A sludge drying process control system based on coating backmixing

Abstract

The present invention relates to the field of control and regulation technology, and in particular to a sludge drying process control system based on coating backmixing. The present invention deploys a corrosion-resistant micro fiber grating sensor array to capture material deformation stress data in real time, and outputs dynamic viscosity parameters through a conversion function; a temperature gradient monitoring module reconstructs the spatial thermal field distribution; and a flow synchronization module generates a dynamic material balance report. The plastic risk analysis module integrates multi-source data. The multi-agent decision module analyzes risk signals, and the Nash equilibrium arbitrator dynamically allocates weights to generate a Pareto optimal instruction set; the equipment collaboration module executes distributed voting to verify the feasibility of instructions. The pulse neural network chip of the safety interlock module implements triple modulation descaling control. The present invention constructs a closed-loop optimization chain of perception, analysis, decision-making and execution to achieve adaptive and precise control of the backmixing ratio, thereby improving drying efficiency and operational stability.

Description

Technical Field

[0001] The present invention relates to the field of control and regulation technology, and in particular to a sludge drying treatment process control system based on coating backmixing. Background Art

[0002] In the field of municipal sludge treatment, coating backmixing technology can quickly reduce the moisture content of the mixed material to below the plastic stage by backmixing high-moisture raw sludge with dried particles, breaking the viscous barrier formed by the sludge colloidal structure, and avoiding scaling of the vessel wall and uneven drying caused by organic matter adhesion during the drying process; the granulated material formed by backmixing is evenly heated by mechanical turning on the indirect heating disc, and maintains a loose physical form during the step-by-step removal of moisture, simultaneously reducing heat consumption and eliminating the risk of internal clumps, thereby improving drying efficiency and product stability, and meeting the process requirements for the physical and chemical properties of sludge for subsequent incineration or resource utilization.

[0003] In the field of automatic control, the technical pain points of the sludge drying process stem from the fact that the dynamic characteristics of the moisture content range in the plastic stage have not been effectively modeled and feedback-controlled. Specifically, when the sludge moisture content is in the range of 60% to 40%, its rheological characteristics show a nonlinear mutation, and the viscosity coefficient increases exponentially. However, the existing control system lacks real-time monitoring and adaptive adjustment mechanisms for material viscosity, and is unable to accurately control the dry powder remixing ratio, resulting in the mixed material failing to remain continuously and stably below the plastic critical point. For example, during the operation of the drying equipment, if the viscosity sensor is missing or the control algorithm is not embedded in the material rheological model, the system will find it difficult to dynamically respond to the instantaneous changes in sludge viscosity. At this time, the insufficiently broken sticky sludge clumps will adhere to the heat exchange surface to form a thermal resistance layer, causing the heat transfer efficiency to decrease and local overheating, and ultimately leading to coking and blockage inside the dryer. Summary of the Invention

[0004] In view of the shortcomings of the existing technology, the present invention provides a sludge drying process control system based on coating backmixing, which solves the problems of lack of real-time viscosity monitoring and inaccurate backmixing ratio adaptive control caused by the nonlinear rheological characteristics of the plastic zone during the sludge drying process.

[0005] In order to solve the above technical problems, the specific invention of the present invention is as follows:

[0006] The present invention provides a sludge drying treatment process control system based on coating backmixing, comprising:

[0007] The rheological property sensing module is deployed at the outlet of the coating machine. It obtains material deformation and stress data through a micro fiber grating sensor array and converts the material deformation and stress data into viscosity parameters.

[0008] The temperature gradient monitoring module is installed on the multi-layer structure surface of the dryer disc layer. It collects temperature distribution time series data through a 5cm×5cm grid thin film thermocouple matrix and reconstructs the spatial thermal field distribution using a spatial interpolation algorithm.

[0009] The flow synchronization module connects the wet sludge input pipeline and the dry powder delivery pipeline, records the instantaneous mass flow of sludge and dry powder in real time and generates a dynamic material balance report;

[0010] The plasticity risk analysis module receives the viscosity parameters from the rheological properties perception module, the spatial thermal field distribution from the temperature gradient monitoring module, and the dynamic material balance report from the flow synchronization module. It solves the viscosity-temperature coupling equation through a physical information neural network to output a moisture content prediction value, triggering the quantum annealing processor to perform moisture content inverse optimization and generate a plasticity risk level signal.

[0011] The multi-agent decision-making module analyzes the plastic risk level signal from the plastic risk analysis module, calculates the dry powder addition ratio correction through the back-mixing regulation agent, generates the thermal oil valve opening array through the thermal field balance agent, and sets the inert gas flow instruction through the safety protection agent. The outputs of the back-mixing regulation agent, thermal field balance agent, and safety protection agent are transmitted to the Nash equilibrium arbitrator to generate the Pareto optimal control instruction set.

[0012] The equipment collaboration module receives the Pareto optimal control instruction set from the multi-agent decision module for distributed voting verification. When the thermal oil pump node confirms that the valve opening array is compatible with the equipment status and the screw feeder node verifies that the dry powder ratio correction is feasible, it executes the weighted voting consensus mechanism to output the equipment execution instruction.

[0013] The safety interlock module, including a pulse neural network chip, monitors the plasticity risk level signal and moisture content prediction value in real time. If three consecutive L5 risk signals are received or the moisture content prediction value suddenly changes by more than 35%, it overrides the conventional instructions and triggers the layered material reduction program.

[0014] The viscosity parameters output by the rheological properties perception module and the spatial thermal field distribution output by the temperature gradient monitoring module are transmitted to the plasticity risk analysis module. The physical information neural network drives the moisture content prediction value to calibrate the valve opening array of the thermal field equilibrium intelligent agent;

[0015] The equipment collaboration module outputs the execution results and feeds them back to the temperature gradient monitoring module to dynamically correct the spatial thermal field distribution reconstruction parameters;

[0016] The quantum annealing processor outputs a moisture content prediction value mutation signal which is input into the safety interlock module, and the pulse neural network chip triggers a high-frequency descaling pulse.

[0017] The equipment collaboration module outputs a voting failure signal to activate the back-mixing adjustment and intelligent weight calculation of the dry powder ratio correction amount;

[0018] The instantaneous flow difference of the material balance report output by the flow synchronization module is transmitted to the safety protection intelligent body as the inert gas flow correction coefficient.

[0019] Furthermore, the sludge drying treatment process control system based on coating backmixing of the present invention further includes:

[0020] The micro fiber grating sensor array of the rheological property sensing module is encapsulated in a corrosion-resistant capillary tube, and the rheological property sensing module includes a shear stress and viscosity conversion function for converting material deformation stress data into viscosity parameters; the shear stress and viscosity conversion function are synchronously configured as follows:

[0021] The shear stress and viscosity conversion function updates the non-Newtonian fluid constitutive equation constraint terms in the physical information neural network of the plastic risk analysis module;

[0022] When the transient fluctuation of the viscosity parameter is greater than 15%, the thermal field equilibrium intelligent agent is triggered to correct the reconstruction parameters of the spatial interpolation algorithm associated with the temperature gradient monitoring module.

[0023] Furthermore, the sludge drying treatment process control system based on coating backmixing of the present invention further includes:

[0024] The quantum annealing processor decomposes the moisture content inversion problem into a 128-dimensional quadratic unconstrained binary optimization model, outputting a plasticity risk level signal and moisture content mutation characteristic parameters, where:

[0025] The plasticity risk level signal includes L1-L5 risk response thresholds, and when the risk reaches L4 or above, the pre-trigger mechanism of the safety interlock module is activated;

[0026] The moisture content mutation characteristic parameter is generated by the first-order derivative of the moisture content prediction value and is used to modulate the pulse trigger frequency of the pulse neural network chip. The upper limit of the pulse frequency is positively correlated with the mutation characteristic parameter.

[0027] When the gradient of the moisture content prediction value is monitored to be greater than 8% / second, the quantum annealing processor sends an excessive gradient signal to the safety interlock module, triggering the voting process of the layered material reduction program covering the equipment collaboration module.

[0028] Furthermore, the sludge drying treatment process control system based on coating backmixing of the present invention further includes:

[0029] The Nash equilibrium arbitrator sets the utility function including drying efficiency weight, safety weight and energy consumption weight, and implements a weight dynamic coupling mechanism;

[0030] When the oxygen concentration change rate calculated by the safety protection agent based on the inert gas flow instruction is greater than 0.5% / second, the safety weight is increased to 0.5 and the energy consumption weight is reduced to 0.1.

[0031] When the heat transfer efficiency attenuation index received by the thermal field balance agent exceeds the threshold, the drying efficiency weight ratio is reduced to 0.3, and the reduced weight value is all allocated to the safety weight to obtain the weight ratio adjustment result;

[0032] The weight ratio adjustment result is transmitted to the plasticity risk analysis module, triggering the quantum annealing processor to reconstruct the optimization objective function.

[0033] Furthermore, the sludge drying treatment process control system based on coating backmixing of the present invention further includes:

[0034] The weighted voting consensus mechanism is used to execute the fault-tolerant cascade response chain of device nodes and agents:

[0035] When the screw feeder node feedback rejects the dry powder ratio correction result, the equipment coordination module reduces the voting weight coefficient of the back-mixing adjustment agent in the Nash equilibrium arbitrator to 50% of the original value;

[0036] The equipment collaboration module triggers the plastic risk analysis module to add a mass conservation constraint term for the sludge fluid in the physical information neural network, recalculate the moisture content, and generate a new dry powder ratio correction;

[0037] If the screw feeder node rejects the new dry powder ratio correction again, the safety interlock module triggers the layered material reduction program and forces the adjustment of the dry powder addition ratio. The equipment collaboration module simultaneously restricts the adjustment authority of non-critical areas in the thermal oil valve opening array that do not affect the oxygen concentration safety. The cascade response action parameters are recorded in the equipment collaboration module, and the equipment historical reliability score and subsequent voting weight distribution are dynamically updated.

[0038] Furthermore, in the sludge drying process control system based on coating backmixing described in the present invention, the pulse neural network chip of the safety interlock module performs dual-channel fusion monitoring and response, and the dual channels include:

[0039] In the thermodynamic response channel, the pulse neural network chip calculates the decay rate based on the heat transfer efficiency decay index output by the temperature gradient monitoring module. When the decay rate is greater than 0.15% / second, a 50Hz base cleaning pulse is generated and directly connected to the rake arm controller.

[0040] In the flow field response channel, the pulse neural network chip superimposes a 20-80Hz variable frequency modulation pulse based on the gradient change rate of the viscosity-temperature coupling field output by the plasticity risk analysis module. The modulation frequency is positively correlated with the gradient change rate.

[0041] The pulse neural network chip integrates and processes the signals of the thermodynamic response channel and the flow field response channel, and transmits them to the plastic risk analysis module to update the mass conservation constraint item, and triggers the real-time update of the response delay calibration coefficient of the safety interlock module based on the integrated signal.

[0042] Furthermore, in the sludge drying process control system based on coating backmixing of the present invention, the spatial thermal field distribution reconstruction deviation of the temperature gradient monitoring module performs three-level conversion control, and the three-level conversion control includes:

[0043] First-level conversion: The temperature gradient monitoring module quantifies the reconstruction deviation into a heat transfer efficiency attenuation index in real time. When the heat transfer efficiency attenuation index is greater than 0.25, the equipment collaboration module locks the adjustment permission for non-critical areas.

[0044] Secondary linkage: The thermal field balancing agent receives the heat transfer efficiency attenuation index. Every time the heat transfer efficiency attenuation index is detected, it increases by 0.1 and reduces the energy consumption weight ratio by 0.05.

[0045] Three-level reconstruction: The thermal field balancing agent calculates the heat transfer efficiency attenuation index of adjacent sampling periods, transmits the difference of the heat transfer efficiency attenuation index of adjacent sampling periods to the quantum annealing processor, reconstructs the boundary constraints of the quadratic unconstrained binary optimization model, and generates plasticity risk level signal pre-correction instructions.

