no response" in traditional systems and reducing the risk of escalating faults.

[0121] Adaptive operation: With thresholds for different operating conditions and dynamic parameter adjustment, energy consumption is reduced by 20% to 30% compared to traditional fixed parameter operation, while avoiding incomplete dust removal.

[0122] Predictive maintenance: Based on historical data and failure trend analysis, it enables "failure prediction and early maintenance" to reduce equipment downtime.

[0123] Technology maturity: The core components (low-power sensors, industrial Ethernet, simplified CNN model) are all mature industrial-grade products with no technological bottlenecks.

[0124] Compatibility: It can be directly integrated into existing ultrasonic atomization dust removal systems (without replacing the main equipment), and is suitable for multiple scenarios such as mining, building materials, and machining, with a compatibility rate of ≥90%.

[0125] Supply chain support: Sensors can be sourced from Shenzhen Topray, and AI edge computing modules can be obtained from Huawei Atlas200I, ensuring a stable supply chain.

[0126] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An ultrasonic atomization dust removal intelligent diagnosis and early warning system, characterized in that: it comprises a basic dust removal unit and an intelligent diagnosis and early warning unit; the basic dust removal unit comprises an ultrasonic atomizer, a fan and a water tank; the intelligent diagnosis and early warning unit comprises a perception layer, a data transmission layer, a control diagnosis layer, an execution layer and a data storage layer for storing historical operation data and fault records; the data transmission layer is connected to the perception layer and the control diagnosis layer; the perception layer is provided with current sensors, liquid level sensors, dust sensors, rotating speed sensors, temperature sensors and vibration sensors; the current sensors are electrically connected to the ultrasonic atomizer, the liquid level sensors are embedded in the water tank, the dust sensors are arranged in a dust removal area, the rotating speed sensors are installed on the shaft end of the fan, the temperature sensors are attached to the equipment shell, and the vibration sensors are fixed to the fan bearing seat; the control diagnosis layer is provided with an edge computing unit and a visual operation and maintenance platform, and the edge computing unit comprises an AI fault identification model and a warning threshold judgment module; the edge computing unit independently acquires single-parameter fault features of any sensor, combines and constructs the correlation between the acquired multiple parameters, and assigns fault weight coefficients to each sensor parameter; according to the fault degree of each sensor parameter and the corresponding fault coefficient weight, the overall risk degree of the ultrasonic atomization system fault state is analyzed and calculated; according to the overall risk degree of the ultrasonic atomization system fault state, the execution layer outputs corresponding degree of linkage action; the execution layer comprises the ultrasonic atomizer and its backup unit, an audible and visual alarm, a power cut-off module and an automatic cleaning device for cleaning the ultrasonic atomizer.

2. The ultrasonic atomization dust removal intelligent diagnosis and early warning system according to claim 1, characterized in that: the AI fault identification model is trained based on historical fault data, wherein the historical fault data includes feature data such as an increase of current by no less than 20% when the atomizer is blocked and a vibration frequency higher than 50Hz when the fan bearing is worn.

3. The ultrasonic atomization dust removal intelligent diagnosis and early warning system according to claim 1, characterized in that: the perception layer adopts low-power sensors, and the dust sensors adopt low-power LoRa type sensors.

4. The ultrasonic atomization dust removal intelligent diagnosis and early warning system according to claim 1, characterized in that: the warning threshold judgment module sets at least three risk threshold values according to the overall risk degree, which are respectively a warning threshold value, an alarm threshold value and an emergency threshold value; the execution layer outputs linkage actions according to the warning levels, increases the atomization amount of the ultrasonic atomizer when warning, starts the backup unit of the ultrasonic atomizer when alarming, and controls the fan to stop and triggers the power cut-off module in an emergency.

5. The ultrasonic atomization dust removal intelligent diagnosis and early warning system according to claim 1, characterized in that: the control diagnosis layer is provided with an adaptive operation optimization function, and the warning threshold judgment module is provided with at least two kinds of working condition warning threshold values, including an industrial production period scene warning threshold value and an industrial shutdown period scene warning threshold value. ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ The industrial production period scene early warning threshold setting time range is from the industrial production start time point to Nmin after the industrial production end, wherein N≥10. The control diagnosis layer obtains historical data matched with the current real-time working condition, predicts the dust concentration change in the future fixed time t, and linearly regulates and controls the atomization frequency and the fan power.

6. The intelligent diagnosis and early warning system for ultrasonic atomization dust removal according to claim 1, characterized in that: The data transmission layer adopts an industrial Ethernet module, and the industrial Ethernet module of the data transmission layer has a transmission rate of ≥100 Mbps. The AI fault identification model is a simplified CNN model, and the data processing delay is ≤500 ms.

7. The intelligent diagnosis and early warning system for ultrasonic atomization dust removal according to claim 1, characterized in that: The control diagnosis layer obtains the fault degrees of the sensor parameters G1, G2, G3,..., Gn, respectively, and the control diagnosis layer assigns the fault weight coefficients β1, β2, β3,..., βn to the sensor parameters, so that the total risk degree ξ of the ultrasonic atomization system fault state is G1*β1+G2*β2+G3*β3+...+Gn*βn. When φx>φ0, the sensor parameter fault degree Gx=φx / φ0, Gx∈[G1, G2, G3,..., Gn], φx is the parameter value detected by the sensor in real time, and φ0 is the ideal reference value preset by the system. When φx≤φ0, the sensor parameter fault degree Gx=0.

8. The ultrasonic atomization dust removal intelligent diagnosis and early warning system according to claim 1, characterized in that, The intelligent diagnosis and early warning method comprises the following steps: Link one, data acquisition: the perception layer performs data acquisition, the working condition of the ultrasonic atomizer is acquired in real time through a current sensor, the water quantity in the water tank is acquired through a liquid level sensor, the concentration in the dust removal area is acquired in real time through a dust sensor, the fan state is acquired in real time through a speed sensor and a vibration sensor, and the temperature rise data of the equipment are acquired in real time through a temperature sensor; Link two, AI diagnosis: the data collected by the perception layer is transmitted to the edge computing unit of the control diagnosis layer through the data transmission layer, the data is analyzed through the AI fault identification model, the equipment fault type is identified, and it is judged whether the early warning threshold is triggered; Link three, early warning control: when the control diagnosis layer triggers the early warning, the sound and light alarm, the short message and the industrial platform push notification are used, and the running parameters are adjusted or the standby equipment is started through the execution layer linkage; In an emergency, the power is cut off through the power cut-off module to stop the equipment and cut off the power supply; Link four, adaptive optimization: the execution layer dynamically adjusts the ultrasonic atomizer frequency and the fan speed according to the real-time dust concentration, so as to realize on-demand dust removal; Link five, data storage: the real-time data collected, the fault diagnosis results and the early warning records are stored in the data storage layer for subsequent fault trend analysis and predictive maintenance.

Images