Sensorless intelligent monitoring system and method for flue gas purification device based on coupling of multiple operating parameters

Abstract

The invention discloses a flue gas purification device sensorless intelligent monitoring system and method based on multi-operation parameter coupling, and relates to the field of flue gas purification, the system comprises a multi-layer distributed flue gas purification terminal and a cloud server unit, and the multi-layer distributed flue gas purification terminal and the cloud server unit realize bidirectional data interaction through a network link. The flue gas purification terminal comprises a physical execution unit, a signal conditioning unit and a time sequence data interaction unit. The non-intrusive electrical parameter coupling technology is used for replacing a traditional physical sensor, and the problems of scaling and corrosion of the probe are solved.

Description

Technical Field

[0001] This application relates to the field of flue gas purification, and in particular to a sensorless intelligent monitoring system and method for flue gas purification devices based on the coupling of multiple operating parameters. Background Technology

[0002] In the field of fume purification in the catering industry, existing technologies generally adopt contact sensors and single-dimensional monitoring strategies. Their core monitoring system is built upon discrete physical quantity detection, facing fundamental technical bottlenecks. Taking hot pot restaurants and other high-fume scenarios as an example, traditional monitoring solutions rely on deploying dedicated sensors such as thermocouples and differential pressure transmitters in key parts of the purification device, connecting limited parameters such as temperature and pressure to the PLC control system via wired connections. This technical approach has revealed three systemic defects in engineering practice: First, the sensor body commonly experiences detection failure under extreme conditions such as high temperature (>100℃), high humidity (RH>85%), and multiphase flow (oil-gas-liquid mixture). For example, when a ceramic-like scale layer (thickness >200μm) forms on the surface of the metal probe due to grease condensation, the dynamic response time of the thermocouple increases from the initial 1.2 seconds to 3.5 seconds. Furthermore, the electrodes undergo intergranular corrosion under chloride ion penetration, with an annual corrosion rate of 0.25mm, resulting in the average lifespan of contact sensors being less than one-third of that in conventional industrial environments. Secondly, the contradiction between the limited scope of monitoring and the complexity of equipment status is becoming increasingly prominent. Existing technologies can only acquire local physical quantities such as differential pressure and temperature, failing to capture key characteristic parameters that reflect the health status of equipment, such as harmonic distortion of drive motors and leakage current ripple of electrostatic units. A typical example is that when the thickness of dust accumulation in pipes reaches 80% of the critical value, traditional differential pressure monitoring is still in the 60% alarm range of the threshold, resulting in a warning lag of more than 48 hours. Thirdly, the rigid cabling requirements of the system architecture are severely mismatched with the dynamic deployment requirements of the catering scenario. The insulation performance of the RS485 bus deteriorates rapidly in the high-temperature and humid environment of the kitchen. When deploying 12 monitoring points on a single device, the cable length exceeds 50 meters, and the annual maintenance cost accounts for 35% of the total investment. Furthermore, the electromagnetic crosstalk caused by dense cabling increases the signal error rate to 10%. - ³ on the order of magnitude.

[0003] A deeper technological constraint lies in the fact that existing monitoring systems fail to effectively extract the information value from the operating parameters of the equipment itself. Taking integrated purification cookware as an example, the three-phase current signal of its drive motor contains characteristics of impeller fouling (the 5th harmonic component shows a strong correlation of 0.92 with the fouling mass), and the high-frequency ripple (>10kHz) of the electrostatic unit's operating current can characterize the trend of plate spacing changes. However, these key parameters have long been ignored by traditional monitoring systems. At the same time, the lack of multi-parameter coupling analysis capabilities makes it impossible to establish a dynamic correlation model between electrical parameters (such as power fluctuation coefficient), mechanical state (impeller dynamic balance), and fluid characteristics (pressure loss gradient). This results in maintenance decisions remaining at the stage of extensive management with fixed-cycle filter replacement, causing over 30% of maintenance costs to be over-maintained. Even more serious is that existing cloud monitoring platforms only achieve simple data storage and threshold alarms, lacking dynamic modeling capabilities based on equipment group knowledge bases. They cannot achieve continuous optimization of diagnostic models through federated learning frameworks, resulting in a model adaptation cycle of up to two weeks for newly installed machines. The root cause of these problems lies in the fact that traditional technical approaches rely too heavily on dedicated sensing elements to obtain isolated parameters. They have neither established a mapping relationship between the electrical parameters and mechanical state of the equipment nor constructed a multi-dimensional feature analysis system that can adapt to complex working conditions. Ultimately, this has resulted in the monitoring system and the purification device itself operating in a disconnected state for a long time. Summary of the Invention

[0004] The purpose of this application is to overcome at least one deficiency in the existing technology and provide a sensorless intelligent monitoring system and method for flue gas purification devices based on the coupling of multiple operating parameters. This application replaces traditional physical sensors with non-invasive electrical parameter coupling technology, avoiding probe scaling and corrosion problems.