[0046] Furthermore, the sludge drying treatment process control system based on coating backmixing of the present invention further includes:

[0047] The thermal field balancing agent executes the weight and index dynamic compensation logic:

[0048] When the thermal field balance agent receives the heat transfer efficiency attenuation index from the temperature gradient monitoring module:

[0049] Calculate the energy consumption weight ratio reduction value, energy consumption weight ratio reduction value = min(0.2, attenuation index 0.7);

[0050] Determine the increase in the inert gas flow command of the security protection intelligent body: increase = attenuation index 30%;

[0051] The thermal field equilibrium agent transmits the weight adjustment results to the plastic risk analysis module, driving the physical information neural network to add the heat conduction time derivative constraint term to the viscosity-temperature coupling equation;

[0052] The thermal field balancing agent synchronously generates the spatial thermal field distribution reconstruction parameter calibration instructions of the temperature gradient monitoring module and outputs them to the temperature gradient monitoring module.

[0053] Furthermore, in the sludge drying process control system based on coating backmixing of the present invention, the mass conservation constraint term of the physical information neural network executes a fluid and pulse joint control chain, and the pulse joint control chain includes: a constraint strengthening mechanism and parameter mutual feedback;

[0054] Constraint strengthening mechanism:

[0055] When the instantaneous flow difference in the material balance report output by the flow synchronization module is greater than 8%, the weight of the mass conservation constraint item is increased to 200%;

[0056] When the flow difference exceeds the limit for 30 seconds, an unsteady flow equation constraint is added;

[0057] Parameter mutual feedback execution:

[0058] The plastic risk analysis module transmits the change in constraint item weight to the safety interlock module and converts it into a response delay calibration coefficient, Δ calibration coefficient = -0.5*weight change;

[0059] The plastic risk analysis module extracts the residual term of the mass conservation equation and generates the descaling pulse frequency offset: offset = |residual|*50 Hz, which is input into the safety interlock module.

[0060] Furthermore, in the sludge drying process control system based on coating backmixing described in the present invention, the pulse neural network chip of the safety interlock module performs triple modulation coordinated control on the 50Hz descaling pulse, and the triple modulation coordinated control includes:

[0061] Frequency dynamic tuning: The pulse neural network chip calculates the base frequency based on the response delay calibration coefficient: ƒ = 50*(1-0.1 calibration coefficient), where the calibration coefficient is the Δ calibration coefficient;

[0062] Amplitude flow field modulation: The pulse neural network chip modulates the pulse amplitude based on the gradient change rate of the viscosity-temperature coupling field output by the plasticity risk analysis module according to the formula: pulse amplitude = base amplitude (1 + |gradient change rate| / 8);

[0063] Phase safety synchronization: When the heat transfer efficiency attenuation index is greater than 0.3, the mechanical vibration cycle characteristics of the rake arm are identified and the descaling pulse phase is locked in synchronization with the vibration cycle;

[0064] The pulse neural network chip outputs modulated pulses to the rake arm controller, triggering a resonance-enhanced scale removal effect.

[0065] Beneficial effects of the present invention:

[0066] The present invention uses a corrosion-resistant micro-fiber Bragg grating sensor array to capture material deformation and stress data in real time, and outputs dynamic viscosity parameters through a built-in conversion function, thus solving the problem of missing viscosity monitoring caused by the failure of traditional sensors in corrosive environments. A physical information neural network integrates viscosity, thermal field distribution, and material balance data to solve the viscosity-temperature coupling equation. Combined with a quantum annealing processor, it optimizes the moisture content prediction accuracy and generates a plasticity risk level signal, thus breaking through the prediction bottleneck under nonlinear rheological conditions. A multi-agent decision module generates a Pareto optimal instruction set based on the dynamic allocation of weights based on Nash equilibrium. The equipment collaboration module verifies the feasibility of the instructions through distributed voting, achieving adaptive adjustment of the back-mixing ratio as the material state changes. The pulse neural network chip of the safety interlock module performs triple modulation scale cleaning control, with frequency tuning matching the system delay characteristics, amplitude modulation responding to flow field gradient changes, and phase synchronization exciting the rake arm resonance effect, which synergistically improves the coke stripping efficiency. The system constructs a closed-loop control chain from viscosity monitoring at the perception layer, risk quantification at the analysis layer, dynamic optimization at the decision layer, to safety linkage at the execution layer, effectively solving the problem of inaccurate back-mixing ratio and improving the stability of the drying process and energy utilization efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] In order to more clearly illustrate the technical solution of the present invention, the following is a brief introduction to the drawings required for use in the embodiments. Obviously, for ordinary technicians in this field, other drawings can be obtained based on the drawings without paying any creative labor.

[0068] Figure 1 This is a system architecture diagram of a sludge drying treatment process control system based on coating backmixing provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0069] To make the technical solution of the present invention clearer, the present invention will be clearly and completely described below in conjunction with specific embodiments of the present invention and corresponding drawings. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. The present invention provided by each embodiment of the present invention is described in detail below in conjunction with the drawings. In order to better understand the purpose of the present invention, the present invention is further described below.

[0070] See also Figure 1 The present invention provides a sludge drying process control system based on coating backmixing, comprising:

[0071] The rheological property sensing module is deployed at the outlet of the coating machine. It obtains material deformation and stress data through a micro fiber grating sensor array and converts the material deformation and stress data into viscosity parameters.

[0072] The rheological properties sensing module deploys a miniature fiber grating (FBG) sensor array at the coating machine outlet. The sensor array utilizes a corrosion-resistant capillary package structure, which isolates the sensor elements from corrosive media such as sludge, maintaining long-term operational stability. The array captures the shear stress signal generated by material extrusion deformation at a fixed sampling frequency, quantifying the deformation amplitude in real time using grating wavelength offset.

[0073] A built-in shear stress and viscosity conversion function converts deformation stress data into dynamic viscosity parameters. This conversion function establishes a stress-viscosity mapping model based on the constitutive relationship of non-Newtonian fluids, and the model parameters are dynamically calibrated based on the rheological properties of the sludge. The viscosity parameter output is connected to a temperature compensation unit, which dynamically adjusts the conversion function coefficients based on ambient temperature changes to eliminate measurement errors caused by thermal drift.

[0074] The viscosity parameters are synchronously transmitted to the physical information neural network of the plasticity risk analysis module. The output signal of the transfer function is directly connected to the constraints of the non-Newtonian fluid constitutive equation in the neural network, and the constraint weight coefficients are updated in real time. This constraint update process enhances the model's ability to characterize the nonlinear rheological properties of sludge and improves the accuracy of moisture content predictions in the plastic zone.

[0075] When transient fluctuations in viscosity parameters are detected exceeding a preset threshold, the rheological properties sensing module sends a correction command to the thermal field balancing agent. This command triggers the parameter reconstruction process of the spatial interpolation algorithm. The algorithm utilizes the coupling relationship between the current viscosity fluctuation amplitude and the temperature gradient distribution data to generate interpolation correction coefficients for the thermal oil valve opening array. These interpolation correction coefficients are distributed to each temperature control unit according to the weight of the dryer disk layer temperature zone, achieving dynamic matching between the thermal field distribution and the changing rheological state.

[0076] The viscosity parameter influences the system through a dual-channel mechanism: a regular channel continuously optimizes the neural network constraints; an emergency channel directly connects to the thermal field control unit in the event of abnormal fluctuations. This dual-channel design balances steady-state control accuracy with transient response speed, meeting the real-time requirements of industrial process control.

[0077] The results of the correction instructions are fed back to the spatial interpolation algorithm in the temperature gradient monitoring module. This feedback data drives the dynamic adjustment of the reconstruction parameter calculation logic, forming a closed-loop chain from viscosity sensing to thermal field control and reconstruction optimization. This closed-loop mechanism continuously reduces temperature field reconstruction deviations through iterative learning, thereby improving heat transfer efficiency.

[0078] An abnormal fluctuation filter processes the viscosity parameter output signal. The filter separates steady-state signals from transient noise based on the rate of change within a time window. The steady-state signal is input into the neural network to update the constraints, while transient mutation signals trigger correction instructions. This signal separation mechanism blocks high-frequency interference from disrupting the control system.

[0079] The temperature gradient monitoring module is installed on the multi-layer structure surface of the dryer disc layer. It collects temperature distribution time series data through a 5cm×5cm grid thin film thermocouple matrix and reconstructs the spatial thermal field distribution using a spatial interpolation algorithm.

[0080] The temperature gradient monitoring module utilizes a gridded thin-film thermocouple matrix arranged on the multi-layered surface of the dryer's disc layer. The thermocouples are distributed at a preset density, with each node connected to a data acquisition unit via high-temperature-resistant wires. The matrix collects time-series temperature distribution data at a fixed sampling interval, simultaneously recording temperature gradients at different depths and radial positions within the disc layer.

[0081] The collected temperature data is fed into a spatial interpolation algorithm. This algorithm reconstructs the three-dimensional thermal field distribution based on the inverse distance weighting principle. It calculates temperature estimates for unmonitored areas using the distance weights of the temperature values ​​at adjacent measurement points. The interpolation process incorporates sludge heat transfer parameters to enhance the model's spatial resolution within the material-covered area.

[0082] The reconstructed spatial thermal field distribution data is transmitted in real time to the plasticity risk analysis module. The thermal field distribution map reveals the location of localized overheating areas and identifies the coordinates of areas with abnormal temperature fluctuations. This coordinate data is synchronously linked to the control logic of the thermal field balancing agent, along with the thermal field gradient parameters, providing a spatial positioning reference for regulating the thermal oil zone valves.

[0083] The execution results of the device collaboration module are fed back to the temperature gradient monitoring module. This feedback data drives the dynamic adjustment of the core parameters of the spatial interpolation algorithm, correcting the material specific heat capacity coefficient in the heat conduction model. The parameter correction process is iteratively optimized based on historical reconstruction deviations, gradually improving the accuracy of the thermal field distribution reconstruction.

[0084] The thermocouple matrix includes a built-in temperature drift compensation mechanism. The compensation unit regularly calibrates the zero-point drift of each node using a standard temperature source, and the calibration data is stored in the sensor signature database. This dynamically compensated temperature data eliminates measurement baseline errors caused by long-term operation, maintaining system stability for temperature gradient monitoring.

[0085] Spatial thermal field distribution data is simultaneously used for heat transfer efficiency analysis. The analysis engine calculates the rate of change of temperature field uniformity per unit time and quantifies this rate of change as a heat transfer efficiency attenuation index. This attenuation index is transmitted to the Nash equilibrium arbitrator to participate in weight allocation decisions, enabling real-time linkage between thermal field status and system control strategies.

[0086] The flow synchronization module connects the wet sludge input pipeline and the dry powder delivery pipeline, records the instantaneous mass flow of sludge and dry powder in real time and generates a dynamic material balance report;

[0087] The flow synchronization module installs high-precision mass flowmeters in both the wet sludge input pipeline and the dry powder conveying pipeline. These mass flowmeters directly acquire material mass flow data using the Coriolis force measurement principle, eliminating measurement errors caused by density variations. Hardware-synchronized clocks align the sampling timestamps of the two flowmeters, achieving millisecond-level synchronization of instantaneous flow data.

[0088] The collected flow data is fed into a dynamic material balance calculation engine. The engine calculates the instantaneous flow difference based on the real-time mass flow rates of wet sludge and dry powder, and combines this with historical flow trends to generate a mixed material ratio parameter. This ratio parameter reflects the real-time mixing uniformity during the backmixing process and serves as a baseline input for inert gas flow control.

[0089] The dynamic material balance report includes three core indicators: instantaneous flow rate difference, cumulative mass deviation, and mixing ratio fluctuation coefficient. The report output is connected to the mass conservation constraint unit in the plasticity risk analysis module. This constraint unit dynamically adjusts the weight coefficient of the fluid continuity equation based on the flow rate difference, enhancing the robustness of the moisture content prediction model to material fluctuations.

