Automatic cleaning device and method for floating matter of wet-type slag conveyer

Abstract

The application relates to the technical field of power plant slag removal equipment, and discloses a wet-type slag conveyor floating object automatic cleaning device and method, which comprises a sensing and sensing unit, a control decision unit and a mechanical execution unit. The apparent elevation of the water surface and the actual water pressure elevation of the bottom layer are respectively acquired by the raw water level and the real water level sensors of the sensing and sensing unit; the control decision unit calculates the liquid level deviation of the two and outputs a start-stop instruction; the mechanical execution unit is controlled and driven to rotate the collection roller to collect the floating objects, the floating objects are peeled off by the anti-winding and detachment components, and then filtered and discharged by the inclined flow guide groove. The method calls a one-dimensional convolutional neural network to extract motor current, torque and rotating speed features to evaluate the winding probability, introduces a small constant in the data preprocessing denominator to prevent zero division overflow, and triggers reverse detachment; for multi-machine cooperation, a space coupling matrix is constructed, a standard unit matrix penalty term is superimposed as a regularization coefficient, matrix singularity under symmetrical conditions is eliminated, and scheduling deadlock is prevented.

Description

Technical Field

[0001] This application relates to the technical field of power plant slag removal equipment, specifically to an automatic cleaning device and method for floating debris in a wet slag removal machine. Background Technology

[0002] During operation, wet slag removal machines continuously accumulate various low-density floating debris on the surface of the tank, requiring automatic slag removal via mechanical cleaning devices. Existing cleaning equipment typically relies on a single surface water level sensor for automatic start / stop triggering. However, when a large amount of floating debris accumulates on the tank surface, a single sensor cannot physically distinguish between the thickness of the surface debris accumulation and the actual water pressure level of the underlying fluid, causing the sensor's apparent water level signal to deviate significantly from the actual fluid state. This physical distortion of the monitoring data directly leads the control module to obtain incorrect logical judgments, resulting in ineffective idling of the mechanical actuator or delayed cleaning actions.

[0003] During the physical process of a mechanical collection device immersing itself in water to perform rotating salvage, flexible fibers and long, thin impurities mixed in the fluid can easily become entangled on the rotating spindle and scraper components. As the machine operates, the mechanical locking force of the entangled material gradually increases, easily causing physical seizure of rotating components and overload of the drive motor. Existing device drivers typically lack feature extraction and prediction mechanisms for progressive mechanical entanglement. When the spindle speed abruptly decreases or even drops to zero due to mechanical jamming, the microprocessor's internal calculation program, when executing mathematical equations involving physical quantities such as rotational speed, often triggers arithmetic division-by-zero overflow errors due to the lack of prior denominator boundary constraints. This abnormal interruption of the underlying logic prevents the control system from outputting reverse torque in a timely manner to perform the decoupling action.

[0004] For multiple automatic cleaning devices arrayed on the tank of a large, long, straight slag removal machine, the existing system suffers from a fundamental algebraic flaw when performing multi-machine collaborative scheduling. When the existing central controller constructs a spatial state model and solves for the startup priorities of multiple devices, if the system encounters an extreme symmetrical operating condition where the local liquid level status of each physical defense zone and the cumulative standby time of the devices are absolutely consistent, the spatial coupling matrix constructed in the underlying derivation equations for jointly solving resource allocation weights will exhibit a highly linear correlation among its internal elements. This algebraic structure of the physical symmetry mapping causes the determinant of the composite matrix to approach zero, inevitably leading to matrix singularity when the controller performs inverse matrix solving. This ultimately causes a computational deadlock in the underlying microprocessor, resulting in the overall paralysis of the multi-machine parallel scheduling system. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this application provides an automatic floating debris cleaning device and method for wet slag removers, which solves the problems of distorted perception of floating debris accumulation leading to false start-stop, mechanical actuators being prone to physical seizure due to flexible fiber entanglement, and matrix singularity easily occurring under extreme symmetrical working conditions in multi-machine collaborative scheduling, leading to deadlock in the underlying computation of the central system.

[0006] To achieve the above objectives, this application provides the following technical solution:

[0007] The first aspect of this application provides an automatic cleaning device for floating debris in a wet slag remover. The device is installed inside the tank of the wet slag remover and includes: a sensing unit, a control decision unit, and a mechanical execution unit.

[0008] The sensing unit is installed in the tank area of ​​the wet slag removal machine, including a raw water level sensor installed inside the tank and a real water level sensor installed on the side wall of the tank. The raw water level sensor obtains the apparent physical elevation of the surface layer containing floating debris through contact measurement, while the real water level sensor uses a penetrating physical beam to penetrate the low-density floating debris layer and obtain the actual water pressure elevation of the fluid at the bottom. The control decision unit mainly consists of a logic control module, which receives the above dual-source detection data, calculates the difference between the apparent elevation and the actual water pressure elevation, converts the invisible floating debris accumulation state into a continuous differential electrical signal, and outputs start / stop control commands accordingly.

[0009] The mechanical actuator is located in the space above the wet slag removal machine tank, including a floating debris collection mechanism horizontally spanning above the water surface, a conveying mechanism located on one side of the tank and directly below the discharge side of the floating debris collection mechanism, and a drive mechanism coaxially connected. The drive mechanism receives start / stop control commands and outputs mechanical torque to drive the floating debris collection mechanism to perform rotational collection operations.

[0010] In one specific embodiment, the physical structure of the floating object collection mechanism is defined as follows: it includes a collection roller and an anti-tangling and desorption component. A filter adhesion strip is fixedly connected to the outer circumferential surface of the collection roller, and the strip has perforated mesh holes arranged in an array.

[0011] During the dynamic process of the collecting roller rotating and cutting into the water surface, the permeable mesh provides a fluid pressure relief channel, discharges the liquid water squeezed out by the collecting roller, and reduces the hydrodynamic resistance experienced by the roller rotation. Simultaneously, solid floating debris is retained on the surface of the belt. Multiple flexible scooping scrapers are spaced apart along the axial direction of the collecting roller on the outer surface of the filter adhesion belt, extending alternately below the water surface as the roller rotates, applying a mechanical lifting force to the floating debris. An anti-entanglement and desorption component is fixedly arranged on the discharge side, with its working end close to the movement trajectory of the flexible scooping scrapers. When the flexible scooping scrapers carry floating debris to the discharge side, the fixing structure of the anti-entanglement and desorption component applies a lateral shearing force to the floating debris attached to it, forcibly breaking the surface tension and mechanical entanglement force of the attached debris, causing it to detach.

[0012] In one specific embodiment, the conveying mechanism is defined as follows: it includes an inclined guide trough and a moisture filter. The feed end of the inclined guide trough is flared and suspended directly below the anti-entanglement and desorption component. The moisture filter is embedded in the bottom receiving surface of the inclined guide trough. Detached floating objects fall into the inclined guide trough under gravity and slide outwards driven by the gravitational component provided by the set inclination angle. During the sliding displacement, the residual moisture carried by the floating objects flows back into the wet slag remover tank under gravity through the moisture filter, achieving solid-liquid separation.

[0013] The second aspect of this application provides an automatic cleaning method for floating debris in a wet slag remover, applied to the automatic cleaning device described in the first aspect of this application. This method controls the start-up, shutdown, and anomaly handling of the equipment through multi-dimensional state assessment and a closed-loop negative feedback mechanism. Specifically, the method includes: The sensing unit performs high-frequency continuous monitoring to acquire timing status signals. The logic control module executes a timestamp alignment procedure, using the system's low-frequency clock signal as a reference, to downsample and slice the high-frequency acquired data, aligning the phase of each sensor's data on the time axis and eliminating time-domain misalignment caused by differences in hardware sampling rates.

