Multi-element spray fluidized bed composite coating PID control method based on particle swarm algorithm
By employing a PID control method based on particle swarm optimization for multi-element spraying fluidized bed composite coating, the problems of hysteresis and thermal shock in fluidized bed temperature control were solved, achieving high-precision temperature regulation and coating uniformity, and improving the stability of agricultural controlled-release fertilizer production.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SHANDONG AGRICULTURAL UNIVERSITY
- Filing Date
- 2026-04-10
- Publication Date
- 2026-06-26
Smart Images
Figure CN121995734B_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of industrial automatic control technology, specifically relating to a PID control method for multi-element spray fluidized bed composite coating based on particle swarm optimization algorithm. Background Technology
[0002] In the production process of controlled-release fertilizers in agriculture, the use of biodegradable polymers such as polyhydroxyalkanoates, polycaprolactone, polybutylene succinate, polylactic acid, and poly(butylene adipate / terephthalate) to coat urea particles in a fluidized bed represents a core direction for industrialization. The physical mechanism of the top-spray fluidized bed coating process lies in the precise adjustment of key operating variables such as inlet air temperature and fluidizing air volume, enabling the polymer solution sprayed into the fluidized bed to spread rapidly and uniformly on the surface of the urea particles and evaporate to form a film. The heat and mass transfer within this physical process is extremely complex, requiring a high-precision bottom-layer control system to maintain a constant temperature for the core materials. Precise temperature control within the fluidized bed is crucial for ensuring the quality of the coating film. Excessive temperature can lead to microscopic cracking on the coating surface and breakage of polymer crosslinks, while excessively low temperatures can cause localized overwetting and adhesion of solid particles, abnormal material agglomeration, and uncontrolled fluctuations in the temperature field. Furthermore, uncontrolled fluctuations in the temperature field directly affect the spreading of the polymer solution on the surface of the urea particles and the solvent evaporation film formation effect. Currently, the industry often uses fixed proportional integral derivative control algorithms to adjust the inlet air temperature and fan air volume. However, since fluidized beds are typical nonlinear systems with large time lag, conventional fixed parameter controllers are difficult to meet the high-precision process requirements, which in turn leads to uncontrolled fluctuations in the temperature field.
[0003] To further improve temperature control accuracy, the industry typically introduces particle swarm optimization (PSO) algorithms for online adaptive optimization of controller parameters. In specific conditions of continuous composite spraying, the system needs to alternately or mix sprays multiple polymer solutions with different physical properties. Different materials exert varying physical binding forces on solvent molecules, and the dynamic kinematic viscosity of the solutions differs significantly. High-viscosity solutions result in slow solvent evaporation, causing a slow and continuous decrease in fluidized bed temperature; while highly volatile solutions absorb a large amount of latent heat of vaporization in a very short time, leading to a transient and drastic drop in the internal temperature of the fluidized bed. Traditional parameter optimization algorithms use a fixed fitness function structure, calculating only the absolute deviation between the temperature setpoint and the actual measured value from a numerical perspective, completely neglecting to incorporate the specific material physical properties causing this deviation into the core evaluation system. This severe lack of a physical evaluation dimension leads to serious control misjudgments in the optimization algorithm when faced with transient high-intensity endothermic changes, resulting in the calculation of extreme adjustment parameters. This causes severe power overshoot in the heating actuator, ultimately leading to premature solidification of subsequently sprayed fluid materials due to abnormally high temperatures, preventing uniform film formation. Summary of the Invention
[0004] This invention provides a PID control method for composite coating of multi-component sprayed fluidized bed based on particle swarm optimization algorithm, in order to solve the technical problems of severe control lag in the bottom layer regulation system caused by the difference in physical evaporation characteristics of multi-component materials in the fluidized bed and local thermal shock damage caused by excessive heating power compensation.
[0005] This invention provides a PID control method for multi-element spraying fluidized bed composite coating based on particle swarm optimization algorithm, comprising the following steps:
[0006] The operating parameters of the fluidized bed equipment and the measured material temperature with hysteresis characteristics are obtained. A state-space model is constructed by using the extended Kalman filter algorithm to deduce and output the real material temperature and transient evaporation rate without hysteresis.
[0007] Based on the actual material temperature and transient evaporation rate, combined with the physical properties of the coating material, the material evaporation hysteresis coefficient and transient thermal shock intensity were calculated respectively.