[0090] When the module detects an instantaneous flow rate difference that consistently exceeds a preset range, it initiates a data verification process. This process uses redundant sensor data for cross-validation, eliminating abnormally fluctuating data points. Flow data that passes verification triggers a self-cleaning command, preventing measurement drift caused by material adhesion and maintaining long-term measurement accuracy.

[0091] Mixing ratio parameters are transmitted to the inert gas control system via an interface within the safety protection agent. These parameters are used to adjust the base flow rate setting proportionally, aligning the inert gas flow rate with the real-time mixing state. This correction logic utilizes closed-loop optimization based on oxygen concentration feedback to prevent the risk of localized oxygen enrichment caused by abnormal material ratios.

[0092] Flow data history is stored in the equipment collaboration module's database. This database analyzes flow fluctuations over time and generates equipment operational status assessment reports. This assessment dynamically adjusts the screw feeder's weight in distributed voting, enabling linked control of equipment reliability and decision-making authority.

[0093] The plasticity risk analysis module receives the viscosity parameters from the rheological properties perception module, the spatial thermal field distribution from the temperature gradient monitoring module, and the dynamic material balance report from the flow synchronization module. It solves the viscosity-temperature coupling equation through a physical information neural network to output a moisture content prediction value, triggering the quantum annealing processor to perform moisture content inverse optimization and generate a plasticity risk level signal.

[0094] The Plasticity Risk Analysis Module receives viscosity parameters from the Rheological Properties Perception Module, spatial thermal field distribution from the Temperature Gradient Monitoring Module, and dynamic material balance reports from the Flow Synchronization Module. A physical information neural network integrates these three input data types and establishes a viscosity-temperature coupling calculation model by embedding constraints from the constitutive equations of non-Newtonian fluids. The neural network solves the problem while simultaneously satisfying the mass and energy conservation equations, transforming the physical laws of sludge rheological properties into prior knowledge for the machine learning model.

[0095] Viscosity parameters update the weight coefficients of non-Newtonian fluid constraints in the neural network in real time, enhancing the model's ability to characterize the nonlinear characteristics of the plastic zone. Spatial thermal field distribution data drives the boundary conditions of the heat conduction equation, dynamically correcting the heat transfer coefficient via temperature gradient parameters. The flow difference parameter in the dynamic material balance report acts on the fluid continuity constraint, improving the model's sensitivity to material mixing conditions.

[0096] After the neural network outputs the moisture content prediction, it triggers a quantum annealing processor to perform inverse optimization. The processor maps the continuous variable prediction value into a discrete quantum bit state, constructing a quadratic unconstrained binary optimization model. The quantum tunneling effect traverses the energy potential well in the solution space, outputting a plasticity risk level signal and moisture content mutation characteristic parameters.

[0097] The risk level signal is classified according to preset grading standards, with thresholds from L1 to L5 corresponding to different control response strategies. When the risk level reaches a high level, the processor simultaneously activates the pre-trigger mechanism of the safety interlock module. The moisture content mutation characteristic parameter is generated from the first-order derivative of the predicted value. After passing through the amplitude threshold limiting module, it is transmitted to the pulse neural network chip to modulate the trigger frequency reference of the descaling pulse.

[0098] The optimization results are verified using a hybrid classical and quantum verification mechanism to eliminate anomalous solutions caused by quantum noise. This verification process employs a cross-validation approach using the confidence probabilities of a physical information neural network to reduce the probability of misjudgment. Validated risk signals are fed into the multi-agent decision-making module, driving the dynamic generation of back-mixing ratio corrections and thermal field control strategies.

[0099] The multi-agent decision-making module analyzes the plastic risk level signal from the plastic risk analysis module, calculates the dry powder addition ratio correction through the back-mixing regulation agent, generates the thermal oil valve opening array through the thermal field balance agent, and sets the inert gas flow instruction through the safety protection agent. The outputs of the back-mixing regulation agent, thermal field balance agent, and safety protection agent are transmitted to the Nash equilibrium arbitrator to generate the Pareto optimal control instruction set.

[0100] The multi-agent decision-making module initiates the collaborative decision-making process upon receiving the plasticity risk level signal. The backmixing adjustment agent calculates the dry powder dosage correction based on the risk level signal strength. The algorithm integrates historical dosage data with risk trends for feedback adjustment. The thermal field balancing agent analyzes spatial thermal field distribution data and generates a thermal oil valve opening array based on the temperature gradient distribution. This array distributes opening values ​​based on the weights of the dryer disc layers, enabling precise thermal field control. The safety protection agent sets inert gas flow commands based on oxygen concentration trends. The command value dynamically adjusts the baseline flow rate based on the material mixing state.

[0101] The output parameters of the three agents are synchronously transmitted to the Nash equilibrium arbitrator. The arbitrator establishes a three-dimensional utility function model for drying efficiency, safety protection, and energy consumption, balancing multi-objective conflicts through a dynamic weight allocation mechanism. When the output parameters of the safety protection agent indicate an abnormal oxygen concentration, the arbitrator elevates the safety weight to a dominant position. When the thermal field equilibrium agent reports a decrease in heat transfer efficiency, the arbitrator transfers the difference in drying efficiency weight to the safety weight. The weight adjustment process uses a gradient smoothing algorithm to avoid command jumps, and a boundary protection unit limits the weight value to the device's executable range.

[0102] The Pareto-optimal instruction set output by the arbitrator includes three control parameters: a dry powder feed ratio correction value, a partitioned valve opening matrix, and an inert gas flow setpoint. This instruction set is transmitted to the device collaboration module for distributed verification. The thermal oil pump node verifies the compatibility of valve openings with the device status, while the screw feeder node verifies the operational feasibility of the dry powder ratio. The verification process utilizes the device's historical operation database and combines it with real-time operating data for compatibility analysis.

[0103] The weight ratio adjustment results are fed back to the plasticity risk analysis module, triggering the quantum annealing processor to reconstruct the optimization objective function. This reconstruction process retains the physical constraints of the original model and only scales the objective coefficient matrices by weight, enabling the moisture inversion model to respond in real time to changes in control strategy priorities. The system establishes a closed-loop optimization chain from decision generation to model iteration, enabling dynamic adaptation of control strategies to physical laws.

[0104] The equipment collaboration module receives the Pareto optimal control instruction set from the multi-agent decision module for distributed voting verification. When the thermal oil pump node confirms that the valve opening array is compatible with the equipment status and the screw feeder node verifies that the dry powder ratio correction is feasible, it executes the weighted voting consensus mechanism to output the equipment execution instruction.

[0105] The equipment collaboration module receives the Pareto optimal control instruction set transmitted by the multi-agent decision module. This instruction set contains three parameters: the dry powder dosage ratio correction, the thermal oil valve opening array, and the inert gas flow rate instruction. The module parses the instruction parameters and distributes them to the corresponding equipment nodes: the thermal oil pump node receives the valve opening array data, and the screw feeder node obtains the dry powder dosage correction.

[0106] Each device node performs command verification based on its local operating status. The thermal oil pump node uses real-time data on oil temperature, pressure, and pump load to verify the compatibility of the valve opening array with the device's physical constraints. The screw feeder node verifies the executable range of the dry powder ratio correction based on silo material inventory and mechanical transmission characteristics. This verification process utilizes a predefined rule engine whose rule base integrates the equipment manufacturer's technical specifications and historical operating experience.

[0107] Node verification results are fed into a weighted voting consensus mechanism. This mechanism assigns voting weights based on the device's historical reliability score, which is derived from the long-term operational database of the device collaboration module. When both the thermal oil pump node and the screw feeder node report passing verification, the consensus mechanism outputs the device execution instruction. If a single node vetoes the decision, the agent weight downgrade process is initiated.

[0108] The execution command outputs three paths: the dry powder feed ratio is sent to the screw feeder controller, the zoned valve opening matrix drives the thermal oil pump actuator, and the inert gas flow setpoint is transmitted to the gas flow control valve. The command execution status is fed back to the temperature gradient monitoring module in real time to drive parameter optimization of the spatial thermal field reconstruction algorithm.

[0109] The voting failure signal triggers the decision-weight downgrade process for the backmixing regulation agent. The module reduces the backmixing regulation agent's voting weight coefficient in the Nash equilibrium arbitrator and simultaneously sends a constraint strengthening instruction to the plasticity risk analysis module, activating the mass conservation constraint enhancement and moisture content recalculation process. This closed-loop feedback mechanism enables dynamic adaptation of equipment status and decision-making authority.

[0110] The safety interlock module, including a pulse neural network chip, monitors the plasticity risk level signal and moisture content prediction value in real time. If three consecutive L5 risk signals are received or the moisture content prediction value suddenly changes by more than 35%, it overrides the conventional instructions and triggers the layered material reduction program.

[0111] The safety interlock module's pulse neural network chip continuously receives plasticity risk level signals and moisture content predictions. The chip's built-in parallel processing unit analyzes these two inputs in real time. The risk level signal processing unit identifies consecutive L5 risk signals and records the number of consecutive triggers using a time window counter. The moisture content prediction processing unit calculates the absolute value of the rate of change between adjacent sampling periods, flagging any sudden change exceeding the limit.

[0112] If three consecutive L5 risk signals or a single sudden change in moisture content exceeding the threshold are detected, the chip activates a priority override mechanism. This override mechanism immediately interrupts the normal command transmission link of the equipment collaboration module and activates the independent control channel of the tiered reduction program. This program operates the dry powder dosing equipment through a direct hardware interface, executing a step-by-step reduction in the feed rate: initially reducing the base rate by a fixed percentage, and then increasing the reduction step by step based on the persistence of the risk, until the system safety baseline is triggered.

[0113] The tiered destocking program synchronously controls the thermal oil valve opening array. The program locks valve access in non-critical areas, allowing adjustments only in safety-critical areas that maintain minimal thermal cycles. Valve control parameters are dynamically calculated based on the current rate of change of oxygen concentration to prevent localized oxygen concentration anomalies during the destocking process.

[0114] A pulse neural network chip synchronously executes triple-modulation descaling control. The base descaling pulse frequency is dynamically adjusted based on the risk level, and the pulse amplitude responds to the rate of change of the viscosity-temperature coupling field gradient. A phase synchronization mechanism for the rake arm vibration is activated when severe heat transfer efficiency degradation is detected. The modulated pulses are directly connected to the rake arm controller via an independent communication link, bypassing conventional control loops and achieving millisecond-level response.

[0115] The program execution status is fed back to the quantum annealing processor in real time. This feedback data includes parameters from the reduction phase, oxygen concentration fluctuation curves, and cleaning intensity records. Based on this feedback, the processor optimizes risk assessment thresholds, forming a closed-loop calibration mechanism for emergency response. The calibration results are written to the configuration registers of the safety interlock module, continuously improving system protection accuracy.

[0116] The viscosity parameters output by the rheological properties perception module and the spatial thermal field distribution output by the temperature gradient monitoring module are transmitted to the plasticity risk analysis module. The physical information neural network drives the moisture content prediction value to calibrate the valve opening array of the thermal field equilibrium intelligent agent;

[0117] The equipment collaboration module outputs the execution results and feeds them back to the temperature gradient monitoring module to dynamically correct the spatial thermal field distribution reconstruction parameters;

[0118] The quantum annealing processor outputs a moisture content prediction value mutation signal which is input into the safety interlock module, and the pulse neural network chip triggers a high-frequency descaling pulse.

[0119] The equipment collaboration module outputs a voting failure signal to activate the back-mixing adjustment and intelligent weight calculation of the dry powder ratio correction amount;

[0120] The instantaneous flow difference of the material balance report output by the flow synchronization module is transmitted to the safety protection intelligent body as the inert gas flow correction coefficient.