[0014] The logic control module performs a three-dimensional start-up condition determination. A start signal is output only if one of the following three conditions occurs: 1) The calculated liquid level difference signal is greater than or equal to the set allowable deviation threshold, indicating that the local accumulation thickness of floating debris has reached the mechanical collection requirement; 2) The actual water level is lower than the set minimum safe water level warning line, indicating a physical risk of water seal damage and leakage in the tank, requiring forced start-up of the equipment for turbulence-induced water replenishment; 3) The system's continuous standby time reaches the preset cycle period, indicating that the equipment meets the periodic no-load operation conditions for anti-jamming.

[0015] In one specific embodiment, the control method of this application includes a driving and self-cleaning linkage step based on mapping the mechanical load state using underlying electrical parameters. Specifically, the operation is as follows: The system extracts three electrical and kinematic physical quantities: the amplitude of the motor stator current, the real-time output torque of the frequency converter, and the spindle encoder speed. The linkage diagnostic module then uses a lightweight one-dimensional convolutional neural network model to extract features from these time-series physical quantities. Because the spindle speed exhibits non-linear decay and the stator current experiences high-frequency fluctuations when fiber entanglement occurs in mechanical components, the network model assesses the probability of mechanical entanglement by calculating the covariance characteristics of these three parameters.

[0016] In the preprocessing mathematical equations for model data standardization, a small positive real number is forcibly introduced as a denominator constant to prevent an arithmetic division-by-zero overflow error from occurring inside the microprocessor when the physical rotational speed detection value drops to zero. When the entanglement probability output by the network model exceeds the set upper limit for multiple consecutive calculation cycles, and the steady-state current of the frequency converter drive feedback reaches the overload multiple boundary of the rated nominal value, it is determined that the mechanical component has physically seized. The control module immediately cuts off the forward output pulse, issues a reverse intervention command, drives the motor to output the maximum starting torque in the reverse direction, and releases the mechanical lock of the flexible fiber through the reverse motion trajectory.

[0017] In one specific embodiment, the control method of this application provides a multi-machine collaboration and matrix regularization anti-deadlock step for managing multiple automatic cleaning devices installed on the same long straight groove.

[0018] When multiple devices operate within the same fluid boundary, start-up and shutdown actions generate surface ripple superposition and fluid interference. The central controller acquires the local liquid level deviation and spatial physical distance of each physical protection zone, constructing a spatial coupling matrix and an urgency column vector. By solving this covariance matrix, the resource allocation weight of each device is calculated, and start-up commands are issued based on the descending order to generate a staggered timing sequence for the physical operation of each device.

[0019] In extreme physical conditions where the liquid levels in multiple tanks are absolutely symmetrical and the standby time is exactly the same, the spatial coupling matrix in the underlying derivation equations will exhibit a linearly dependent state with a determinant approaching zero (i.e., matrix singularity). In this case, the controller's matrix inversion operation will cause the central scheduler to deadlock. To eliminate this risk, a penalty term (regularization coefficient) based on the standard identity matrix is ​​superimposed along the main diagonal of the original spatial coupling matrix in the underlying solution equations. This mathematical operation forcibly increases the absolute values ​​of the main diagonal elements, disrupting the original extremely symmetrical matrix eigenvalue structure. From an algebraic topological perspective, this ensures that the synthesized matrix satisfies the strictly positive definite and invertible mathematical boundary constraints under any operating condition, guaranteeing that the controller can still output a deterministic finite solution sequence even under extremely symmetrical conditions.

[0020] This application provides an automatic cleaning device and method for floating debris in a wet slag remover. It has the following beneficial effects: 1. This application involves arranging a primary water level sensor for detecting the apparent water level and a secondary water level sensor for detecting the actual water pressure at the bottom in the tank area of ​​a wet slag removal machine. A logic control module calculates the absolute value of the difference between the two sensors as the liquid level deviation. This hardware layout and differential calculation process transforms the invisible physical accumulation thickness of floating debris on the water surface into a continuously quantified electrical signal, providing a definite numerical basis for the start and stop of the mechanical execution unit. This solves the physical mechanism defect where a single sensor outputs erroneous trigger commands under complex liquid surface conditions, leading to ineffective idling of the equipment.

[0021] 2. This application utilizes a linkage diagnostic module to call a one-dimensional convolutional neural network model to extract features from the motor stator current, real-time output torque, and spindle speed to assess the winding probability. A small constant is introduced into the denominator of the data standardization preprocessing. When both the winding probability and overload current exceed their limits, a reversal intervention command is output. This control flow directly maps and processes the flexible fiber winding state of the mechanical components using the covariance characteristics of the underlying electrical and kinematic parameters. Simultaneously, the physical setting of the small constant in the denominator eliminates the risk of an arithmetic division-by-zero overflow error occurring within the microprocessor when the spindle speed decays to zero, ensuring the continuous execution of the underlying logic.

[0022] 3. This application constructs a spatial coupling matrix containing the physical spacing between equipment and an urgency column vector of local liquid level deviations through a central controller to solve for equipment startup priority. In the underlying derivation equations, a penalty term based on the standard identity matrix is ​​superimposed along the main diagonal of the spatial coupling matrix as a regularization coefficient. This mathematical operation step forcibly increases the absolute values ​​of the elements on the main diagonal of the composite matrix, disrupting the linear correlation state exhibited by multiple devices under extreme symmetrical operating conditions, eliminating matrix singularity, ensuring that the composite matrix is ​​strictly positive definite and invertible, and guaranteeing that the multi-machine collaborative system can output a definite staggered startup timing solution under any physical operating condition. Attached Figure Description

[0023] Figure 1 This is a front view of the device according to an embodiment of this application; Figure 2 This is a schematic diagram of the overall device architecture of an embodiment of this application; Figure 3 This is a macroscopic flowchart of the automatic cleaning method according to an embodiment of this application; Figure 4 This is a single-machine dual-source water level signal curve diagram according to an embodiment of this application; Figure 5 This is a time-domain response curve of the cleanup action in an embodiment of this application; Figure 6 This is a comparison diagram of the multi-machine operation effect under extreme symmetrical conditions according to an embodiment of this application.

[0024] The components include: 1. Wet slag removal machine tank; 2. Original water level sensor; 3. Real water level sensor; 4. Collection roller; 5. Filter water adhesion belt; 6. Flexible scooping scraper; 7. Guide channel; 8. Water filter screen; and 9. High-pressure spray pipeline. Detailed Implementation

[0025] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0026] In this embodiment, refer to the appendix Figure 1 and attached Figure 2 Based on the actual operating conditions of the underlying equipment in coal-fired power plants, this automatic cleaning device is installed entirely inside the wet slag remover tank 1 to solve the technical problems of water level signal distortion and potential equipment operation hazards caused by the accumulation of floating debris on the slag remover's surface. At the hardware level, the device mainly includes a sensing unit, a control decision unit, and a mechanical execution unit.

[0027] Specifically, the sensing unit is installed in the area of ​​the wet slag removal machine tank 1, and includes a raw water level sensor 2 and a real water level sensor 3. Regarding the specific structure of the raw water level sensor 2, those skilled in the art can use existing ultrasonic level gauges or float-type level gauges; their basic measurement principles and equipment selection are well-known technologies in the field and will not be elaborated here. As a preferred approach, considering the limitations of conventional measurement principles, the apparent water level signal detected by the raw water level sensor 2 will drift or distort when porous, lightweight floating debris covers the water surface due to diffuse sound reflection or changes in buoyancy. The real water level sensor 3 is specifically implemented as an infrared dual-beam level gauge. This infrared dual-beam level gauge is installed on the side wall of the wet slag removal machine tank 1 and has the physical characteristic of penetrating the floating debris layer on the water surface, used to obtain the real water level signal inside the wet slag removal machine tank 1 that is not disturbed by floating debris.