[0005] To achieve the above objectives, in a first aspect, this application discloses a sensorless intelligent monitoring system for a flue gas purification device based on the coupling of multiple operating parameters. The system includes a flue gas purification terminal and a cloud server, which interact bidirectionally via a network link.

[0006] The flue gas purification terminal includes a physical execution unit with electrical parameter detection, a signal conditioning unit for acquiring electrical parameter detection data from the physical execution unit, and a timing data interaction unit for sending data processed by the information conditioning unit.

[0007] The physical execution unit for achieving flue gas purification includes at least a power drive component, an electrostatic treatment component, a photocatalytic treatment component, and a detection component for detecting the electrical parameters of each component.

[0008] The detection components include a voltage monitoring circuit and a current detection return circuit installed at the power input port of the power drive component, an impedance detection circuit installed at the output port of the electrostatic treatment component, and a shunt sampling module installed on the power supply bus of the photocatalytic treatment component.

[0009] Furthermore, the power drive component is a constant speed motor.

[0010] Furthermore, the voltage monitoring circuit achieves non-contact signal acquisition through parallel voltage sensors; the current monitoring circuit uses a closed-loop sensor based on the Faraday electromagnetic induction principle to form current sampling.

[0011] Furthermore, the impedance detection circuit extracts high-frequency characteristic signals through capacitive coupling technology.

[0012] Furthermore, the DC power supply bus is connected to a shunt sampling module for current and voltage detection.

[0013] Furthermore, the signal conditioning unit includes a distributed signal preprocessing submodule and a synchronous sampling control submodule. The signal preprocessing submodule integrates electrically isolated amplifier circuits and adaptive filter circuits.

[0014] Furthermore, the synchronous sampling control submodule is based on a field-programmable gate array to build a multi-channel parallel sampling architecture and is equipped with an analog-to-digital converter to achieve accurate acquisition of wide dynamic range signals.

[0015] Furthermore, the timing data interaction unit includes a wired communication module and / or a wireless communication module.

[0016] Furthermore, the cloud server unit adopts a distributed computing architecture, which includes parallel processing computing nodes and time-series data storage clusters.

[0017] When this system is in operation, the raw data from the detection components within the flue gas purification terminal undergoes multi-level filtering processing before being uploaded to the cloud server unit via a breakpoint resume and data verification mechanism. The cloud server unit then generates an equipment health status assessment report using feature extraction algorithms and pattern recognition models, which is transmitted back to the field terminal equipment via an encrypted channel.

[0018] Compared with existing technologies, this system achieves operation management and detection based on sensorless technology.

[0019] Secondly, this application discloses an intelligent monitoring method applied to a sensorless intelligent monitoring system for a flue gas purification device based on the coupling of multiple operating parameters. The method includes the following steps:

[0020] Step 1: Synchronously acquire device identification and signal.

[0021] Each monitoring terminal is assigned a globally unique identifier. The voltage and current signals of the drive motor are synchronously collected through an isolated sampling circuit. The raw signals are encapsulated into data packets after analog-to-digital conversion, which include packet header verification sequence, UUID and timestamp information.

[0022] Step 2: Dynamic Feature Parameter Extraction

[0023] Instantaneous power calculations are performed on voltage and current signals, and the active power component is separated using orthogonal decomposition. The spectral distribution characteristics of the electrostatic unit's operating current are analyzed simultaneously, and the proportion of high-frequency harmonic energy is extracted. The dynamic impedance parameters of the drive circuit are calculated, generating a multidimensional feature vector containing power fluctuation coefficients, harmonic distortion rate, and impedance changes.

[0024] Step 3: Local Edge State Assessment

[0025] The feature vector is input into a pre-trained machine learning model to perform a preliminary diagnosis based on the random forest algorithm. When the power parameter is detected to deviate positively from the benchmark value by more than a preset threshold, or the high-frequency harmonic index increases for three consecutive cycles, a local early warning signal is triggered. The evaluation results and the original data are cached together in the local storage area.