[0121] The coating remixing-based sludge drying process control system provided by this invention achieves precise control under nonlinear rheological conditions through the collaboration of multiple modules. The rheological properties sensing module deploys a micro-fiber Bragg grating sensor array at the coating machine outlet to capture material deformation and stress data in real time. This data is converted into dynamic viscosity parameters using a built-in conversion function, establishing real-time sensing capabilities for the material's rheological properties.

[0122] The temperature gradient monitoring module arranges a grid-shaped thin-film thermocouple matrix on the surface of the dryer disc layer, continuously collects temperature distribution time series data, and uses a spatial interpolation algorithm to reconstruct the three-dimensional thermal field distribution to accurately reflect the temperature gradient changes inside the dryer.

[0123] The flow synchronization module connects the wet sludge input pipeline and the dry powder conveying pipeline, and synchronously records the instantaneous mass flow of the two materials through a high-precision mass flow meter, generates a dynamic material balance report, and provides real-time feedback on the material mixing ratio status.

[0124] The plasticity risk analysis module receives viscosity parameters, spatial thermal field distribution, and dynamic material balance reports. A physical information neural network integrates these three inputs, solves the viscosity-temperature coupled constitutive equation for sludge, and outputs a moisture content prediction. A quantum annealing processor performs moisture content inverse optimization calculations, converting the prediction into a plasticity risk level signal, enabling a quantitative risk assessment of the material's condition.

[0125] The multi-agent decision-making module analyzes the plasticity risk level signal, the back-mixing control agent calculates the dry powder dosage correction, the thermal field balance agent generates the thermal oil zone valve opening array, and the safety protection agent sets the inert gas flow command. The outputs of these three parties are then optimized through a Nash equilibrium arbitrator for multi-objective optimization, generating a Pareto-optimal control instruction set that balances drying efficiency and safety objectives.

[0126] The device collaboration module performs distributed verification of the control instruction set. The thermal oil pump node verifies the compatibility of the valve opening array with the device status, and the screw feeder node verifies the executable value of the dry powder ratio correction. A weighted voting consensus mechanism is implemented between nodes, and after reaching a valid consensus, the device execution instruction is output.

[0127] The safety interlock module monitors risk signals and moisture content prediction values ​​in real time through a pulse neural network chip. When continuous high-level risks or sudden changes in moisture content exceeding the limit are detected, it immediately overrides the conventional control channel, triggering the layered material reduction protection program to block accident risks.

[0128] The system establishes a closed-loop optimization mechanism: Viscosity parameters and spatial thermal field distribution drive moisture content predictions to calibrate the valve opening array; equipment execution results are fed back to the temperature monitoring module to optimize the thermal field reconstruction algorithm; moisture content mutation signals are directly connected to safety interlocks to trigger emergency responses; voting failure signals activate the dry powder ratio recalculation process; and instantaneous flow difference parameters correct the inert gas control baseline. Through data interaction, these modules form a self-optimizing control chain, enabling adaptive adjustment of the backmixing ratio based on the material's rheological state.

[0129] The quantum annealing processor and the pulse neural network chip establish cross-domain collaboration. The risk level signal modulates the pulse frequency benchmark, simultaneously triggering the emergency coverage mechanism of the device collaboration module. The thermal field balancing agent dynamically adjusts the weight distribution based on the heat transfer efficiency decay index, driving the physical information neural network to enhance the heat conduction constraint term and simultaneously generating temperature field reconstruction and calibration instructions.

[0130] The fluid continuity constraint implements joint control of the flow field and pulses. Material balance anomalies dynamically increase the constraint weight. Constraint changes are converted into safety response delay calibration factors. The fluid residual term generates a frequency shift instruction for the descaling pulse, forming a cross-border linkage between fluid mechanics and control execution. The descaling pulse achieves precise mechanical control through a triple modulation system. The base frequency responds to the system delay parameter, the pulse amplitude matches the viscosity-temperature gradient change rate, and a phase synchronization mechanism stimulates the rake arm resonance effect, improving coke removal efficiency.

[0131] Through the above technical solutions, the system realizes multi-dimensional coordinated control of the entire sludge drying process, captures rheological characteristics and thermal field distribution in real time at the perception layer, optimizes risk prediction through physical models and quantum computing at the analysis layer, adopts multi-agent dynamic weight distribution at the decision layer, and relies on distributed verification and emergency response mechanisms at the execution layer to form a closed-loop optimization control chain, effectively improving the control accuracy and system stability under nonlinear rheological conditions.

[0132] Specifically, the sludge drying treatment process control system based on coating backmixing of the present invention also includes:

[0133] The micro fiber grating sensor array of the rheological property sensing module is encapsulated in a corrosion-resistant capillary tube, and the rheological property sensing module includes a shear stress and viscosity conversion function for converting material deformation stress data into viscosity parameters; the shear stress and viscosity conversion function are synchronously configured as follows:

[0134] The shear stress and viscosity conversion function updates the non-Newtonian fluid constitutive equation constraint terms in the physical information neural network of the plastic risk analysis module;

[0135] When the transient fluctuation of the viscosity parameter is greater than 15%, the thermal field equilibrium intelligent agent is triggered to correct the reconstruction parameters of the spatial interpolation algorithm associated with the temperature gradient monitoring module.

[0136] In a further optimized technical solution, the rheological properties sensing module's microfiber Bragg grating sensor array utilizes a corrosion-resistant capillary package. This package physically isolates the sensor element from corrosive sludge media, maintaining the sensor's measurement stability over long periods of operation. The sensor array captures the shear stress signal generated by material extrusion deformation in real time at the coating machine outlet. Using a built-in stress-to-viscosity conversion function, this shear stress data is converted into dynamic viscosity parameters, establishing continuous monitoring capabilities for the material's rheological properties.

[0137] The stress and viscosity conversion function is applied synchronously to the physical information neural network of the plasticity risk analysis module. The output of the conversion function is connected to the constraints of the non-Newtonian fluid constitutive equation in the neural network, updating the constraint parameters in real time and enhancing the model's accuracy in characterizing the nonlinear rheological properties of sludge. The constraint update mechanism, based on the mapping relationship between shear stress and viscosity, uses a dynamic calibration coefficient to eliminate measurement drift caused by environmental interference, improving the reliability of viscosity parameters in plasticity risk prediction.

[0138] When abnormal transient fluctuations in viscosity parameters are detected, the rheological properties sensing module sends a correction command to the thermal field balancing agent. This command triggers the thermal field balancing agent to recalculate the core parameters of the spatial interpolation algorithm in the temperature gradient monitoring module. This parameter adjustment process combines the coupling relationship between the current viscosity fluctuation amplitude and the spatial thermal field distribution data to generate interpolation correction coefficients for the thermal oil valve opening array. These interpolation correction coefficients are distributed to each temperature control unit based on the dryer disk layer temperature partition weights, achieving dynamic matching between the thermal field distribution and the changing rheological state.

[0139] The thermal field balancing agent uses temperature gradient data collected by the thin-film thermocouple matrix to perform partitioned optimization of the interpolation algorithm parameters. The optimization results drive local adjustments to the thermal oil valve opening array, modifying only the opening values ​​of the temperature control units corresponding to areas of viscosity anomalies to avoid thermal field oscillations caused by global adjustments. The valve opening corrections undergo a distributed verification process within the device collaboration module, and the thermal oil pump node verifies their compatibility with the device's operating status before executing the corrections.

[0140] Viscosity parameters influence system control through two pathways: a conventional pathway continuously updates neural network constraints to optimize the moisture content prediction model; an emergency pathway directly connects to the thermal field equilibrium agent to reduce response latency when transient fluctuations exceed a threshold. This dual-path collaborative mechanism balances steady-state control accuracy with rapid response to abnormal conditions, meeting the dual requirements of real-time and stability in industrial process control.

[0141] The results of the correction instructions are fed back to the spatial interpolation algorithm in the temperature gradient monitoring module. This feedback data dynamically modifies the calculation logic of the thermal field reconstruction parameters, forming a closed-loop chain from viscosity sensing to thermal field control and reconstruction optimization. This closed-loop chain continuously reduces temperature field reconstruction deviations through an iterative learning mechanism, improving heat transfer efficiency during the drying process.

[0142] The corrosion-resistant package integrates a temperature compensation unit, whose output signal is fed into the calculation of the calibration coefficients for the viscosity conversion function. This calibration coefficient dynamically adjusts the stress-viscosity mapping relationship based on the fiber Bragg grating wavelength, eliminating system errors caused by thermal drift. The calibrated viscosity parameter is then fed into an abnormal fluctuation filter, which separates steady-state signals from transient noise based on the rate of change within a time window, thereby blocking high-frequency interference from disrupting the control system.

[0143] Specifically, the sludge drying treatment process control system based on coating backmixing of the present invention also includes:

[0144] The quantum annealing processor decomposes the moisture content inversion problem into a 128-dimensional quadratic unconstrained binary optimization model, outputting a plasticity risk level signal and moisture content mutation characteristic parameters, where:

[0145] The plasticity risk level signal includes L1-L5 risk response thresholds, and when the risk reaches L4 or above, the pre-trigger mechanism of the safety interlock module is activated;

[0146] The moisture content mutation characteristic parameter is generated by the first-order derivative of the moisture content prediction value and is used to modulate the pulse trigger frequency of the pulse neural network chip. The upper limit of the pulse frequency is positively correlated with the mutation characteristic parameter.

[0147] When the gradient of the moisture content prediction value is monitored to be greater than 8% / second, the quantum annealing processor sends an excessive gradient signal to the safety interlock module, triggering the voting process of the layered material reduction program covering the equipment collaboration module.

[0148] The quantum annealing processor maps the moisture content inversion problem into a quadratic unconstrained binary optimization model. Using quantum tunneling, it traverses the energy potential well space and outputs a plasticity risk level signal and moisture content mutation characteristic parameters. The risk level signal classifies the severity of the sludge plasticity risk according to pre-set criteria, driving the multi-agent decision-making module to adjust the risk response strategy. When the risk level reaches a high threshold, the quantum annealing processor simultaneously activates the pre-trigger mechanism of the safety interlock module, activating the standby state of the safety protection equipment.

[0149] The moisture content mutation characteristic parameter is calculated based on the first-order derivative of the predicted moisture content value and represents the instantaneous intensity of the material state change. This parameter is directly connected to the pulse modulation unit of the pulse neural network chip, dynamically adjusting the trigger frequency of the cleaning pulse in direct proportion to the material's rheological state change. An upper limit protection mechanism is implemented during the pulse frequency modulation process to prevent equipment damage caused by mechanical resonance.

[0150] The quantum annealing processor continuously monitors the gradient of the predicted moisture content. If it detects a gradient exceeding a critical range, it immediately generates a top-priority override command. This override command is transmitted to the safety interlock module via a dedicated communication channel, triggering the tiered material reduction program to forcibly take over the voting process of the equipment coordination module. This emergency channel, independent of the main control system, enables millisecond-level response to interruptions.

[0151] Risk signal hierarchical control and parameter modulation form a synergistic architecture: conventional risk level signals drive the multi-agent decision-making process; mutation characteristic parameters modulate the cleaning pulse frequency; and a super-gradient change direct start safety interlock override mechanism is implemented. This three-level architecture covers all control scenarios, from steady-state operation to extreme operating conditions.

[0152] The quantum annealing processor incorporates a solution space compression algorithm, preserving key dimensions based on the rheological physics of sludge to maintain model prediction accuracy. Output results are verified using a hybrid classical-quantum validation mechanism to eliminate anomalous solutions caused by quantum noise. The pre-trigger mechanism employs a dual-channel confirmation strategy, cross-validating the risk level signal with the confidence probability of the physical information neural network to avoid false triggering from a single data source.

[0153] Upon receiving a signal indicating an excessive gradient, the safety interlock module immediately terminates the normal instruction execution chain. The tiered reduction program directly operates the dry powder dosing equipment and thermal oil valves via an independent control bus, implementing a graded reduction in feed rate. Program execution status is fed back to the quantum annealing processor in real time, forming a closed-loop verification loop for emergency response.