[0028] To achieve comprehensive processing of the aforementioned heterogeneous sensor data, the control decision unit mainly consists of a logic control module. The specific lower-level technical feature of the logic control module is a programmable logic controller (PLC) system or a substation module of a distributed control system (DCS). The logic control module is electrically connected to the original water level sensor 2 and the actual water level sensor 3 via signal cables, respectively, to receive real-time detection data of the apparent and actual water levels. The logic control module's memory is pre-loaded with a deviation calculation algorithm and start / stop judgment threshold parameters. Its microprocessor generates electrical signals based on the sensor data comparison results and outputs start / stop control commands to the subsequent mechanical execution unit.

[0029] Based on the aforementioned spatially distributed hardware system architecture, the mechanical execution unit is arranged in the space above the wet slag removal machine tank 1, mainly including a floating debris collection mechanism, a conveying mechanism, and a drive mechanism. The floating debris collection mechanism is horizontally mounted on two guide rails above the water surface of the wet slag removal machine tank 1, with its bottom area immersed in the water surface of the wet slag removal machine tank 1. The conveying mechanism is arranged on one side of the wet slag removal machine tank 1, with its feed end fixed directly below the discharge side of the floating debris collection mechanism, forming a material receiving and receiving cooperation in spatial position. The drive mechanism is specifically implemented as an integrated motor and reducer with a waterproof shell. This drive mechanism is coaxially connected to the main shaft of the floating debris collection mechanism, and the electrical control terminal of the drive mechanism is electrically connected to the output terminal of the logic control module.

[0030] See attached document Figure 3 In actual industrial operation scenarios, the logic control module controls the entire automatic cleaning device to operate in a closed loop according to a specific workflow. The specific operating steps are described below.

[0031] Step S1: The sensing unit performs real-time monitoring. After the system is powered on, the original water level sensor 2 continuously detects the apparent water level signal in the wet slag removal machine tank 1, while the real water level sensor 3 simultaneously detects the real water level signal after penetrating the floating debris layer. The two sensors transmit the acquired analog or digital water level signals to the logic control module in real time at a fixed sampling period. Considering the inherent delay differences in the hardware response of multi-source sensors, the logic control module has a timestamp alignment program. Using the signal period with the lower sampling frequency as a benchmark, it performs downsampling time slice alignment on the high-frequency acquired signals to ensure strict alignment of the apparent water level signal and the real water level signal in the time domain within the same calculation period, avoiding false deviations introduced by time misalignment.

[0032] Step S2: The logic control module performs liquid level deviation calculation and start-up condition determination. Based on the multi-source input data aligned in the time domain, the logic control module executes internal program code to extract interference features according to the general technical principle of differential compensation. The liquid level deviation is calculated using the following mathematical formula: ; In this formula, This represents the apparent water level measured by the original water level sensor 2; This represents the actual water level measured by the real water level sensor 3. This represents the absolute value of the difference between the apparent water level and the actual water level, i.e., the actual calculated liquid level deviation. The physical meaning of this formula is that it filters out the positive and negative interference caused by liquid level fluctuations through absolute value calculation, and directly quantifies the amount of false liquid level rise caused by conventional acoustic or mechanical buoyancy detection due to the floating object layer.

[0033] Based on this data, the logic control module determines whether the current operating condition meets the start-up conditions according to the calculated liquid level deviation and system running time parameters. To avoid frequent malfunctions caused by relying on a single extreme value, the judgment logic of the output result is constructed as a parallel trigger network with three dimensions: when the liquid level deviation is greater than or equal to the preset deviation allowable threshold (in this embodiment, the deviation allowable threshold is set to a range of 50mm to 150mm, and its specific value is determined by weighting the physical overflow tolerance of the slag removal machine tank and the inherent error bands of the two sets of sensors); Alternatively, the actual water level may fall below the preset minimum safe water level warning line; or the cumulative time the system has been in standby monitoring mode may reach the preset cyclic waiting period. When any of the above triggering conditions are met, the logic control module determines that there is floating debris interference or that periodic cleaning and maintenance are required, and the system meets the startup conditions.

[0034] In step S3, the mechanical execution unit performs the floating debris collection and discharge operation. When the start-up conditions are met, the logic control module outputs a high-level or pulse start control command to the drive mechanism. The drive mechanism is energized and provides rotational driving force to drive the floating debris collection mechanism to rotate continuously and uniformly. The partially submerged floating debris collection mechanism scoops up the floating debris from the water surface and flips and transfers it to its discharge side along the rotational trajectory. The conveying mechanism located below the discharge side receives the floating debris that falls off the floating debris collection mechanism and discharges it to a designated area outside the wet slag removal machine tank 1 through gravity guidance caused by the inclined structure.

[0035] Step S4: The logic control module performs stop condition verification and system reset. During the operation of the floating debris collection mechanism and the conveying mechanism, the original water level sensor 2 and the actual water level sensor 3 continuously feed back the water level signal after the water surface state changes to the logic control module. The logic control module verifies the stop condition in real time. The determination of the stop condition must simultaneously meet three constraints: the recalculated liquid level deviation is less than the preset allowable deviation threshold, indicating that the floating debris causing signal distortion has been effectively removed; the actual water level is greater than or equal to the standard safe water level, indicating that the water level in the tank is within the safe operating range; and the current continuous operating time of the device since receiving the start command is greater than or equal to the preset minimum shutdown delay (in this embodiment, the value of the minimum shutdown delay ranges from 30 seconds to 120 seconds, which is derived from the total mechanical transmission travel time of a single floating debris movement from the collection end to the collection box).

[0036] The introduction of the aforementioned minimum shutdown delay effectively consolidates the cleaning effect and prevents premature shutdown of the mechanism due to the brief attainment of the liquid level signal caused by dynamic fluctuations in the water surface. When all three judgment criteria mentioned above are met simultaneously, the logic control module outputs a stop control command to the drive mechanism. The drive mechanism is powered off and shuts down, the floating debris collection mechanism stops rotating, the state variables and timers inside the logic control module are cleared and reset, and the entire automatic cleaning device re-enters the real-time monitoring standby state of step S1.

[0037] See attached document Figure 3 The sensing and logic control center constitutes the control unit of the entire automatic cleaning device. Its main task is to acquire on-site status data and identify whether there are floating objects on the water surface of the slag remover that affect normal operation after filtering out interference.

[0038] In this embodiment, the basic hardware of the sensing unit relies on a dual-source detection architecture consisting of the original water level sensor 2 and the actual water level sensor 3. The original water level sensor 2 is fixedly installed inside the wet slag removal machine tank 1. For the ranging component and signal conversion circuit of the original water level sensor 2, those skilled in the art can use an industrial-grade ultrasonic probe or a float resistance transmitter. The specific hardware selection and wiring are well-known technologies in the field and will not be described in detail here. Due to the limitations of the measurement mechanism, the original water level sensor 2 is mainly used to detect the apparent water level in the wet slag removal machine tank 1. When there is a large amount of floating ash or porous flocculent matter on the water surface, the feedback value of the apparent water level will be significantly higher than the actual liquid level due to the rise of the physical interface. In order to obtain a comparison benchmark that is not affected by the physical covering layer, the actual water level sensor 3 is installed on the wet slag removal machine tank 1. As a preferred embodiment, the lower feature of the actual water level sensor 3 is an infrared dual-beam level gauge. This infrared dual-beam level gauge utilizes the physical properties of infrared light waves at specific frequency bands, which have high transmittance to lightweight porous floating objects and high absorption rate to liquid water, to penetrate the floating object layer on the water surface and detect the true water level in the tank 1 of the wet slag removal machine.

[0039] Based on the dual-source analog signals acquired by the aforementioned hardware, the logic control module is electrically connected to both the original water level sensor 2 and the actual water level sensor 3 via shielded twisted-pair cables for interference suppression. The underlying microprocessor of the logic control module receives the apparent and actual water levels in real time and outputs start / stop control commands based on the deviation comparison results after alignment in the time domain. To ensure the rigor and robustness of this logical decision-making process, the logic control module internally incorporates an automated calculation and verification program with fault-tolerant mechanisms, specifically divided into the following logical operation steps.