[0026] Step 4: In-depth cloud-based analysis and diagnosis

[0027] Feature data and equipment operation logs are uploaded to the cloud server via an encrypted channel. Time-series pattern analysis based on long short-term memory networks is performed, and the physical rationality of the feature parameters is verified by combining a fluid dynamics simulation model. The predicted value of pipe ash accumulation thickness and the remaining effective operating time are calculated, and a result report including maintenance priority score is generated.

[0028] Step 5: Control Strategy Optimization and Feedback

[0029] Based on cloud-based diagnostic results, the equipment operating parameters are dynamically adjusted: power compensation control is implemented for the drive motor to maintain system stability, and the operating voltage of the electrostatic unit is optimized to suppress abnormal discharge. The decision threshold parameters of the local evaluation model are updated to form an adaptive control closed loop.

[0030] Step 6: Model Iteration and Data Source Tracing

[0031] Establish a device group knowledge base, aggregate feature data from various terminals using a federated learning framework, and optimize the weight parameters of the cloud-based analysis model. Initiate end-to-end data tracing for anomaly detection events to verify the data integrity of each stage of signal acquisition, feature extraction, and analysis decision-making.

[0032] Step 7: Maintain Decision Generation and Execution

[0033] By combining local early warning signals with cloud-based diagnostic reports, a tiered maintenance instruction is generated: immediate maintenance, planned maintenance, or continuous monitoring. The instruction is then sent to the flue gas purification terminal, and the equipment maintenance record database is updated simultaneously.

[0034] Furthermore, in step 1, the sampling frequency is set to be more than 15 times the fundamental frequency.

[0035] Compared with the prior art, this application has at least one of the following beneficial technical effects:

[0036] 1. Eliminate the risk of sensor failure and improve system reliability.

[0037] By replacing traditional physical sensors with non-invasive electrical parameter coupling technology, the problem of probe scaling and corrosion is avoided. Combined with impedance matching network and electromagnetic shielding design, it can stably acquire equipment operating characteristic parameters in high temperature / high humidity environments, and its lifespan is increased by more than 70% compared with traditional sensors.

[0038] 2. Achieve multi-dimensional state perception and predictive maintenance

[0039] By employing a multi-feature vector-based analysis system, sub-health conditions such as pipe dust accumulation can be identified 3-5 operating cycles in advance, overcoming the lag limitations of traditional threshold alarms. The cloud-based federated learning framework supports dynamic model iteration, continuously optimizing diagnostic accuracy.

[0040] 3. Reduce system deployment and maintenance costs

[0041] Modular design reduces wiring requirements by 80%, while intelligent switching circuits and breakpoint resume mechanisms reduce maintenance downtime to one-third of traditional solutions.

[0042] The beneficial effects listed above are not exhaustive of all advantages. Other potential beneficial effects and detailed technical implementation methods will be further disclosed in the embodiments or other descriptive sections of this application. Attached Figure Description

[0043] A better understanding of various aspects of this disclosure will be achieved by reading the following detailed description in conjunction with the accompanying drawings. The positions, dimensions, and extents of the structures shown in the drawings, etc., do not always represent actual positions, dimensions, and extents. In the drawings:

[0044] Figure 1 This is a hardware connection block diagram of one embodiment disclosed in this application.

[0045] Figure 2 This is a flowchart of a method according to an embodiment of this application. Detailed Implementation

[0046] The present disclosure will now be described with reference to the accompanying drawings, which illustrate several embodiments of the present disclosure. However, it should be understood that the present disclosure can be presented in many different ways and is not limited to the embodiments described below; in fact, the embodiments described below are intended to make the disclosure more complete and to fully illustrate the scope of protection of the present disclosure to those skilled in the art. It should also be understood that the embodiments disclosed herein can be combined in various ways to provide further additional embodiments.

[0047] In one embodiment of this application, the system overcomes the problems of sensor failure and high maintenance costs in traditional monitoring methods, and realizes accurate and efficient monitoring and intelligent evaluation of flue gas purification devices, ensuring the stable operation of flue gas purification devices and providing reliable technical support for environmental protection in industrial production.