[0154] The pulse frequency modulation unit integrates an amplitude limiting protection circuit to monitor the rake arm drive motor load current in real time. When the current exceeds the safety threshold, the pulse amplitude is automatically reduced, and a smooth transition algorithm is used to prevent mechanical shock. The protection parameters are matched to the mechanical characteristics of the dryer, ensuring equipment safety during descaling operations.

[0155] Specifically, the sludge drying treatment process control system based on coating backmixing of the present invention also includes:

[0156] The Nash equilibrium arbitrator sets the utility function including drying efficiency weight, safety weight and energy consumption weight, and implements a weight dynamic coupling mechanism;

[0157] When the oxygen concentration change rate calculated by the safety protection agent based on the inert gas flow instruction is greater than 0.5% / second, the safety weight is increased to 0.5 and the energy consumption weight is suppressed to 0.1;

[0158] When the heat transfer efficiency attenuation index received by the thermal field balance agent exceeds the threshold, the drying efficiency weight ratio is reduced to 0.3, and the reduced weight value is all allocated to the safety weight to obtain the weight ratio adjustment result;

[0159] The weight ratio adjustment result is transmitted to the plasticity risk analysis module, triggering the quantum annealing processor to reconstruct the optimization objective function.

[0160] The Nash equilibrium arbitrator establishes a utility function system composed of drying efficiency, safety, and energy consumption weights, achieving multi-objective optimization through a dynamic weighted coupling mechanism. When the rate of change in oxygen concentration calculated by the safety protection agent based on the inert gas flow command exceeds a preset threshold, the arbitrator immediately increases the safety weight allocation and reduces the energy consumption weight allocation. The weight adjustment operation recalculates the output equilibrium point of the three agents based on the Pareto optimality principle, forming a safety-focused control strategy.

[0161] When the heat transfer efficiency decay index received by the thermal field balancing agent exceeds the critical range, the arbitrator's weight redistribution process is triggered. The arbitrator automatically reduces the proportion of drying efficiency weights allocated, transferring all the released weight differences to safety weights. This weight transfer process uses a gradient smoothing algorithm to avoid system oscillations caused by control command jumps and maintain the stability of the drying process.

[0162] The weight ratio adjustment results are transmitted via a data bus to the plasticity risk analysis module. This module analyzes the weight changes and drives the quantum annealing processor to reconstruct the constraints of the optimization objective function. The objective function reconstruction process is embedded in the weight influence factor calculation module, enabling the optimization objective of the moisture content inversion model to respond in real time to changes in the control strategy's priority, forming a closed-loop coupling between the decision-making layer and the physical model layer.

[0163] The oxygen concentration rate of change parameter and the heat transfer efficiency attenuation index establish a cross-agent collaborative mechanism: in the oxygen concentration-dominated mode, the safety protection agent's output parameter directly triggers weight suppression; in the heat transfer attenuation-dominated mode, the thermal field balance agent's output parameter triggers weight difference redistribution. These two modes automatically switch based on the operating conditions, achieving multi-dimensional coverage of safety risks.

[0164] The arbitrator has a built-in weight boundary protection unit to prevent weight allocation ratios from exceeding the valid range. This unit sets an upper threshold for safety weights and a lower threshold for energy weights. These thresholds are dynamically adjusted based on the device's real-time operating status. If a weight adjustment instruction exceeds the threshold range, the unit activates a smoothing algorithm to output a feasible instruction set that complies with the device's physical constraints.

[0165] After receiving the reconstruction instruction, the quantum annealing processor recalculates the optimization objective function using the weighted scaling parameters. This objective function reconstruction process retains the physical constraints of the original model and adjusts only the coefficient matrices of each optimization objective. The coefficient matrices are linearly scaled by the weighted scaling, and the scaling is calibrated using orthogonal testing to ensure convergence stability. The reconstructed objective function outputs a risk level signal, which is fed back to the arbitrator to form a closed loop for policy iteration.

[0166] The weight adjustment results are synchronously written to the device collaboration module's historical decision database. This database records weight changes and triggering conditions by timestamp, and generates optimization recommendations for the weight allocation strategy through statistical analysis. These recommendations are fed back to the Nash equilibrium arbitrator's initialization parameter setting unit, enabling self-learning and evolution of the system's strategy.

[0167] Specifically, the sludge drying treatment process control system based on coating backmixing of the present invention also includes:

[0168] The weighted voting consensus mechanism is used to execute a fault-tolerant cascade response chain between device nodes and agents:

[0169] When the screw feeder node feedback rejects the dry powder ratio correction result, the equipment coordination module reduces the voting weight coefficient of the back-mixing adjustment agent in the Nash equilibrium arbitrator to 50% of the original value;

[0170] The equipment collaboration module triggers the plastic risk analysis module to add a mass conservation constraint term for the sludge fluid in the physical information neural network, recalculate the moisture content, and generate a new dry powder ratio correction;

[0171] If the screw feeder node rejects the new dry powder ratio correction again, the safety interlock module triggers the layered material reduction program and forces the adjustment of the dry powder addition ratio. The equipment collaboration module simultaneously restricts the adjustment authority of non-critical areas in the thermal oil valve opening array that do not affect the oxygen concentration safety. The cascade response action parameters are recorded in the equipment collaboration module, and the equipment historical reliability score and subsequent voting weight distribution are dynamically updated.

[0172] A weighted voting consensus mechanism implements a fault-tolerant cascade response process between device nodes and agents. When the screw feeder node reports a negative dry powder ratio correction, the device coordination module immediately adjusts the voting weight coefficient of the backmixing agent in the Nash equilibrium arbitrator, reducing it to a fixed percentage of the original value. This voting weight coefficient modification is synchronized in real time to the utility function calculation unit of the Nash equilibrium arbitrator, weakening the decision-making influence of the corresponding agent.

[0173] The Equipment Collaboration Module sends a constraint-enhancing instruction to the Plasticity Risk Analysis Module, driving the physical information neural network to add mass conservation constraints for the sludge fluid to the computational model. The neural network recalculates the moisture content prediction based on the updated constraints. The quantum annealing processor then performs moisture content inverse optimization to generate a new correction for the dry powder ratio, forming a self-correcting closed loop after the initial rejection.

[0174] If the screw feeder node again rejects the new dry powder ratio correction, the device collaboration module activates the safety interlock module's tiered material reduction program. This program forces the dry powder feed ratio to be adjusted via an independent control channel, overriding the device's local control logic. Simultaneously, the device collaboration module restricts access to non-safety-related areas of the thermal oil valve opening array, retaining only the necessary permissions to maintain basic operations.

[0175] The parameters of all cascading response actions are recorded and stored in the device collaboration module's historical database. This database records the number of vetoes, authority adjustment ranges, and response types by node, and dynamically updates the device's historical reliability score using a reliability assessment algorithm. The score serves as the basis for weight allocation in the subsequent weighted voting consensus mechanism, achieving long-term dynamic matching of device status and decision-making authority.

[0176] The tiered reduction program utilizes a hierarchical control strategy, directly controlling the dry powder dosing equipment via an independent bus to implement a step-by-step reduction in feed rate. Program execution status is fed back to the quantum annealing processor for closed-loop verification. Valve access restrictions implement a zoning management strategy, freezing only temperature control zone adjustment permissions unrelated to oxygen concentration control to maintain core safety functions.

[0177] The weight coefficient recovery mechanism for the remixed regulation agent is activated after the risk is resolved. The device collaboration module gradually increases the agent's voting weight based on historical reliability scores. The recovery process uses an incremental smoothing algorithm to avoid jumps in control instructions. The weight recovery threshold is dynamically calculated based on system stability indicators, forming an adaptive closed-loop adjustment loop for decision-making authority.

[0178] The cascading response process builds a three-tiered fault-tolerance architecture: the first rejection triggers model optimization and permission demotion; the second rejection initiates mandatory security intervention; and historical reliability scores drive long-term decision optimization. This three-tiered architecture covers the entire fault-tolerance process, from immediate response to long-term learning.

[0179] Specifically, in the sludge drying process control system based on coating backmixing described in the present invention, the pulse neural network chip of the safety interlock module performs dual-channel fusion monitoring and response, and the dual channels include:

[0180] In the thermodynamic response channel, the pulse neural network chip calculates the decay rate based on the heat transfer efficiency decay index output by the temperature gradient monitoring module. When the decay rate is greater than 0.15% / second, a 50Hz base cleaning pulse is generated and directly connected to the rake arm controller.

[0181] In the flow field response channel, the pulse neural network chip superimposes a 20-80Hz variable frequency modulation pulse based on the gradient change rate of the viscosity-temperature coupling field output by the plasticity risk analysis module. The modulation frequency is positively correlated with the gradient change rate.

[0182] The pulse neural network chip integrates and processes the signals of the thermodynamic response channel and the flow field response channel, and transmits them to the plastic risk analysis module to update the mass conservation constraint item, and triggers the real-time update of the response delay calibration coefficient of the safety interlock module based on the integrated signal.

[0183] The safety interlock module's pulse neural network chip implements a dual-channel fusion monitoring and response mechanism. The thermodynamic response channel calculates the thermal field degradation rate based on the heat transfer efficiency decay index output by the temperature gradient monitoring module. When the rate exceeds a preset threshold, a reference frequency cleaning pulse is generated. This pulse is directly connected to the rake arm controller via a dedicated communication link, driving the mechanical rake arm to perform the cleaning operation. The reference frequency matches the inherent mechanical characteristics of the dryer's coke layer.

[0184] The flow field response channel receives the viscosity-temperature coupled field gradient change rate output by the plasticity risk analysis module and superimposes a dynamic pulse component using variable frequency modulation technology. The modulation frequency maintains a positive correlation with the viscosity-temperature gradient change rate, and the pulse frequency range is limited to the equipment's safe operating frequency band to avoid the risk of mechanical resonance. The variable frequency modulation process utilizes sideband separation technology to isolate interference between frequency and amplitude modulation.

[0185] The spiking neural network chip performs spatiotemporal integration of the dual-channel signals. The integration algorithm uses pulse temporal coding to fuse thermodynamic decay characteristics with flow field gradient characteristics to generate a composite response eigenvector. This eigenvector is transmitted via a high-speed data bus to the plasticity risk analysis module, driving real-time updates of mass conservation constraints. The constraint update process incorporates the calculation of thermodynamic and flow field coupling factors, enhancing the model's ability to characterize the interaction between heat transfer and flow.

[0186] An integrated signal synchronization mechanism triggers the safety interlock module's response delay calibration mechanism. The plasticity risk analysis module extracts constraint item changes and calculates response delay calibration coefficients using a pre-calibrated mapping function. These coefficients are written to the safety interlock chip's clock calibration register, optimizing system response delays from risk detection to action execution in real time. This addresses protection lags caused by communication delays in industrial sites.

[0187] A cross-validation mechanism is established between the thermodynamic and flow field channels. When a single channel triggers a response threshold, the chip uses data from the other channel for confidence verification. In the event of a data conflict between the two channels, a re-monitoring procedure is initiated. Once verification is successful, the execution instruction is released, reducing the probability of false triggering. The descaling pulse generation unit integrates an amplitude limiting protection circuit that monitors the rake arm drive motor load current in real time and automatically reduces the pulse amplitude when the current exceeds a safe threshold.

[0188] The response delay calibration results are fed back to the quantum annealing processor. Based on the calibrated delay parameters, the processor optimizes the time step setting of the moisture inversion model, improving the real-time prediction performance under extreme operating conditions. This closed feedback loop enables cross-module collaboration from monitoring response to model iteration, enhancing the system's adaptability.

[0189] Historical data from dual-channel signals is stored in the safety interlock module's cache. This cached data is used to train the abnormal pattern recognition model of the spiking neural network chip. Through offline learning, the channel weight distribution ratio is continuously optimized, improving the monitoring system's adaptability to complex operating conditions.