[0040] The logic control module continuously samples the voltage or current signals output by the sensors at a set millisecond interval and converts them into engineering units. Considering the potential for electromagnetic interference or transient extreme values ​​caused by violent water churning in industrial environments, the logic control module performs a moving average filtering algorithm on the introduced apparent water level signal and the actual water level signal before entering the core calculation. This preprocessing logic effectively eliminates isolated noise points that deviate from the normal distribution, preventing subsequent misjudgments caused by a single abnormal data point. Furthermore, to address the issue of inconsistent sampling frequencies between the two types of sensors, the logic control module introduces a timestamp synchronization mechanism. Using the lower-frequency sensor sampling period as a benchmark, the high-frequency signal is downsampled and truncated, ensuring that the two sets of data entering subsequent calculation steps are in the same working condition slice on the physical timeline.

[0041] Based on the filtered steady-state data, the logic control module executes its internal arithmetic logic unit according to the theoretical basis of differential compensation to calculate the level deviation between the apparent water level signal and the actual water level signal. The level deviation is calculated using the following mathematical formula: ; In the above formula, Represents the current time point The apparent water level value output by the original water level sensor 2; Represents the same time point The actual water level value output by the lower actual water level sensor 3; This represents the calculated liquid level deviation, i.e., the absolute value of the difference between the apparent water level signal and the true water level signal. The physical meaning of this calculation logic is that, by calculating the absolute difference between the two, the false liquid level rise caused by the accumulation of floating debris is completely separated from the base liquid level, making it an independent quantitative indicator characterizing the thickness and degree of interference of floating debris.

[0042] To avoid the device starting and stopping from getting stuck in a logical dead loop due to relying on a single parameter, the specific logic for determining whether the start-up conditions are met is constructed as a judgment network containing three parallel trigger branches.

[0043] Specifically, the system determines that the startup conditions are met when one of the following three operating conditions occurs.

[0044] The first operating condition is deviation exceeding the limit trigger. The system is triggered when the calculated liquid level deviation is greater than or equal to a preset allowable deviation threshold. This allowable deviation threshold is typically set between 50mm and 150mm, and its specific value is determined by on-site process personnel based on the physical height margin of the overflow port of the wet slag remover tank 1, combined with the maximum combined error band of the two sets of sensors.

[0045] The second operating condition is forced triggering at extremely low water levels. The system is activated when the actual water level signal falls below a preset minimum safe water level warning line. The preset minimum safe water level warning line is determined by the water seal safety height of the wet slag remover tank 1. This setting aims to prevent false safety feedback from the original water level sensor 2 indicating a high water level due to severe accumulation of floating debris, while the actual water body is facing the crisis of drying out or seal failure.

[0046] The third operating condition is periodic forced triggering. When the cumulative standby time of the system reaches a preset cycle waiting period (in this embodiment, the preset cycle waiting period is a fixed value within the range of 2 to 8 hours), the system is triggered regardless of the sensor readings, as this preset cycle waiting period is limited by the average generation rate of floating objects under historical operating conditions. This logic is used to deal with the signal freezing caused by contamination and scaling of the sensor probe itself, and to break the dead zone through periodic mechanical cleaning.

[0047] When the system is in the cleanup phase, the logic control module continuously verifies the stop conditions. Unlike the broad triggering of start conditions, the determination of stop conditions uses a serial logic structure. A stop condition must simultaneously meet the following three criteria.

[0048] The first criterion is that the source of interference has been eliminated, meaning that the liquid level deviation calculated in real time must be less than the preset allowable deviation threshold.

[0049] The second criterion is to characterize the safety of the system itself, that is, the actual water level signal must be greater than or equal to the standard safe water level, indicating that the water replenishment and sealing status in the tank has been restored to normal.

[0050] The third criterion is consolidation delay control, which requires that the current continuous operating time after the device starts must be greater than or equal to the minimum downtime. The value of this minimum downtime is calculated by dividing the physical transmission distance of the floating object from the collection end to the discharge end by the transmission linear velocity, and is usually between 30 seconds and 120 seconds.

[0051] Considering that the transmission linear velocity is used as the denominator in the division operation, the logic control module has a preset lower limit protection value for the linear velocity. When the motor stalls and causes the linear velocity to approach zero, the system automatically outputs an alarm command and terminates the division operation to avoid a division-to-zero crash in the underlying program. The physical purpose of this constraint is to overcome the illusion of backflow caused by the rapid emergence of new floating debris after a localized cleanup of the water surface, ensuring that the mechanical mechanism operates for sufficient time to complete the material transfer across the entire cross-section of the tank.

[0052] The logic control module will cut off the output level and complete a closed-loop control with high adaptability only when all three of the above criteria are met simultaneously.

[0053] Based on the above overall system architecture, refer to the appendix Figure 1In this embodiment, the floating debris collection mechanism, as the core component of the mechanical execution unit, is used to remove impurities from the water surface and transfer them outwards. Structurally, this mechanism mainly includes a collection roller 4, a filter adhesion belt 5, a flexible scooping scraper 6, and an anti-entanglement and desorption component. Through the spatial coordination of these multiple components, the above structure can adapt to the on-site working conditions of retrieval and separation of multiphase mixed solid materials inside the wet slag remover.

[0054] The two ends of the collecting roller 4 are rotatably fixed across the sides of the wet slag removal machine trough 1. For the bearing supports and waterproof sealing structures at both ends of the collecting roller 4, those skilled in the art can use industrial standard mounted spherical bearings and end-face mechanical seal assemblies, which will not be elaborated further here. Considering the high steam temperature and corrosive media in the operating environment, the base of the collecting roller 4 is made of passivated alloy stainless steel to ensure structural rigidity during long-term operation.

[0055] To achieve simultaneous initial separation during the salvage process, the filter adhesion belt 5 is wrapped and fixed to the outer circumferential surface of the collecting roller 4.

[0056] As a preferred alternative implementation, the filter adhesion belt 5 is a porous, heat-resistant rubber belt with a specific surface roughness. In this embodiment, the surface roughness Ra value of the belt is preferably between 12.5 μm and 25 μm to provide sufficient microscopic physical contact area. This material can adhere to fine floating objects such as lightweight fly ash by relying on physical surface tension upon contact with the water surface. The filter adhesion belt 5 has perforated holes arranged in an array. During the stroke of the collection mechanism rotating and lifting out of the water surface, the free water carried out by the material will preferentially collect along the perforated holes and drip back into the slag removal machine tank, thus completing the initial gravity solid-liquid separation at the beginning of collection.

[0057] To address the technical challenge of removing large, coked slag chunks or long, strip-shaped foreign objects solely through surface adhesion, multiple flexible scooping scrapers 6 are provided. These flexible scooping scrapers 6 are spaced apart along the axial direction of the collecting roller 4 on the outer surface of the filter adhesion belt 5. To avoid excessive instantaneous impact load caused by multiple scrapers simultaneously cutting into the water surface, the flexible scooping scrapers 6 are arranged in a staggered, circumferential pattern on the collecting roller 4. The flexible scooping scrapers 6 are made of polyurethane elastomer, and their ends rotate with the collecting roller 4, alternately extending below the water surface to scoop up floating debris.

[0058] Based on the general principles of resistance calculation in fluid mechanics, the flexible scraper 6 inevitably encounters water resistance when it operates underwater and cuts into the liquid surface. To avoid the mechanism's operation causing violent turbulence in the water surface and generating false secondary liquid level signals, the system design strictly constrains the rotational dynamic parameters of the collecting roller 4. The hydrodynamic resistance experienced by the scraper is calculated using the following theoretical derivation formula: ; In this formula, This represents the hydrodynamic resistance experienced by the flexible scooping scraper 6 when it rotates underwater; The dimensionless fluid resistance coefficient corresponding to a specific shape of the scraper; The density of the water in the working condition of tank 1 of the wet slag remover; This represents the effective water-facing projected area of ​​the flexible scooping scraper 6 along the direction of motion at the current cutting angle. This represents the rotational linear velocity at the end of the flexible scooping scraper 6.