[0048] In the hardware components, refer to the appendix. Figure 1 This application provides an improved embodiment, relating to a sensorless intelligent monitoring system for flue gas purification devices based on the coupling of multiple operating parameters. This system effectively solves the problems of sensor failure and high maintenance costs in traditional monitoring methods, achieving accurate and efficient monitoring and intelligent evaluation of flue gas purification devices, ensuring stable equipment operation, and providing reliable technical support for industrial environmental protection.

[0049] In terms of hardware architecture, the system mainly consists of a flue gas purification terminal and a cloud server, which establish a two-way data interaction link through wired or wireless networks. The flue gas purification terminal adopts a layered architecture, including physical execution units, signal conditioning units, and timing data interaction units. The functional units achieve seamless connection and command transmission through standard interface protocols, ensuring efficient collaborative operation of the system.

[0050] The physical actuators used for flue gas purification include a power drive module, an electrostatic treatment module, a photocatalytic treatment module, and detection components. The detection components include a voltage monitoring circuit, a circuit detection circuit, an impedance detection circuit, and a shunt sampling module. Specifically, the power input terminal of the power drive module is equipped with a voltage monitoring circuit, which achieves non-contact signal acquisition through a parallel voltage sensor; the current monitoring circuit uses a closed-loop sensor based on the principle of electromagnetic induction, connected in series with the main power cable to form a lossless current sampling path. The electrostatic treatment module is equipped with an impedance detection circuit, which extracts high-frequency characteristic signals through capacitive coupling technology, and its transmission link uses a shielded twisted-pair structure to enhance anti-interference capabilities. The DC bus of the photocatalytic module is connected to the shunt sampling module, and a four-terminal Kelvin connection method is used to eliminate wire resistance errors and ensure sampling accuracy.

[0051] The signal conditioning unit includes an electrically isolated amplification module and a multi-channel synchronous sampling module. The electrically isolated amplification module uses opto-isolation technology to achieve complete electrical isolation between signal channels; the multi-channel synchronous sampling module is based on an FPGA architecture and equipped with a Σ-Δ analog-to-digital converter. Each channel is synchronized through phase-locked loop technology to ensure the consistency of data acquisition.

[0052] The timing data interaction unit integrates a dual-mode communication protocol stack, including either a wired or wireless communication module. The wired module features an RJ45 interface and opto-isolation circuitry, employing differential signal transmission. The wireless module incorporates a programmable RF front-end, supports dual-band communication, and its antenna interface is configured with an impedance matching network to optimize signal radiation efficiency. Data transmission utilizes an end-to-end encryption protocol to ensure transmission security and integrity.

[0053] The cloud server adopts a distributed architecture, including parallel computing nodes and a time-series data storage cluster. The computing nodes support real-time inference for machine learning models, and the storage cluster uses columnar compression storage technology to optimize write performance. After preprocessing, the data collected by the flue gas purification terminal is uploaded to the cloud via a breakpoint resume mechanism. The cloud then generates an equipment health assessment report through feature extraction and pattern recognition and sends it back to the on-site terminal.

[0054] In practical applications, taking a kitchen exhaust gas purification system as an example, this system replaces traditional physical sensors with non-invasive electrical parameter coupling technology, avoiding probe scaling and corrosion problems. It can stably acquire equipment operating characteristic parameters even in high-temperature and high-humidity environments. Experiments show that the system's service life is increased by more than 70% compared to traditional sensors, significantly improving reliability while reducing maintenance time and costs.

[0055] For the monitoring methods of this system, please refer to the appendix. Figure 2 The first stage involves device identification and signal synchronization. In this stage, a globally unique identifier (UUID) is assigned to each monitoring terminal, generated as a 128-bit code according to the ISO / IEC 9834 standard. Three-phase voltage and current signals of the drive motor are synchronously acquired through an isolated sampling circuit, with the sampling frequency set to at least 15 times the fundamental frequency. The raw signals are then converted from analog to digital and encapsulated into data packets. These data packets include a header checksum sequence, the UUID, and timestamp information. This design ensures accurate identification and synchronization of the acquired signals, laying a solid foundation for subsequent processing and analysis.