[0190] Specifically, in the sludge drying process control system based on coating backmixing of the present invention, the spatial thermal field distribution reconstruction deviation of the temperature gradient monitoring module performs three-level conversion control, which includes:

[0191] First-level conversion: The temperature gradient monitoring module quantifies the reconstruction deviation into a heat transfer efficiency attenuation index in real time. When the heat transfer efficiency attenuation index is greater than 0.25, the equipment collaboration module locks the adjustment permission for non-critical areas.

[0192] Secondary linkage: The thermal field balancing agent receives the heat transfer efficiency attenuation index. Every time the heat transfer efficiency attenuation index is detected, it increases by 0.1 and reduces the energy consumption weight ratio by 0.05.

[0193] Three-level reconstruction: The thermal field balancing agent calculates the heat transfer efficiency attenuation index of adjacent sampling periods, transmits the difference of the heat transfer efficiency attenuation index of adjacent sampling periods to the quantum annealing processor, reconstructs the boundary constraints of the quadratic unconstrained binary optimization model, and generates plasticity risk level signal pre-correction instructions.

[0194] The temperature gradient monitoring module inputs the spatial thermal field distribution reconstruction deviation into the quantization conversion unit. Based on a preset mapping relationship, the quantization conversion unit outputs a heat transfer efficiency attenuation index, establishing a quantitative correlation between the thermal field reconstruction accuracy and the system's heat transfer performance. When the index exceeds a preset threshold, the device collaboration module immediately locks adjustment permissions for non-critical areas of the thermal oil valve opening array. This permission lock is achieved by modifying the control instruction priority flag. Valves in non-critical areas maintain their current state until the system unlocks them.

[0195] The thermal field balancing agent receives a real-time data stream of the heat transfer efficiency decay index. It monitors the incremental changes in the index and reduces the energy consumption weight allocation of the Nash equilibrium arbitrator by a fixed proportional coefficient. The weight adjustment is achieved through the arbitrator's utility function redistribution algorithm, maintaining the sum of the drying efficiency and safety weights. A gradient smoothing algorithm is used in the weight transfer process to avoid system oscillations caused by jumps in control commands.

[0196] The thermal field balancing agent calculates the exponential difference parameters between adjacent sampling periods. These difference parameters are transmitted to the quantum annealing processor via a high-speed interface, driving the reconstruction of boundary constraints in the quadratic unconstrained binary optimization model. Boundary constraint reconstruction utilizes the real-time calibration value of the sludge specific heat capacity to generate pre-correction instructions for the plasticity risk level signal. This pre-correction instruction compensates for prediction deviations caused by thermal field anomalies, improving the real-time nature of risk prediction.

[0197] A three-level conversion forms a vertical control chain: reconstructing deviations into quantified heat transfer performance indices; index increments trigger weight redistribution; and index differences drive model boundary reconstruction. This control chain enables cross-level collaboration from field monitoring to decision optimization.

[0198] The quantization conversion unit integrates a temperature drift compensation algorithm. This algorithm uses historical thermocouple calibration data to eliminate sensor zero-point drift. The compensated deviation is then fed into the exponential calculation engine. The engine employs an adaptive sliding window filter to suppress measurement noise, with the window width dynamically adjusted based on thermal field stability.

[0199] The weight adjustment process implements a gradual control strategy. When a continuous exponential change is detected, the thermal field equilibrium agent initiates a smooth weight ratio migration procedure. The migration rate is positively correlated with the rate of exponential change, preventing control oscillations caused by weight jumps. The smooth migration parameters are written to the Nash equilibrium arbitrator's strategy buffer.

[0200] Boundary constraint reconstruction introduces a coupling factor between heat conduction and the flow field. After receiving the exponential difference, the quantum annealing processor uses the viscosity-temperature coupling field gradient data to calculate the coupling factor. This coupling factor serves as a correction coefficient for the boundary constraint equation, dynamically scaling the solution space for moisture content inversion and enhancing the model's adaptability to complex operating conditions.

[0201] The results of the pre-correction instructions are fed back to the temperature gradient monitoring module. This feedback data drives the iterative optimization of the reconstruction deviation calculation logic, forming a closed-loop learning mechanism from model correction to thermal field monitoring. This closed-loop mechanism continuously improves the accuracy of thermal field reconstruction through incremental learning.

[0202] Specifically, the sludge drying treatment process control system based on coating backmixing of the present invention also includes:

[0203] The thermal field balancing agent executes the weight and index dynamic compensation logic:

[0204] When the thermal field balance agent receives the heat transfer efficiency attenuation index from the temperature gradient monitoring module:

[0205] Calculate the energy consumption weight ratio reduction value, energy consumption weight ratio reduction value = min(0.2, attenuation index 0.7);

[0206] Determine the increase in the inert gas flow command of the security protection intelligent body: increase = attenuation index 30%;

[0207] The thermal field equilibrium agent transmits the weight adjustment results to the plastic risk analysis module, driving the physical information neural network to add the heat conduction time derivative constraint term to the viscosity-temperature coupling equation;

[0208] The thermal field balancing agent synchronously generates the spatial thermal field distribution reconstruction parameter calibration instructions of the temperature gradient monitoring module and outputs them to the temperature gradient monitoring module.

[0209] After receiving the heat transfer efficiency attenuation index from the temperature gradient monitoring module, the thermal field balance agent executes the weighted and exponential dynamic compensation logic. Based on the attenuation index, the agent calculates the energy consumption weight ratio adjustment using a composite constraint rule consisting of a preset upper threshold and a proportional coefficient. The calculation results are fed into the weight allocation unit of the Nash equilibrium arbitrator, which updates the energy consumption weight ratio in the utility function in real time.

[0210] The safety protection agent simultaneously calculates the magnitude of the inert gas flow rate increase command. This magnitude, which is proportional to the attenuation exponent, is transmitted to the safety protection agent via a communication interface. The safety protection agent adds this magnitude to the base flow rate setpoint to generate a new inert gas control command to compensate for oxygen concentration fluctuations caused by thermal field anomalies.

[0211] The weight adjustment results are transmitted to the plasticity risk analysis module via the data bus. This module drives the physical information neural network to add a heat conduction time derivative constraint term to the viscosity-temperature coupling equation. This constraint term enhances the neural network's ability to characterize transient heat transfer processes and optimizes the accuracy of moisture content predictions under sudden thermal field changes. A progressive strategy is used to increase the constraint's influence in stages.

[0212] The thermal field equalization agent simultaneously generates calibration instructions for the temperature gradient monitoring module's spatial thermal field distribution reconstruction parameters. These instructions, containing the attenuation exponent and residual analysis data for the heat conduction equation, are transmitted to the temperature gradient monitoring module via a point-to-point communication protocol. The temperature gradient monitoring module uses these instructions to modify the core parameters of the spatial interpolation algorithm, improving the accuracy of the thermal field distribution reconstruction.

[0213] The compensation logic creates a dual-path output: a weight adjustment path acts on the decision-making arbitrator; a constraint-enhanced path optimizes the physical model layer. The dual-path outputs are synchronized in time, focusing on thermal anomalies in space. The weight recovery mechanism activates once the index falls back to a safe range, and the arbitrator gradually restores the energy consumption weight ratio based on historical data.

[0214] The Reconstruction Parameter Calibration command has a built-in priority flag. When the heat transfer efficiency decay index exceeds a critical range, the command triggers the temperature gradient monitoring module's real-time parameter refresh mechanism, pausing the current thermal field reconstruction process and immediately applying the new parameters. This refresh mechanism uses differential write technology to update only the local parameters affected by the decay index.

[0215] The constraint enhancement results are fed back to the thermal field equilibrium agent. This feedback data is used to optimize the weight ratio calculation logic, forming a closed-loop learning mechanism from model correction to decision compensation. This closed-loop mechanism continuously improves the accuracy of thermal field anomaly response through incremental learning.

[0216] Specifically, in the sludge drying process control system based on coating backmixing of the present invention, the mass conservation constraint term of the physical information neural network executes a fluid and pulse joint control chain, and the pulse joint control chain includes: a constraint strengthening mechanism and parameter mutual feedback;

[0217] Constraint strengthening mechanism:

[0218] When the instantaneous flow difference in the material balance report output by the flow synchronization module is greater than 8%, the weight of the mass conservation constraint item is increased to 200%;

[0219] When the flow difference exceeds the limit for 30 seconds, an unsteady flow equation constraint is added;

[0220] Parameter mutual feedback execution:

[0221] The plastic risk analysis module transmits the constraint item weight change to the safety interlock module and converts it into a response delay calibration coefficient, Δ calibration coefficient = -0.5*weight change;

[0222] The plastic risk analysis module extracts the residual term of the mass conservation equation and generates the descaling pulse frequency offset: offset = |residual|*50 Hz, which is input into the safety interlock module.

[0223] The mass conservation constraint in the physical information neural network implements a joint fluid and pulse control chain. A constraint reinforcement mechanism, triggered by the material balance report from the flow synchronization module, automatically increases the weight of the mass conservation constraint to a multiple of the baseline value when the instantaneous flow difference exceeds a preset threshold. This operation enhances the influence of the constraint in the neural network and strengthens the representation of the sludge fluid mass conservation characteristics in the moisture content prediction model.

[0224] If the flow rate differential persists beyond a critical range, the neural network dynamically adds constraints to the unsteady flow equation. These constraints incorporate transient flow physics models, enhancing the system's ability to characterize abnormal flow phenomena such as turbulence and air locks. Model parameters are calibrated in real time based on current viscosity-temperature coupled field data, addressing the prediction blind spots of the steady-state model under sudden flow changes.

[0225] During the parameter feedback execution phase, the plasticity risk analysis module monitors changes in constraint weights and converts them into response delay calibration coefficients using a linear mapping function. These coefficients are written into the safety interlock module's clock management unit, optimizing the system response delay from risk identification to action execution in real time, addressing protection lags caused by control link time delays.

[0226] The numerical residual term of the mass conservation equation is simultaneously extracted. This residual term characterizes the degree of local non-conservation in the material flow. The residual amplitude is normalized to generate the cleaning pulse frequency offset parameter, which is input into the safety interlock module via a data interface. This offset parameter participates in the real-time modulation calculation of the pulse neural network chip, ensuring precise matching of cleaning intensity with flow anomalies.

[0227] The joint control chain establishes a bidirectional interaction mechanism: the forward path enhances physical model accuracy through constraint enforcement; the reverse path converts model outputs into execution-level control parameters. This bidirectional interaction forms a closed loop between fluid mechanics and control execution, improving the system's adaptability to complex rheological conditions.

[0228] Constraint weights are increased using a progressive loading strategy. The neural network strengthens the constraints in stages based on the duration of the flow differential. Initially, the basic constraint weights are increased, and when the constraint exceeds the limit for a sustained period, higher-order flow equations are superimposed. The loading curves are calibrated using fluid dynamics simulations to avoid numerical instability caused by sudden changes in the model.

[0229] The frequency offset generation module integrates a database of equipment resonance characteristics. This database stores the rake arm's mechanical vibration signature and maps the calculated offset to a safe operating frequency band. Bandpass filtering is used in this mapping process to eliminate dangerous frequencies that could trigger mechanical resonance, ensuring safe cleaning operations.

[0230] The delayed calibration results are fed back to the flow synchronization module. This feedback data drives iterative optimization of the material balance monitoring algorithm, forming a reverse optimization closed loop from the execution response to the perception layer. This closed-loop mechanism reduces flow measurement errors through continuous learning and improves system stability.