[0059] Since hydrodynamic resistance is directly proportional to the square of the linear velocity, to avoid surface instability caused by a nonlinear increase in resistance, the preset rotational linear velocity is strictly limited by the hardware frequency converter to the range of 0.1 m / s to 0.3 m / s. This range was derived by comprehensively balancing the amount of floating debris to be removed and the upper limit of the critical kinetic energy required to maintain a stable water surface pressure.

[0060] The anti-entanglement and desorption component is fixedly arranged on the discharge side. The working end of the anti-entanglement and desorption component faces the surface of the collecting roller 4 and is set close to the movement trajectory of the flexible scooping scraper 6. It is used to forcibly peel off the floating objects that are attached to the filter adhesion belt 5 and the flexible scooping scraper 6.

[0061] As a specific substructure of this general feature, the anti-winding and desorption component can be configured as an anti-winding paddle or an anti-winding brush. During the mechanical assembly stage, a small assembly gap of 2mm to 5mm is maintained between the working end of the anti-winding and desorption component and the surface of the filter adhesion belt 5. This gap is set to ensure that the hard scraping end does not directly scrape the rubber belt to prevent wear and damage caused by long-term operation, while also ensuring that its interference depth is sufficient to cut off the capillary bridge formed between the wet floating material and the belt surface, thereby achieving the desorption of viscous materials.

[0062] Based on the aforementioned three-dimensional physical architecture, the floating object collection mechanism performs continuous salvage operations after receiving a start command. The specific operating principle is broken down into the following steps.

[0063] The flexible scooping scraper 6 cuts into the liquid surface at a specific entry angle, pushing and lifting the lumpy floating matter gathered on the upper layer of the liquid surface into the angled area formed by the scraper and the belt. In the parallel stage of this mechanical pushing process, the surface of the filter adhesion belt 5 comes into contact with the water, simultaneously capturing the floating light ash bubbles.

[0064] As the main shaft continues to rotate, the flexible scraper 6 carrying various types of floating matter and the filter adhesion belt 5 flip upwards and detach from the liquid surface. In the transport section exposed to air, excess water mixed in with the material accumulation is drawn downwards by gravity. The water is drained back into the water tank through the permeable mesh on the filter adhesion belt 5, while the solid floating matter intercepted on the outside of the belt continues to transfer towards the discharge side with the rollers.

[0065] As the collecting roller 4 rotates past its highest point and enters the downward section on the discharge side, most of the loose floating debris naturally falls off under the influence of gravity. Some stubborn material, adhering to the belt surface due to its own adhesiveness, is mechanically blocked and sheared by the working end of the anti-winding and desorption component when it rotates to the area where the component is located. The anti-winding and desorption component forcibly peels off the floating debris remaining on the filter adhesion belt 5 and the flexible scooping scraper 6. The desorbed material loses its support and falls freely onto the guide surface of the lower conveying mechanism, avoiding the equipment hazard of flexible impurities accumulating and entangled on the rotating shaft system.

[0066] Based on the physical unloading trajectory of the aforementioned floating debris collection mechanism, the conveying mechanism needs to form a seamless material transfer with it in space in order to receive and transfer the stripped impurities outward.

[0067] In this embodiment, the conveying mechanism is arranged on one side of the wet slag removal machine tank 1, and its overall shape is inclined outward and downward. The conveying mechanism mainly includes a guide channel 7 and a moisture filter screen 8, which work together to achieve passive gravity discharge of floating objects and secondary solid-liquid separation.

[0068] The guide channel 7 forms the basic framework of the conveying mechanism. The feed end of the guide channel 7 is flared and is fixedly suspended directly below the discharge side of the floating material collection mechanism to collect free-falling desorbed material. In order to utilize the Earth's gravitational field to achieve zero-power material transfer and avoid energy consumption and failure points caused by additional drive motors, the guide channel 7 extends downward at a set angle from the feed end.

[0069] Based on classical Newtonian mechanics and the theory of inclined plane friction, the dynamic characteristics of the floating material sliding down the guide channel 7 directly determine the solid-liquid separation effect. If the material slides down too quickly, free water will not have enough time to penetrate the filter screen; if the material slides down too slowly or stagnates, it will lead to severe accumulation and blockage at the discharge end. Therefore, the system design introduces a theoretical derivation model of material residence time. The forced sliding residence time of the material on the inclined plane is calculated using the following mathematical relationship: ; In this formula, This represents the dwell time of the floating object as it slides down the guide channel 7; This represents the effective sliding surface length of guide channel 7; Represents gravitational acceleration; This represents the installation angle between the guide channel 7 and the horizontal plane; This represents the combined sliding friction coefficient between the wet floating matter and the surface of the guide channel 7. The physical purpose of this calculation logic is to constrain a suitable residence time range by matching the sliding length and the installation tilt angle, ensuring that the material has a sufficient time window to release the internally entrained free moisture during the sliding process.

[0070] Considering the division and square root operations in the above calculation formula, in order to ensure the validity of the physical model and avoid dead-point stagnation caused by the denominator approaching zero, the installation angle of the guide channel 7 must meet the following requirements. The absolute boundary conditions are used to ensure that the denominator is always positive. As a preferred method, the friction coefficients of fly ash and coke slag under wet conditions (usually...) are comprehensively considered. (fluctuating between 0.4 and 0.6) In this embodiment, the installation angle of the guide channel 7 is strictly defined between 35° and 50°. This range overcomes the critical static friction to maintain smooth material discharge, while avoiding insufficient residence time due to an excessively steep angle.

[0071] To achieve deep dehydration, a moisture filter screen 8 is embedded in the bottom receiving surface of the guide channel 7. The lower end of the moisture filter screen 8 is suspended above the water surface inside the wet slag removal machine tank 1, forming a water return channel.

[0072] After the automatic cleaning operation is triggered by the control module, the operation of the conveying and solid-liquid separation mechanism is specifically manifested in the following steps.

[0073] Under the mechanical peeling action of the anti-entanglement and desorption components, the mixed floating matter carrying moisture is peeled off from the collecting roller 4 and falls to the flared feed end of the guide channel 7 under the action of gravity. The material impacts the surface of the moisture filter screen 8, converting part of the vertical falling kinetic energy into the initial kinetic energy of sliding down at an inclined angle. Under the impact force at the moment of impact, the loose structure inside the material clump is destroyed, and the free water wrapped on the surface is squeezed out for the first time.

[0074] Relying on the difference between the forced gravity component and the inclined plane friction, the mixed material slides outward at a uniform acceleration along the surface of the moisture filter screen 8. During the sliding residence time, the material continuously tumbles and rubs, and the water it carries passes through the screen gaps of the moisture filter screen 8 under the pull of gravity. The filtered free water gathers into streams, flows along the bottom of the filter screen, and drips directly back into the interior of the wet slag removal machine tank 1, avoiding environmental pollution and heat loss from external drainage. The intercepted pure solid floating matter continues to slide to the discharge end of the guide channel 7 and is finally discharged into the external waste collection bin, completing the entire slag removal and dewatering closed loop.

[0075] In floating debris collection and transportation operations, the moving parts of the equipment inevitably face the problem of increased load caused by the gradual adhesion of sticky impurities. In this embodiment, a drive and self-cleaning linkage mechanism is used to provide the core rotational power and automatically perform desorption intervention when abnormal operating conditions are detected. At the hardware level, this mechanism mainly includes a variable frequency drive component, a high-pressure spray pipeline 9, and a linkage diagnostic module based on a neural network.