[0056] Next is the dynamic feature parameter extraction stage. In this stage, instantaneous power calculation is performed on the acquired voltage and current signals, and the active power component is separated using orthogonal decomposition. Simultaneously, the spectral distribution characteristics of the electrostatic unit's operating current are analyzed, and the proportion of high-frequency harmonic energy is extracted. At the same time, the dynamic impedance parameters of the drive circuit are calculated, generating a multi-dimensional feature vector containing power fluctuation coefficients, harmonic distortion rate, and impedance changes. Specifically, the Clark orthogonal transform formula for instantaneous power calculation is as follows:

[0057] Eliminating zero-sequence components and constructing orthogonal axes for instantaneous active power

[0058]

[0059] in, These are the three-phase voltages, respectively. These are two-phase orthogonal currents. This transformation converts a three-phase unbalanced system into a two-phase orthogonal system, thereby eliminating neutral point offset errors and enabling accurate calculation of instantaneous active power. Simultaneously, high-frequency harmonic distortion analysis employs a windowed FFT algorithm, as shown in the following formula:

[0060]

[0061] in, This is a current sampling sequence. This is the Hanning window function. This algorithm can effectively extract high-frequency harmonic components from a signal, and then calculate the high-frequency harmonic distortion rate (THD), providing crucial harmonic information for subsequent analysis.

[0062] In addition, dynamic impedance monitoring uses the effective value of a sliding window for calculation, as shown in the following formula:

[0063]

[0064] in, and These are the voltage and current sample values, respectively. This is the window length. In this way, the dynamic impedance changes of the system can be monitored in real time. When the impedance change exceeds a preset threshold, an early warning can be triggered in a timely manner, reminding relevant personnel to conduct inspections and handle the situation.

[0065] After feature parameter extraction, the local edge state assessment stage begins. In this stage, the generated feature vectors are input into a pre-trained machine learning model to perform preliminary diagnosis based on the random forest algorithm. When a power parameter deviates positively from the baseline value by more than a preset threshold, or when high-frequency harmonic indicators show a continuous increase for three consecutive cycles, a local early warning signal is immediately triggered. Simultaneously, the assessment results and the original data are cached together in the local storage area. Specifically, the input features of the random forest model include power deviation, harmonic distortion rate changes, impedance change rate, and power fluctuation coefficient. Through comprehensive analysis of these features, the model can quickly determine whether the equipment's operating status is abnormal and issue timely early warning signals in abnormal situations, allowing for appropriate measures to be taken to prevent potential faults from worsening.

[0066] Following the local edge state assessment, the next step is a deep cloud-based analysis and diagnostic process. In this process, feature data and device operation logs are uploaded to a cloud server via an encrypted channel, leveraging the powerful computing capabilities of the cloud to perform time-series pattern analysis based on a Long Short-Term Memory (LSTM) network. The hidden layer equations of the LSTM network are as follows:

[0067]

[0068] in,,, These are the outputs of the forget gate, input gate, and output gate, respectively. Candidate memory cell states, This is the updated state of memory cells. The data is in a hidden state. The input dimension is 20 (including features such as power, harmonics, and impedance), the number of hidden units is 128, and the time step is 36 (6 hours of data, 10-minute intervals). The physical rationality of the feature parameters is verified using a fluid dynamics simulation model. The predicted value of pipe ash accumulation thickness and remaining effective operating time are further calculated, ultimately generating a result report including maintenance priority scoring. The cloud server, leveraging its powerful computing capabilities and storage resources, can perform in-depth mining and analysis of large amounts of historical and real-time data, uncovering potential patterns and trends in equipment operation. This enables more accurate assessment and prediction of equipment status, providing a strong basis for formulating reasonable maintenance strategies.

[0069] The next step is the control strategy optimization and feedback phase. In this phase, based on cloud-based diagnostic results, the system dynamically adjusts equipment operating parameters. For example, it implements power compensation control for the drive motor to maintain system stability and optimizes the operating voltage of the electrostatic unit to suppress abnormal discharge. The power compensation algorithm is as follows: [Parameters omitted].

[0070]

[0071] The execution cycle is 100ms, and .

[0072] This PID control algorithm can adjust the compensation power in real time according to the system's power deviation, maintain stable system operation, effectively cope with dynamic changes and disturbances in the system, and ensure the normal operation of the flue gas purification device. Simultaneously, it updates the decision threshold parameters of the local evaluation model, forming an adaptive control closed loop. This allows the system to continuously optimize the control strategy based on real-time feedback information, improving control accuracy and adaptability, and ensuring efficient and stable operation of the equipment under different operating conditions.

[0073] In terms of model iteration and data traceability, a device group knowledge base is established, and the feature data of each terminal is aggregated with the help of the federated learning framework to optimize the weight parameters of the cloud analysis model.