[0231] Specifically, in the sludge drying process control system based on coating backmixing described in the present invention, the pulse neural network chip of the safety interlock module performs triple modulation coordinated control on the 50Hz descaling pulse, and the triple modulation coordinated control includes:

[0232] Frequency dynamic tuning: The pulse neural network chip calculates the base frequency based on the response delay calibration coefficient: ƒ = 50*(1-0.1 calibration coefficient), where the calibration coefficient is the Δ calibration coefficient;

[0233] Amplitude flow field modulation: The pulse neural network chip modulates the pulse amplitude based on the gradient change rate of the viscosity-temperature coupling field output by the plasticity risk analysis module according to the formula: pulse amplitude = base amplitude (1 + |gradient change rate| / 8);

[0234] Phase safety synchronization: When the heat transfer efficiency attenuation index is greater than 0.3, the mechanical vibration cycle characteristics of the rake arm are identified and the descaling pulse phase is locked in synchronization with the vibration cycle;

[0235] The pulse neural network chip outputs modulated pulses to the rake arm controller, triggering a resonance-enhanced scale removal effect.

[0236] The safety interlock module's pulse neural network chip performs triple modulation coordinated control on the baseline cleaning pulse. The dynamic frequency tuning process uses the response delay calibration coefficient stored in the safety interlock module to dynamically calculate the base frequency using a linear compensation algorithm. This compensation algorithm converts the system response delay into a frequency adjustment, ensuring that the pulse frequency matches the control link's delay characteristics in real time, optimizing the timeliness of the cleaning action.

[0237] The amplitude flow field modulation stage receives the viscosity-temperature coupling field gradient rate parameter output by the plasticity risk analysis module. The chip modulates the pulse amplitude in direct proportion to the absolute value of the gradient rate, using a base amplitude superimposed increment model. A limiting protection mechanism is embedded in the modulation process to prevent overload of the mechanical actuator and maintain safe equipment operation.

[0238] The phase safety synchronization mechanism activates when the heat transfer efficiency decay index exceeds a critical threshold. A pulse neural network chip collects time-domain signals from the rake arm vibration sensor and identifies the dominant frequency characteristics of the mechanical vibration through spectral analysis. Upon identification, the cleaning pulse phase angle is immediately synchronized with the vibration period, leveraging the resonance effect to amplify the energy required to remove the coke layer. Phase synchronization is only activated when heat transfer efficiency is severely degraded, preventing the equipment from operating in continuous resonance.

[0239] The triple-modulated signals undergo carrier aggregation processing within the chip. Orthogonal frequency division multiplexing (OFDM) is used to separate the frequency, amplitude, and phase modulation components, eliminating mutual interference between the signals. The modulated pulses are then output via a power amplifier to the rake arm controller, driving the mechanical rake arm to perform resonant cleaning operations.

[0240] Collaborative control builds a three-dimensional optimization system: the frequency dimension compensates for system response delays; the amplitude dimension matches changes in flow field states; and the phase dimension stimulates mechanical resonance. This three-dimensional collaboration overcomes the limitations of a single modulation dimension, forming a scale removal control mechanism with directional energy transfer.

[0241] The frequency tuning unit incorporates a boundary protection circuit. This circuit monitors the harmonic distribution of the cleaning pulse. When it detects harmonic components entering the device's resonant frequency band, it automatically activates a notch filter to suppress hazardous frequency components. Protection parameters are calibrated through frequency sweep experiments to ensure the spectral safety of the pulse output.

[0242] The phase synchronization module implements a pre-trigger verification mechanism. Before identifying a vibration cycle, the chip verifies that the rake arm bearing temperature and vibration amplitude are within acceptable ranges. If either parameter exceeds the limit, synchronization is suspended and the system switches to fixed-phase pulse output mode. This verification mechanism mitigates the risk of mechanical overload and embodies fail-safe design principles.

[0243] The modulated pulse output waveform undergoes edge smoothing. A finite impulse response filter smoothes the rising and falling edges of the pulse, eliminating current surges caused by high-frequency glitches. The filter cutoff frequency is dynamically set based on the electrical characteristics of the rake arm motor to maintain smooth drive current.

[0244] The system's physical information neural network embeds the physical laws of sludge rheological properties into a machine learning model. Using the constitutive equations for non-Newtonian fluids, it converts real-time viscosity parameters into prior knowledge for model training. When solving the viscosity-temperature coupling equation, the neural network integrates temperature field distribution and material balance data, ensuring that moisture content predictions simultaneously adhere to the principles of fluid mechanics and the law of conservation of energy. This physical constraint mechanism significantly improves the model's generalization capabilities in data-sparse regions, avoiding the inaccuracies of traditional purely data-driven models in the plastic zone.

[0245] The quantum annealing processor uses a quadratic unconstrained binary optimization model to solve the moisture content inversion problem, mapping continuous variables into discrete quantum bit states. The processor efficiently searches the solution space through quantum tunneling, outputting a risk level signal and mutation signature parameters. The optimization process preserves key physical constraint dimensions and uses a hybrid classical-quantum verification mechanism to eliminate noisy solutions, ensuring output reliability in industrial control scenarios.

[0246] The Nash equilibrium arbitrator establishes a dynamic multi-objective weight allocation mechanism. When the safety protection agent detects an abnormal oxygen concentration, the arbitrator automatically elevates the safety weight to a dominant position. When the thermal field balance agent reports a decrease in heat transfer efficiency, the arbitrator shifts the difference in drying efficiency weight to the safety weight. The weight adjustment process uses a gradient smoothing algorithm to prevent command jumps, and a boundary protection unit ensures that weight values ​​remain within the device's executable range, achieving engineering adaptability for multi-objective optimization.

[0247] The equipment collaboration module incorporates a historical reliability scoring system when implementing a weighted voting consensus mechanism. When a screw feeder node vetoes a control instruction, the module reduces the voting weight of the corresponding agent, triggering the plasticity risk analysis module to add a mass conservation constraint and regenerate the instruction. A second veto activates the safety interlock's tiered material reduction process, simultaneously freezing permissions for non-critical equipment. Cascaded response data dynamically updates node scores, forming a fault learning mechanism.

[0248] The pulse neural network chip achieves dual-channel signal fusion. The thermodynamic channel generates base descaling pulses based on the rate of heat transfer efficiency decay, while the flow field channel superimposes a variable-frequency modulation component based on the rate of change of the viscosity-temperature gradient. The dual-channel signals are integrated through spatiotemporal encoding to drive two optimizations: updating the weights of the neural network's mass conservation constraints and generating a response delay calibration coefficient. This cross-domain fusion correlates thermal anomalies with flow conditions, enhancing the adaptability of descaling control to operating conditions.

[0249] The temperature gradient monitoring module's three-level conversion control constructs a closed-loop optimization chain. Spatial thermal field reconstruction deviations are quantified as a heat transfer efficiency attenuation index, triggering device permission adjustments. Incremental changes in the index drive arbitrator weight redistribution, and exponential differences reconstruct the quantum optimization model boundaries. The output of each level is fed back to the preceding module, forming a reverse correction path from monitoring to decision-making.

[0250] The dynamic compensation logic of the thermal field balancing agent enables cross-module linkage. The energy consumption weight reduction and inert gas flow rate increase calculated based on the attenuation exponent are applied to the decision-making and execution layers, respectively. This simultaneously drives the physical information neural network to add a heat conduction time derivative constraint and generate temperature field reconstruction and calibration instructions. This closed-loop compensation ensures that abnormal heat transfer conditions simultaneously affect the control strategy, physical model, and perception system.

[0251] A combined fluid and pulse control chain establishes a bidirectional interaction mechanism. Flow synchronization anomalies trigger an increase in the weight of the mass conservation constraint term. The change in the constraint term is converted into a safety response delay calibration coefficient. The residual of the mass conservation equation generates the frequency shift of the descaling pulse. This design directly maps fluid dynamics characteristics to control execution parameters, breaking the response delay bottleneck of traditional hierarchical control architectures.

[0252] The descaling pulse triple modulation system coordinates time-frequency characteristics. A frequency tuning module dynamically adjusts the base frequency based on system delay; an amplitude modulation unit amplifies pulse intensity based on the viscosity-temperature gradient; and a phase synchronization mechanism locks the mechanical vibration period when heat transfer efficiency is severely degraded. This three-dimensional synergy focuses pulse energy on the natural frequency band of the coke layer, improving stripping efficiency through mechanical resonance while minimizing the risk of equipment overload.

[0253] All algorithms are implemented using a hardware architecture that is practical for industrial applications. Microfiber Bragg grating sensors and thin-film thermocouple matrices provide physical-layer sensing; quantum processors and pulsed neural network chips are deployed at the edge computing nodes; and a distributed voting mechanism operates at the device control layer. Modules exchange information via a real-time data bus, forming a complete control loop from physical sensing to decision execution.

[0254] The sludge drying process control system of this invention addresses the challenges of viscosity monitoring and backmixing control during the plastic phase of municipal sludge treatment. A corrosion-resistant micro-fiber Bragg grating sensor array is deployed at the coating machine outlet to capture material extrusion deformation stress data in real time. A built-in stress-to-viscosity conversion function converts deformation stress into dynamic viscosity parameters, addressing the lack of viscosity monitoring caused by traditional sensor failure under corrosive conditions. The temperature gradient monitoring module utilizes a gridded thin-film thermocouple matrix on the dryer's disc layer, employing a spatial interpolation algorithm to reconstruct the three-dimensional thermal field distribution and accurately locate localized overheating areas.

[0255] The flow synchronization module connects the wet sludge and dry powder conveying pipelines, synchronously collecting instantaneous mass flow rates via a high-precision mass flowmeter and generating a dynamic material balance report. The plasticity risk analysis module integrates viscosity parameters, thermal field distribution, and material balance data. A physical information neural network solves the viscosity-temperature coupled constitutive equation to output a moisture content prediction. A quantum annealing processor performs moisture content inverse optimization, generating a plasticity risk level signal and quantifying the risk of material status.

[0256] The multi-agent decision-making module analyzes risk signals: the back-mixing control agent calculates the dry powder dosage ratio correction; the thermal field balance agent generates a zoned valve opening array; and the safety protection agent sets the inert gas flow command. A Nash equilibrium arbitrator dynamically assigns weights to the three-party outputs to generate a Pareto-optimal control instruction set. The equipment coordination module verifies instruction feasibility through a distributed voting mechanism. The thermal oil pump node verifies valve opening compatibility, and the screw feeder node verifies the dry powder ratio enforceability.

[0257] When continuous high-level risks or sudden moisture content exceeding limits are detected, the safety interlock module's pulse neural network chip overrides the normal command chain, triggering the tiered material reduction program to forcibly adjust the feed rate. The system incorporates a closed-loop optimization mechanism: viscosity parameters calibrate the valve opening array; equipment execution results optimize the thermal field reconstruction algorithm; moisture content mutation signals are directly linked to emergency response; and flow rate difference parameters correct the inert gas baseline. Each module forms a self-optimizing control chain through cross-layer data exchange.

[0258] Descaling control utilizes a triple modulation mechanism: the base frequency responds to the system delay parameter; the pulse amplitude matches the viscosity-temperature gradient change rate; and when heat transfer efficiency is severely degraded, the rake arm vibration phase is locked to stimulate resonance. Under abnormal thermal conditions, the thermal field balance agent dynamically reduces the energy efficiency weight, driving the physical information neural network to enhance the heat conduction constraint, while simultaneously generating temperature field reconstruction calibration instructions. The fluid continuity constraint dynamically increases in weight based on flow anomalies, and the constraint change is converted into a safety response delay calibration factor, forming a closed-loop linkage between fluid mechanics and control execution.

[0259] Through the above technical solutions, the system realizes real-time capture of sludge rheological characteristics and thermal field distribution at the perception layer, optimizes risk prediction through physical models and quantum computing at the analysis layer, adopts multi-agent dynamic weight distribution at the decision layer, and relies on distributed verification and emergency response mechanisms at the execution layer to form a closed-loop optimization chain from data acquisition to control execution, effectively solving the problems of missing viscosity monitoring and inaccurate backmixing ratio under nonlinear rheological conditions.