[0076] For the selection of motors and reducers for the frequency converter drive components, those skilled in the art can use standard industrial-grade AC asynchronous motors paired with planetary gear reducers. The basic electromechanical energy conversion and speed reduction / torque increase principles are well-known technologies in this field and will not be elaborated here. The high-pressure spray pipe 9 is arranged parallel to the side and above the collecting roller 4, and the water inlet end of the pipe is controlled by a solenoid valve.

[0077] Based on the general technical principles of electromechanical system dynamics, when severe material entanglement occurs on the surface of the filter adhesion belt 5, the rotational resistance torque of the main shaft will increase significantly, leading to distortion of the motor operating parameters. Therefore, a rotational dynamics monitoring model is established for the system, and its basic torque balance equation adopts the following mathematical relationship: ; In this formula, Represents the electromagnetic torque output by the variable frequency drive component; This represents the combined load torque borne by the collecting roller 4 during underwater propulsion and material retrieval; The equivalent moment of inertia of the entire rotating axis system; This represents the derivative of the rotational angular velocity with time, i.e., angular acceleration; Represents the viscous friction coefficient of bearings and transmission components; This represents the real-time rotational angular velocity. The physical meaning of this formula is that it decouples the motor's output torque into three parts: steady-state load, inertial impact, and viscous loss, thereby revealing the underlying causal relationship between sudden load changes and the resulting surge in torque.

[0078] Considering the randomness of water surface fluctuations in industrial settings, relying solely on static torque thresholds can easily lead to frequent malfunctions such as equipment start-stop. As a preferred approach, a lightweight one-dimensional convolutional neural network (1D-CNN) model is deployed within the linkage diagnostic module. This model is used to deeply mine the temporal fluctuation characteristics of motor operating parameters and accurately identify the actual winding and stalled state. Specifically, the construction and operation of this model includes the following steps.

[0079] The motor stator current, real-time output torque, and spindle speed are selected as input parameters, as these three directly map the electromechanical coupling load state of the transmission shaft system in physics. These parameters, within a set time window (preferably sampled at 50Hz within the past 2 seconds in this embodiment), form a 3×100 time series matrix. Before being input into the network, the raw data needs to undergo Z-score normalization preprocessing. ; In this formula, This represents standardized data; This represents the original sensor value at the current sampling point; This represents the sample mean of the data within that time window. This represents the sample standard deviation of the data within that time window. Considering the completeness of the division operation, Set to 10 -5 The tiny constant, the technical purpose of which is to reduce the standard deviation when the system is in an absolutely no-load constant-velocity state. When the value approaches 0, it prevents the underlying calculation program from crashing due to division by zero.

[0080] The internal hierarchical structure of the model is clearly described. This 1D-CNN model consists of an input layer, a one-dimensional convolutional layer, a max-pooling layer, and a fully connected layer connected in series. Data flows from the input layer, where the one-dimensional convolutional layer uses a 1×5 sliding convolutional kernel to extract local abrupt changes in current and torque. The max-pooling layer performs dimensionality reduction to retain the most obvious load impact response. Finally, the data enters the fully connected layer and is processed by the sigmoid activation function. The specific business meaning of the output is to collect the probability value of severe material entanglement on roller 4, with a value ranging from 0 to 1.

[0081] To ensure that those skilled in the art can fully reproduce the model, its offline training steps are disclosed here. The model's sample data comes from historical sensor records of similar salvage equipment under normal operation and simulated entanglement conditions. Labels are defined as binary discrete values: 0 represents normal operation, and 1 represents entanglement and stalling. The training process uses a binary cross-entropy loss function for backpropagation. ; In this formula, This represents the calculated loss value; This represents the total number of samples in a single training batch. Representing the The true classification label of each sample; This represents the entanglement probability predicted by the model. Similarly, the bias constant... The introduction of this loss function avoids the mathematical singularity deadlock that occurs when the logarithmic function has zero input. Based on this loss function, gradient descent optimization leads to convergence of the model weights during iteration.

[0082] The output result is determined based on multi-dimensional weighted logic to avoid erroneous operations caused by the failure of a single algorithm. When the entanglement probability output by the model is greater than the preset judgment threshold (set to 0.85 in this embodiment, determined by the F1-Score of precision and recall) for three consecutive calculation cycles, and the preset multiple is limited by the short-term overload safety margin allowed by the drive component, in this embodiment, when the steady-state current fed back by the frequency converter drive component is greater than 1.2 times the rated current, the linkage diagnostic module determines that the self-cleaning trigger condition is met.

[0083] Upon triggering, the linkage diagnostic module sends a reverse intervention command to the frequency converter drive component, causing the collecting roller 4 to rotate in reverse within a set time of 3 to 5 seconds, using the reverse water flow resistance to loosen the surface entanglement. Immediately afterwards, the logic level drives the solenoid valve to open, and industrial-grade fire-fighting water or recycled water in the high-pressure spray pipe 9, pressurized by the system pump, is atomized and sprayed through an array of conical nozzles. The kinetic energy of the high-speed fluid particles directly acts on the surface of the filter adhesion belt 5, overcoming the capillary adhesion between the damp fly ash and the rubber belt. The stubborn impurities forcibly removed fall back into the wet slag removal machine tank 1 with the flushing water flow. After a 15-second flushing cycle, the solenoid valve is de-energized and closed, the frequency converter drive component resumes forward operation, the automatic cleaning device resets its internal state variables, and seamlessly transitions to regular retrieval operations.

[0084] Based on the aforementioned single-machine mechanical architecture and underlying closed-loop logic, the wet slag removal machine tank 1 typically has a physical span of tens of meters in large thermal power generating units. Since the mechanical slag removal coverage radius of a single automatic cleaning device has an objective geometric boundary, in this embodiment, to achieve efficient cleaning of the entire water section without blind spots, the system deploys multiple identical automatic cleaning devices in an array along the length of the tank. These multiple devices establish bidirectional communication with the central controller via an industrial fieldbus, jointly executing a multi-machine collaborative automatic cleaning method that includes physical actions and global scheduling. This method integrates the independent control logic of a single machine with the dynamic load balancing of a global cluster, specifically including the following steps: In a distributed hardware architecture, the underlying microprocessor of each automatic cleaning device independently collects the apparent water level signal and the actual water level signal within its physical defense zone. Considering the physical clock drift of the quartz crystal oscillators of each node in the distributed computing architecture, directly aggregating unaligned data would lead to timing misalignment in the global state assessment. To eliminate this time-domain error and ensure that the data is within the same physical condition slice, the central controller periodically broadcasts high-precision time synchronization messages to downstream nodes. For clock synchronization protocols in industrial Ethernet environments (such as the IEEE 1588PTP protocol), those skilled in the art can perform conventional settings according to the network topology; its underlying timestamp distribution and delay compensation mechanisms are well-known technologies in the field and will not be elaborated here. Each node device aligns high- and low-frequency sensor data under a unified global time reference and calculates the local liquid level deviation characterizing the degree of floating debris accumulation in the corresponding defense zone.

[0085] Under high-load operation conditions where multiple defense zones simultaneously meet the activation conditions, if all automatic cleaning devices are started at full load at the same time, the transient starting current of the motor group will cause a sudden drop in the voltage of the power grid in the plant area. Furthermore, the simultaneous cutting of water by multiple sets of collection rollers can easily trigger hydrodynamic interference and mechanical resonance in the water within the tank. Based on the general technical principles of linear algebra and optimal allocation, a spatial state coupling model is constructed within the central controller to calculate the activation priority and resource allocation weight of each device. The allocation weight is solved using the following matrix operation formula: ; In the above formula, This represents the weights that indicate the operating priority of each device. dimensional column vector (where (Total number of automatic cleaning devices). Represents the construction 3D space coupling matrix; Represents the spatial coupling matrix The transpose of the matrix; The representation is composed of the urgency level of each defense zone. Urgency column vector; represent A standard identity matrix of 3D; This represents the preset regularization coefficient.