[0074] The parameter aggregation formula is as follows: where is the terminal loss value.

[0075]

[0076] The aggregation cycle is 24 hours, and the 50% of terminals with the lowest loss values ​​are selected to participate in the aggregation.

[0077] The federated learning mechanism enables the sharing and optimization of model parameters while protecting the data privacy of each terminal, thereby improving the intelligence level and diagnostic capabilities of the entire system. This allows the system to continuously learn and adapt to new operating conditions and fault modes, and continuously optimize the monitoring effect.

[0078] In addition, full-link data tracing is initiated for anomaly detection events to verify the data integrity of each stage of signal acquisition, feature extraction, and analysis and decision-making, ensuring the accuracy of data at each stage and providing reliable data support for anomaly diagnosis. It also helps to identify and improve weak links in the system, further enhancing the system's reliability and stability.

[0079] Finally, there is the maintenance decision generation and execution phase. In this phase, by integrating local early warning signals and cloud-based diagnostic reports, tiered maintenance instructions are generated, specifying whether immediate maintenance, planned maintenance, or continuous monitoring is required. These instructions are then precisely distributed to the flue gas purification terminal via the industrial bus, simultaneously updating the equipment maintenance record database. This design ensures the timeliness and effectiveness of maintenance work, allowing for the rational arrangement of maintenance plans based on the actual condition of the equipment. This avoids work interruptions caused by equipment failures, reduces unnecessary maintenance costs, and improves the overall operating efficiency and lifespan of the equipment.

[0080] Understandably, flue gas purification equipment is crucial in the treatment of flue gas in various industries and catering establishments. Taking a barbecue restaurant as an example, its flue gas purification grill table is one of the core pieces of equipment. This equipment uses non-invasive electrical parameter coupling technology to synchronously collect the three-phase voltage and current signals of the drive motor, thereby achieving intelligent monitoring of the equipment's operating status. When the power parameters of the drive motor deviate positively from the benchmark value by more than a preset threshold, or when the high-frequency harmonic index increases for three consecutive cycles, the system can promptly trigger an early warning signal. This process is based on a pre-trained machine learning model, analyzing multi-dimensional feature vectors such as power fluctuation coefficient, harmonic distortion rate, and impedance change. When an abnormality is detected, the system automatically sends an alarm to the maintenance terminal in the restaurant, reminding staff to replace filter consumables or clean the channels in a timely manner to ensure optimal flue gas purification. Furthermore, the system can generate equipment status assessment reports through a cloud server, providing detailed maintenance suggestions and decision support for the restaurant, thereby effectively reducing maintenance costs, improving equipment operating efficiency and lifespan, ensuring indoor air quality, and creating a comfortable and healthy dining environment for customers.

[0081] While exemplary embodiments of this disclosure have been described, those skilled in the art will understand that various changes and modifications can be made to the exemplary embodiments of this disclosure without departing from the spirit and scope thereof. Therefore, all changes and modifications are included within the scope of protection of this disclosure as defined by the claims. This disclosure is defined by the appended claims, and equivalents of those claims are also included.

Claims

1. A sensorless intelligent monitoring system for flue gas purification devices based on the coupling of multiple operating parameters, characterized in that, The system includes a flue gas purification terminal and a cloud server, which interact bidirectionally via a network link; The flue gas purification terminal includes a physical execution unit with electrical parameter detection, a signal conditioning unit for acquiring electrical parameter detection data from the physical execution unit, and a timing data interaction unit for sending data processed by the information conditioning unit. The physical execution unit for achieving flue gas purification includes at least a power drive component, an electrostatic treatment component, a photocatalytic treatment component, and a detection component for detecting the electrical parameters of each component. The detection components include a voltage monitoring circuit and a current detection return circuit located at the power input port of the power drive component, an impedance detection circuit located at the output port of the electrostatic treatment component, and a shunt sampling module located on the power supply bus of the photocatalytic treatment component.

2. The sensorless intelligent monitoring system for flue gas purification devices based on multi-operating parameter coupling as described in claim 1, characterized in that, The power drive component is a constant speed motor.

3. The sensorless intelligent monitoring system for flue gas purification devices based on multi-operating parameter coupling as described in claim 1, characterized in that, The voltage monitoring circuit achieves non-contact signal acquisition through parallel voltage sensors; the current monitoring circuit uses a closed-loop sensor based on the Faraday electromagnetic induction principle to form current sampling.