[0260] The present invention deploys a rheological properties sensing module at the coating machine outlet, capturing material deformation and stress data in real time using a micro-fiber Bragg grating sensor array. This module incorporates a built-in shear stress and viscosity conversion function, converting deformation and stress data into dynamic viscosity parameters. This process provides continuous online monitoring of sludge rheological properties, enabling real-time sensing of nonlinear viscosity in the plastic zone, and compensating for the data collection deficiencies of conventional systems due to the lack of sensors.

[0261] Based on real-time viscosity parameters and other input data, the plasticity risk analysis module solves the viscosity-temperature coupling equation through a physical information neural network and outputs a moisture content prediction. A quantum annealing processor further optimizes and generates a plasticity risk level signal. The multi-agent decision module analyzes the risk signal, and the backmixing control agent calculates the dry powder dosing ratio correction. The equipment collaboration module performs distributed voting verification to ensure command feasibility. The safety interlock module overrides conventional control in extreme operating conditions, triggering the stratified material reduction process. This mechanism dynamically adjusts the backmixing ratio as the material state changes, eliminating the inaccuracy issues associated with fixed-ratio strategies.

[0262] Each module forms a closed-loop optimization mechanism through data interaction. Viscosity parameters calibrate the valve opening array of the thermal field balancing agent; equipment execution results are fed back to the temperature monitoring module to optimize the thermal field reconstruction algorithm; moisture content mutation signals are directly connected to the safety interlock to trigger emergency response. A quantum annealing processor and a pulse neural network chip collaborate to enhance prediction accuracy and execution efficiency. The overall system implements a closed-loop control chain from perception to decision-making to execution, improving adaptive control capabilities under nonlinear rheological conditions.

Claims

1. A sludge drying process control system based on coating backmixing, characterized in that: include: The rheological property sensing module is deployed at the outlet of the coating machine. It obtains material deformation and stress data through a micro fiber grating sensor array and converts the material deformation and stress data into viscosity parameters. The temperature gradient monitoring module is installed on the multi-layer structure surface of the dryer disc layer. It collects temperature distribution time series data through a 5cm×5cm grid thin film thermocouple matrix and reconstructs the spatial thermal field distribution using a spatial interpolation algorithm. The flow synchronization module connects the wet sludge input pipeline and the dry powder delivery pipeline, records the instantaneous mass flow of sludge and dry powder in real time and generates a dynamic material balance report; The plasticity risk analysis module receives the viscosity parameters from the rheological properties perception module, the spatial thermal field distribution from the temperature gradient monitoring module, and the dynamic material balance report from the flow synchronization module. It solves the viscosity-temperature coupling equation through a physical information neural network to output a moisture content prediction value, triggering the quantum annealing processor to perform moisture content inverse optimization and generate a plasticity risk level signal. The multi-agent decision-making module analyzes the plastic risk level signal from the plastic risk analysis module, calculates the dry powder addition ratio correction through the back-mixing regulation agent, generates the thermal oil valve opening array through the thermal field balance agent, and sets the inert gas flow instruction through the safety protection agent. The outputs of the back-mixing regulation agent, thermal field balance agent, and safety protection agent are transmitted to the Nash equilibrium arbitrator to generate the Pareto optimal control instruction set. The equipment collaboration module receives the Pareto optimal control instruction set from the multi-agent decision module for distributed voting verification. When the thermal oil pump node confirms that the valve opening array is compatible with the equipment status and the screw feeder node verifies that the dry powder ratio correction is feasible, it executes the weighted voting consensus mechanism to output the equipment execution instruction. The safety interlock module includes a pulse neural network chip that monitors the plasticity risk level signal and moisture content prediction value in real time. When three L5 risk signals are received consecutively or the moisture content prediction value suddenly changes by more than 35%, the conventional instructions are overwritten to trigger the layered material reduction program.

2. The sludge drying treatment process control system based on coating back mixing according to claim 1 is characterized in that: Also includes: The micro fiber grating sensor array of the rheological property sensing module is encapsulated in a corrosion-resistant capillary tube, and the rheological property sensing module includes a shear stress and viscosity conversion function for converting material deformation stress data into viscosity parameters; the shear stress and viscosity conversion function are synchronously configured as follows: The shear stress and viscosity conversion function updates the non-Newtonian fluid constitutive equation constraint terms in the physical information neural network of the plastic risk analysis module; When the transient fluctuation of the viscosity parameter is greater than 15%, the thermal field equilibrium intelligent agent is triggered to correct the reconstruction parameters of the spatial interpolation algorithm associated with the temperature gradient monitoring module.

3. The sludge drying treatment process control system based on coating back mixing according to claim 2 is characterized in that: Also includes: The quantum annealing processor decomposes the moisture content inversion problem into a 128-dimensional quadratic unconstrained binary optimization model, outputting a plasticity risk level signal and moisture content mutation characteristic parameters, where: The plasticity risk level signal includes L1-L5 risk response thresholds, and when the risk reaches L4 or above, the pre-trigger mechanism of the safety interlock module is activated; The moisture content mutation characteristic parameter is generated by the first-order derivative of the moisture content prediction value and is used to modulate the pulse trigger frequency of the pulse neural network chip. The upper limit of the pulse frequency is positively correlated with the mutation characteristic parameter. When the gradient of the moisture content prediction value is monitored to be greater than 8% / second, the quantum annealing processor sends an excessive gradient signal to the safety interlock module, triggering the voting process of the layered material reduction program covering the equipment collaboration module.

4. The sludge drying treatment process control system based on coating backmixing according to claim 3 is characterized in that: Also includes: The Nash equilibrium arbitrator sets the utility function including drying efficiency weight, safety weight and energy consumption weight, and implements a weight dynamic coupling mechanism; When the oxygen concentration change rate calculated by the safety protection agent based on the inert gas flow instruction is greater than 0.5% / second, the safety weight is increased to 0.5 and the energy consumption weight is reduced to 0.

1. When the heat transfer efficiency attenuation index received by the thermal field balance agent exceeds the threshold, the drying efficiency weight ratio is reduced to 0.3, and the reduced weight value is all allocated to the safety weight to obtain the weight ratio adjustment result; The weight ratio adjustment result is transmitted to the plasticity risk analysis module, triggering the quantum annealing processor to reconstruct the optimization objective function.

5. The sludge drying treatment process control system based on coating back mixing according to claim 4 is characterized in that: Also includes: The weighted voting consensus mechanism is used to execute the fault-tolerant cascade response chain of device nodes and agents: When the screw feeder node feedback rejects the dry powder ratio correction result, the equipment coordination module reduces the voting weight coefficient of the back-mixing adjustment agent in the Nash equilibrium arbitrator to 50% of the original value; The equipment collaboration module triggers the plastic risk analysis module to add a mass conservation constraint term for the sludge fluid in the physical information neural network, recalculate the moisture content, and generate a new dry powder ratio correction; If the screw feeder node rejects the new dry powder ratio correction again, the safety interlock module triggers the layered material reduction program and forces the adjustment of the dry powder addition ratio. The equipment collaboration module simultaneously restricts the adjustment authority of non-critical areas in the thermal oil valve opening array that do not affect the oxygen concentration safety. The cascade response action parameters are recorded in the equipment collaboration module, and the equipment historical reliability score and subsequent voting weight distribution are dynamically updated.

6. The sludge drying treatment process control system based on coating backmixing according to claim 5 is characterized in that: The pulse neural network chip of the safety interlock module performs dual-channel fusion monitoring and response, and the dual channels include: In the thermodynamic response channel, the pulse neural network chip calculates the decay rate based on the heat transfer efficiency decay index output by the temperature gradient monitoring module. When the decay rate is greater than 0.15% / second, a 50Hz base cleaning pulse is generated and directly connected to the rake arm controller. In the flow field response channel, the pulse neural network chip superimposes a 20-80Hz variable frequency modulation pulse based on the gradient change rate of the viscosity-temperature coupling field output by the plasticity risk analysis module. The modulation frequency is positively correlated with the gradient change rate. The pulse neural network chip integrates and processes the signals of the thermodynamic response channel and the flow field response channel, and transmits them to the plastic risk analysis module to update the mass conservation constraint item, and triggers the real-time update of the response delay calibration coefficient of the safety interlock module based on the integrated signal.

7. The sludge drying treatment process control system based on coating backmixing according to claim 6 is characterized in that: The spatial thermal field distribution reconstruction deviation of the temperature gradient monitoring module performs three-level conversion control, which includes: First-level conversion: The temperature gradient monitoring module quantifies the reconstruction deviation into a heat transfer efficiency attenuation index in real time. When the heat transfer efficiency attenuation index is greater than 0.25, the equipment collaboration module locks the adjustment permission for non-critical areas. Secondary linkage: The thermal field balancing agent receives the heat transfer efficiency attenuation index. Every time the heat transfer efficiency attenuation index is detected, it increases by 0.1 and reduces the energy consumption weight ratio by 0.

05. Three-level reconstruction: The thermal field balancing agent calculates the heat transfer efficiency attenuation index of adjacent sampling periods, transmits the difference of the heat transfer efficiency attenuation index of adjacent sampling periods to the quantum annealing processor, reconstructs the boundary constraints of the quadratic unconstrained binary optimization model, and generates plasticity risk level signal pre-correction instructions.

8. The sludge drying process control system based on coating backmixing according to claim 7 is characterized in that: Also includes: The thermal field balancing agent executes the weight and index dynamic compensation logic: When the thermal field balance agent receives the heat transfer efficiency attenuation index from the temperature gradient monitoring module: Calculate the energy consumption weight ratio reduction value, energy consumption weight ratio reduction value = min(0.2, attenuation index 0.7); Determine the increase in the inert gas flow command of the security protection intelligent body: increase = attenuation index 30%; The thermal field equilibrium agent transmits the weight adjustment results to the plastic risk analysis module, driving the physical information neural network to add the heat conduction time derivative constraint term to the viscosity-temperature coupling equation; The thermal field balancing agent synchronously generates the spatial thermal field distribution reconstruction parameter calibration instructions of the temperature gradient monitoring module and outputs them to the temperature gradient monitoring module.

9. The sludge drying treatment process control system based on coating backmixing according to claim 8, characterized in that: The mass conservation constraint term of the physical information neural network executes a fluid and pulse joint control chain, and the pulse joint control chain includes: a constraint strengthening mechanism and parameter mutual feedback; Constraint strengthening mechanism: When the instantaneous flow difference in the material balance report output by the flow synchronization module is greater than 8%, the weight of the mass conservation constraint item is increased to 200%; When the flow difference exceeds the limit for 30 seconds, an unsteady flow equation constraint is added; Parameter mutual feedback execution: The plastic risk analysis module transmits the constraint item weight change to the safety interlock module and converts it into a response delay calibration coefficient, Δ calibration coefficient = -0.5*weight change; The plastic risk analysis module extracts the residual term of the mass conservation equation and generates the descaling pulse frequency offset: offset = |residual|*50 Hz, which is input into the safety interlock module.

10. The sludge drying treatment process control system based on coating backmixing according to claim 9, characterized in that: The pulse neural network chip of the safety interlock module performs triple modulation coordinated control on the 50Hz descaling pulse, and the triple modulation coordinated control includes: Frequency dynamic tuning: The pulse neural network chip calculates the base frequency based on the response delay calibration coefficient: ƒ = 50*(1-0.1 calibration coefficient), where the calibration coefficient is the Δ calibration coefficient; Amplitude flow field modulation: The pulse neural network chip modulates the pulse amplitude based on the gradient change rate of the viscosity-temperature coupling field output by the plasticity risk analysis module according to the formula: pulse amplitude = base amplitude (1 + |gradient change rate| / 8); Phase safety synchronization: When the heat transfer efficiency attenuation index is greater than 0.3, the mechanical vibration cycle characteristics of the rake arm are identified and the descaling pulse phase is locked in synchronization with the vibration cycle; The pulse neural network chip outputs modulated pulses to the rake arm controller, triggering a resonance-enhanced scale removal effect.

Images