[0086] The reason for choosing physical distance and local liquid level deviation as key input parameters is that physical distance directly maps the interference intensity of water fluctuations during the salvage of adjacent equipment, while liquid level deviation objectively reflects the urgency of on-site cleanup needs. Specifically, the elements within the spatial coupling matrix are based on the defense zone. With defense zone The interference coefficient is generated by mapping the inverse decay function of the physical distance between them; the smaller the distance, the larger the interference coefficient. The elements inside the urgency column vector are generated by the first... The local liquid level deviation of the defense zone is calculated by weighting the cumulative standby time proportionally. This output logic based on multi-dimensional parameter matrix mapping effectively avoids the drawbacks of one-sided scheduling caused by relying solely on the highest water level at a single point.

[0087] The physical purpose of this operational logic is to comprehensively evaluate the fluid interference effects of adjacent equipment and smoothly allocate power resources and salvage operations to the most urgent defense zones, achieving staggered salvage. Considering extreme symmetrical conditions (e.g., the floating object status and liquid level data from all defense zones are completely consistent), the column vectors of the spatial coupling matrix will exhibit a high degree of linear correlation, causing the target matrix to suffer rank deficiency and exhibit singularity. Singular matrices do not mathematically have an inverse. To ensure the absolute completeness of the underlying algorithm and avoid stack crashes during matrix inversion operations, a penalty term based on the identity matrix is ​​introduced in the above formula.

[0088] As a preferred approach, the regularization coefficient is strictly defined between 0.01 and 0.05. This value ensures that the synthesized matrix is ​​strictly positive definite and invertible, while preserving the physical distribution characteristics of the original weights to the greatest extent possible.

[0089] Based on the dynamic priority scheduling derived from the matrix operations described above, the central controller issues start commands to the automatic cleaning devices with higher priority in descending order of the output weight vectors. The activated automatic cleaning device drives its collection roller 4 to cut into the water surface at a set speed, relying on the mechanical cooperation of the filter adhesion belt 5 and the flexible scooping scraper 6 to complete the combined retrieval of floating debris. During continuous rotation cycles, impurities carrying moisture fall into the conveying and solid-liquid separation mechanism under the interference of the anti-entanglement and desorption components, and are dehydrated and transferred outward by gravity through the inclined guide of the water filter screen 8. During this physical transfer process, the actual water level and liquid level deviation in the local defense zone are continuously monitored, serving as a closed-loop feedback for dynamically evaluating the retrieval progress.

[0090] During the ongoing cleaning operation, when a single operating automatic cleaning device detects that the liquid level deviation within its protected zone is less than a preset threshold and the actual water level is within the safety zone, the system enters a shutdown verification phase. To prevent the equipment from malfunctioning due to momentary low sensor values ​​caused by water surface ripples, the control program mandates that the liquid level deviation must be maintained at a minimum physical delay window over time. This minimum physical delay parameter is mathematically defined as the physical length (i.e., characteristic length) covered by a single automatic cleaning device within its protected zone along the extension direction of the slag removal machine trough, divided by the real-time underwater linear velocity of the collection roller 4. To ensure the robustness of the algorithm, when the linear velocity fed back by the underlying frequency converter is below the lower limit dead zone of 0.05 m / s, the division operation is immediately bypassed, and the delay window is directly assigned the system's preset maximum safety constant (120 seconds in this embodiment) to ensure the continuity of the program logic chain and prevent division by zero overflow. After the above-mentioned multi-dimensional combination of shutdown conditions are met, the node equipment stops the main shaft rotation, reports the status word that the defense zone has been cleaned to the central controller, and releases the global scheduling token for subsequent downgraded equipment to take over operation, ultimately achieving steady-state cleaning of the entire water area by the slag removal machine.

[0091] To further verify the engineering effectiveness of this application, a detailed description is provided below with reference to a specific industrial application embodiment and its experimental comparison data.

[0092] In this embodiment, the application scenario is selected as the bottom wet slag removal system of a megawatt-class coal-fired power generation unit. The total length of the slag removal tank is 42 meters. Due to the physical coverage boundary in the large-scale space, the system deploys four automatic cleaning devices (numbered as nodes 1 to 4) at equal intervals along the length of the tank. The central controller establishes 100-megabit bidirectional communication with each node via the Profinet industrial fieldbus.

[0093] See attached document Figure 4 and attached Figure 5 To verify the anti-interference capability of heterogeneous sensor fusion, this experiment recorded the data stream of node 3 under severe slag discharge conditions for 120 consecutive minutes. Figure 4 In the upper subplot, the horizontal axis represents time, and the vertical axis represents the liquid level. The dashed curve represents the apparent water level (Ho) measured by a traditional ultrasonic level gauge, while the solid curve represents the true water level (Hr) measured by an infrared dual-beam level gauge. As time progresses, porous, lightweight ash gradually accumulates in the tank, causing the apparent water level to exhibit a significant false upward trend due to the physical lifting of the acoustic wave diffuse reflection interface, reaching as high as above the overflow warning line; while the true water level curve remains consistently within the standard replenishment range. Figure 5In the figure, the solid line represents the liquid level deviation calculated in real time by the logic control module. When this deviation crosses the preset 100mm deviation allowable threshold at the 45-minute mark, the control program precisely issues a start command. The curve then shows that within the following 15 minutes of the cleaning mechanism's intervention, the apparent water level rapidly drops and approaches the actual water level, with the liquid level deviation converging to near the zero dead zone. This experimental data intuitively and fully demonstrates that the triggering logic based on dual-source deviation extraction can completely avoid the false alarm problem when a single sensor is exposed to water surface cover.

[0094] See attached document Figure 6 This experiment aims to verify the technical contribution of the matrix regularization multi-machine collaborative algorithm to global load balancing. The experiment simulated an extreme symmetrical operating condition where four defense zones simultaneously reached their highest urgency levels. In the traditional control group operating mode with simultaneous startup, the high-power drive motors of the four automatic cleaning devices were fully loaded and engaged in the water surface within the same clock cycle, causing the transient starting current of the local low-voltage busbar in the plant area to surge to 4.5 times its rated value. Furthermore, the water in the tank was stimulated to produce severe mechanical resonance, resulting in a high peak value for the water surface fluctuation variance. In contrast, in the experimental group applying the algorithm of this embodiment, the central controller constructed a spatial coupling matrix based on physical distance and liquid level deviation. Under the mathematical intervention of the Tikhonov regularization term, the algorithm successfully broke the matrix singularity deadlock caused by the symmetrical operating condition, outputting a set of descending weight vectors with reasonable time gradients. The drive commands were smoothly and staggered, significantly reducing the global maximum grid impact current, and the water surface fluctuation variance was always suppressed within a safe threshold. The experimental results confirm that the matrix scheduling logic provided in this application not only has rigorous underlying mathematical completeness, but can also realize power resource allocation in real macroscopic multi-machine interference scenarios.

Claims

1. A floating matter automatic cleaning device for a wet-type slag conveyer, characterized by, The automatic cleaning device is installed inside the wet slag removal machine tank (1), and the automatic cleaning device includes: The sensing unit includes an original water level sensor (2) installed inside the wet slag removal machine tank (1) for detecting the apparent water level and a real water level sensor (3) installed on the side wall of the wet slag removal machine tank (1) for penetrating the floating debris layer to detect the real water level. The control decision unit is used to receive the detection data of the apparent water level and the actual water level in real time, and output start and stop control commands based on the comparison results of the detection data. A mechanical execution unit is arranged in the space above the wet slag removal machine tank (1). The mechanical execution unit includes a floating object collection mechanism that is horizontally connected and erected above the water surface, a conveying mechanism arranged on one side of the tank and directly below the discharge side of the floating object collection mechanism, and a drive mechanism that is coaxially connected to the main shaft of the floating object collection mechanism. The electrical control terminal of the drive mechanism is electrically connected to the output terminal of the control decision unit. The drive mechanism is used to receive the start / stop control command and drive the floating object collection mechanism to perform rotational collection operations.