4. The sensorless intelligent monitoring system for flue gas purification devices based on multi-operating parameter coupling as described in claim 1, characterized in that, The impedance detection circuit extracts high-frequency characteristic signals through capacitive coupling technology.

5. The sensorless intelligent monitoring system for flue gas purification devices based on multi-operating parameter coupling as described in claim 1, characterized in that, The DC power supply bus is connected to a shunt sampling module for current and voltage detection.

6. The sensorless intelligent monitoring system for flue gas purification devices based on multi-operating parameter coupling as described in claim 1, characterized in that, The signal conditioning unit includes a distributed signal preprocessing submodule and a synchronous sampling control submodule; the signal preprocessing submodule integrates an electrically isolated amplifier circuit and an adaptive filter circuit.

7. The sensorless intelligent monitoring system for flue gas purification devices based on multi-operating parameter coupling as described in claim 1, characterized in that, The synchronous sampling control submodule is based on a field-programmable gate array to build a multi-channel parallel sampling architecture and is equipped with an analog-to-digital converter to achieve accurate acquisition of wide dynamic range signals.

8. The sensorless intelligent monitoring system for flue gas purification devices based on multi-operating parameter coupling as described in claim 1, characterized in that, The timing data interaction unit includes a wired communication module and / or a wireless communication module.

9. The sensorless intelligent monitoring system for flue gas purification devices based on multi-operating parameter coupling as described in claim 1, characterized in that, The cloud server unit adopts a distributed computing architecture, which includes parallel processing computing nodes and time-series data storage clusters.

10. An intelligent monitoring method, characterized in that, The sensorless intelligent monitoring system for flue gas purification devices based on the coupling of multiple operating parameters, as described in claims 1 to 9, comprises the following steps: Step 1: Synchronously acquire device identification and signal. Each monitoring terminal is assigned a globally unique identifier. The voltage and current signals of the drive motor are synchronously collected through an isolated sampling circuit. The raw signals are encapsulated into data packets after analog-to-digital conversion, which include packet header verification sequence, UUID and timestamp information. Step 2: Dynamic Feature Parameter Extraction Instantaneous power calculations are performed on voltage and current signals, and the active power component is separated using orthogonal decomposition. The spectral distribution characteristics of the electrostatic unit's operating current are analyzed simultaneously, and the proportion of high-frequency harmonic energy is extracted. The dynamic impedance parameters of the drive circuit are calculated, generating a multi-dimensional feature vector containing power fluctuation coefficients, harmonic distortion rate, and impedance changes. Step 3: Local Edge State Assessment The feature vector is input into a pre-trained machine learning model to perform a preliminary diagnosis based on the random forest algorithm; when the power parameter is detected to deviate positively from the benchmark value by more than a preset threshold, or when the high-frequency harmonic index increases for three consecutive cycles, a local early warning signal is triggered; the evaluation results and the original data are cached together in the local storage area. Step 4: In-depth cloud-based analysis and diagnosis Feature data and equipment operation logs are uploaded to the cloud server through an encrypted channel. Time-series pattern analysis based on long short-term memory network is performed. The physical rationality of the feature parameters is verified by combining fluid dynamics simulation model. The predicted value of pipe ash accumulation thickness and remaining effective operating time are calculated, and a result report including maintenance priority score is generated. Step 5: Control Strategy Optimization and Feedback Based on cloud-based diagnostic results, the equipment operating parameters are dynamically adjusted: power compensation control is implemented on the drive motor to maintain system stability, the operating voltage of the electrostatic unit is optimized to suppress abnormal discharge, and the decision threshold parameters of the local evaluation model are updated to form an adaptive control closed loop. Step 6: Model Iteration and Data Source Tracing Establish a device group knowledge base, aggregate feature data from various terminals through a federated learning framework, and optimize the weight parameters of the cloud-based analysis model; initiate full-link data tracing for anomaly detection events to verify the data integrity of each stage of signal acquisition, feature extraction, and analysis decision-making. Step 7: Maintain Decision Generation and Execution By combining local early warning signals with cloud-based diagnostic reports, a tiered maintenance instruction is generated: immediate maintenance, planned maintenance, or continuous monitoring. The instruction is then sent to the flue gas purification terminal, and the equipment maintenance record database is updated simultaneously.

Images