2. A device for automatic cleaning of floating matter from a wet-type slag conveyer according to claim 1, characterized in that, The floating object collection mechanism includes a collection roller (4) and an anti-entanglement and desorption component. The two ends of the collection roller (4) are rotatably connected to the two sides of the wet slag removal machine tank (1). A filter adhesion belt (5) is fixedly connected to the outer circumferential surface of the collection roller (4). The filter adhesion belt (5) has permeable mesh holes arranged in an array. Multiple flexible scooping scrapers (6) are provided on the outside of the collection roller (4). The multiple flexible scooping scrapers (6) are spaced out along the axial direction of the collection roller (4) on the outer surface of the filter adhesion belt (5). The ends of the flexible scooping scrapers (6) rotate with the collection roller (4) and alternately extend below the water surface to scoop up the floating objects. The anti-entanglement and desorption component is fixedly arranged on the discharge side. The working end of the anti-entanglement and desorption component faces the surface of the collecting roller (4) and is set close to the movement trajectory of the flexible scooping scraper (6) to peel off the floating objects that are attached to the filter adhesion belt (5) and the flexible scooping scraper (6).

3. A device for automatic cleaning of floating matter from a wet-type slag conveyer according to claim 1, characterized in that, The feed end of the conveying mechanism is fixed directly below the discharge side of the floating object collection mechanism, and is used to receive the floating objects falling from the discharge side and guide the floating objects outward to the outside of the wet slag removal machine tank (1). The conveying mechanism includes a guide trough (7) and a moisture filter (8). The feed end of the guide trough (7) is located directly below the anti-entanglement and desorption component. The guide trough (7) is arranged at an inclination, and the end of the guide trough (7) leads to a collection box located outside the wet slag removal machine tank (1). The moisture filter (8) is laid on the bottom receiving surface of the guide trough (7) to filter the moisture carried by the floating objects back into the wet slag removal machine tank (1).

4. A device for automatic cleaning of floating matter from a wet type slag conveyer according to claim 2, characterized in that, The real water level sensor (3) is an infrared dual-beam level gauge; The drive mechanism includes an integrated motor and reducer with a waterproof housing that are coaxially connected to the collecting roller (4); The automatic cleaning device also includes a high-pressure spray pipe (9) electrically connected to the control decision unit. The high-pressure spray pipe (9) includes a side-above arrangement parallel to the collecting roller (4), and the high-pressure spray pipe (9) is provided with an array of conical nozzles.

5. A method for automatically cleaning floating matter of a wet-type slag conveyer, characterized by, An automatic floating debris cleaning device for a wet slag remover according to any one of claims 1-4 includes the following steps: The sensing unit performs real-time monitoring, and the control decision unit has a timestamp alignment program that uses the period of the signal with a lower sampling frequency as a reference to perform downsampling time slice alignment on the high-frequency acquired signal. The control decision unit performs liquid level deviation calculation and start-up condition determination. It calculates the absolute value of the difference between the apparent water level signal and the actual water level signal as the liquid level deviation, and determines whether the current operating condition meets the start-up condition based on the liquid level deviation and the system running time parameters. When the start-up conditions are met, the mechanical actuator performs the floating object collection and discharge operation. The floating object collection mechanism is partially immersed in the water surface and rotates continuously at a constant speed. It captures the floating objects through a combination of scooping and physical surface tension and transfers them to its discharge side. The conveying mechanism receives the detached floating objects and discharges them through gravity guidance. The control decision unit performs stop condition verification and system reset actions. When the stop conditions are met simultaneously, the recalculated liquid level deviation is less than the preset allowable deviation threshold, the actual water level is greater than or equal to the standard safe water level, and the current continuous running time of the automatic cleaning device since receiving the start command is greater than or equal to the preset minimum shutdown delay, a stop control command is output.

6. A method of automatic cleaning of floating matter in a wet type slag conveyer according to claim 5, wherein The output of the determination of whether the start-up condition is met is constructed as a parallel triggering network containing three dimensions. The start-up condition is determined to be met if and only if one of the following three conditions occurs: The calculated liquid level deviation is greater than or equal to the preset allowable deviation threshold; The actual water level is lower than the preset minimum safe water level warning line; The system has been in standby monitoring mode for an extended period of time, reaching the preset cyclic waiting period. The allowable threshold for deviation is determined and limited based on the physical overflow tolerance of the wet slag removal machine tank (1) and the inherent error band weighted by the original water level sensor (2) and the real water level sensor (3). The minimum safe water level warning line is limited by the water seal safety height of the wet slag removal machine tank (1); The cyclic waiting period is defined by the average generation rate of floating objects under historical operating conditions.

7. A method of automatic cleaning of floating matter in a wet type slag conveyer according to claim 5, wherein The minimum downtime is dynamically calculated based on the proportional mapping relationship between the physical transmission distance and the transmission linear speed. To prevent abnormalities in the underlying program, the control decision unit has preset a lower limit protection value for the transmission linear speed. When the transmission linear speed is detected to be lower than the lower limit protection value, the system terminates the division operation logic and outputs an alarm command. The lower limit protection value for linear velocity is defined to prevent division-by-zero crashes in the underlying program.

8. The automatic cleaning method for floating debris in a wet slag remover according to claim 5, characterized in that, It also includes an execution-driven and self-cleaning linkage mechanism, the specific steps of which are as follows: Using a lightweight one-dimensional convolutional neural network model deployed within the linkage diagnostic module, dynamic feature acquisition and standardized preprocessing are performed on the collected motor stator current, real-time output torque, and spindle speed, and a small constant is introduced into the denominator constraint of the standardized preprocessing process. The tiny constant is defined to prevent division by zero errors in the calculation. When the entanglement probability output by the lightweight one-dimensional convolutional neural network model is greater than the preset judgment threshold for multiple consecutive calculation cycles, and the steady-state current fed back by the frequency conversion drive component is greater than the preset multiple of the rated current, the linkage diagnosis module determines that the self-cleaning trigger condition is met and issues a reverse intervention command to perform forced desorption and rinsing. The determination threshold is determined and limited by the F1-Score evaluation of precision and recall. The preset multiple is limited by the short-term overload safety margin allowed by the drive component.

9. The automatic cleaning method for floating debris in a wet slag remover according to claim 5, characterized in that, It also includes a multi-machine collaborative automatic cleaning process for global load balancing, which is jointly executed by multiple automatic cleaning devices deployed in an array along the length of the tank. The multi-machine collaborative automatic cleaning process specifically includes: During the multi-region data synchronization phase, the central controller periodically broadcasts time synchronization messages to downstream nodes, enabling each node device to align sensor data under a unified global time reference and calculate the local liquid level deviation that characterizes the degree of floating debris accumulation in their respective physical defense zones. During the collaborative scheduling and allocation phase, the central controller constructs a spatial state coupling model to comprehensively evaluate the impact of fluid interference and calculate the start-up priority and resource allocation weight of each device. Then, based on the descending order of the output weight vector, it sequentially issues start-up commands to the automatic cleaning devices with higher priority.

10. A method of automatic cleaning of floating matter in a wet type slag conveyer according to claim 9, wherein The specific scheduling logic of the internal spatial state coupling model of the central controller is as follows: A spatial coupling matrix is ​​generated based on the physical distance feature mapping between each automatic cleaning device, and an emergency level column vector is constructed by proportionally weighting the local liquid level deviation and cumulative standby time of each physical defense zone. In the underlying derivation equation for jointly solving the resource allocation weights using the spatial coupling matrix and the urgency column vector, a penalty term based on the standard identity matrix is ​​introduced as a regularization coefficient to maintain the mathematical boundary constraints of the synthesized matrix in the operation process, which are strictly positive definite and invertible.

Images