[0008] The comprehensive dynamic penalty weight is calculated in parallel using the material evaporation hysteresis coefficient and transient thermal shock intensity, and the target fitness function of the particle swarm algorithm is reconstructed using the comprehensive dynamic penalty weight.
[0009] The running parameters are substituted into the reconstructed target fitness function to trigger the particle swarm optimization algorithm for iterative search, obtain the optimal control parameter set, and overwrite the optimal control parameter set into the underlying control closed-loop module for closed-loop control execution and dynamic parameter distribution, thereby realizing proportional-integral-derivative control of multi-element spraying fluidized bed composite coating based on particle swarm algorithm.
[0010] This invention can effectively eliminate the physical thermal inertia hysteresis of temperature sensors caused by metal protective sleeves. When processing multi-element sprayed materials, it directly introduces the physical evaporation characteristics of the materials into the control and evaluation system, realizing the decoupling and adaptive matching of control for materials with different physical properties. This avoids uncontrolled fluctuations in the temperature field caused by material replacement, thereby ensuring the compactness and uniformity of the composite coating.
[0011] Furthermore, based on the actual material temperature and transient evaporation rate, combined with the physical properties of the coating material, the material evaporation hysteresis coefficient and transient thermal shock intensity are calculated, including:
[0012] Based on the kinematic viscosity of the mixed solution in the current pipeline, the actual material temperature, and the physical boiling point constant of the specific solvent, the material evaporation hysteresis coefficient is calculated using the following formula:
[0013]
[0014] In the formula, This represents the evaporation hysteresis coefficient of the material. This represents the real-time dynamic viscosity of the mixed solution at the current formulation ratio. This represents the basic dynamic viscosity of the pure solvent under standard atmospheric pressure. The absolute thermodynamic temperature value that represents the actual temperature of the material. The absolute thermodynamic temperature value representing the physical boiling point constant of a specific solvent.
[0015] This invention accurately captures the physical nature of evaporation lag caused by high-viscosity fluids by quantifying the diffusion resistance of mixed solutions at specific temperatures and viscosities. This enables the system to more scientifically predict thermal response delays when facing different solvent components, thereby providing accurate data support for subsequent compensation calculations.
[0016] Furthermore, based on the actual material temperature and transient evaporation rate, combined with the physical properties of the coating material, the material evaporation hysteresis coefficient and transient thermal shock intensity are calculated, including:
[0017] Based on the transient evaporation rate, the latent heat of vaporization of the solvent, and the comprehensive specific heat capacity of the solid materials inside the fluidized bed, the transient thermal shock intensity is calculated using the following formula:
[0018]
[0019] In the formula, Represents transient thermal shock intensity. Represents the transient evaporation rate. This represents the control sampling cycle time of the host computer. The latent heat of vaporization of the solvent. This represents the overall specific heat capacity of the solid phase material inside the fluidized bed. This represents the total cumulative mass of materials currently within the bed. This represents the maximum allowable temperature fluctuation tolerance boundary value in the process specification.
[0020] This invention constructs a mathematical model that includes latent heat of vaporization, specific heat capacity, and material mass, accurately measuring the transient impact intensity of the explosive volatilization of materials at the moment of contact on the temperature field. This prevents power overshoot during the switching of easily film-forming materials and effectively avoids the physical damage to the coating structure caused by local high temperature.
[0021] Furthermore, the comprehensive dynamic penalty weight is calculated in parallel using the material evaporation hysteresis coefficient and transient thermal shock intensity, including:
[0022] The preset pre-basic allocation coefficient is multiplied by the material evaporation hysteresis coefficient to obtain the pre-product result. The preset post-basic allocation coefficient is multiplied by the transient thermal shock intensity to obtain the post-product result. The pre-product result and the post-product result are added together to obtain the comprehensive dynamic penalty weight.
[0023] This invention constructs a composite evaluation index that can comprehensively measure system response lag and thermal field damage by weighting and fusing two key physical indicators, evaporation hysteresis and thermal shock. This enables the control system to have stronger adaptability and robustness in complex and ever-changing physical environments.
[0024] Furthermore, the objective fitness function of the particle swarm optimization algorithm is reconstructed using a comprehensive dynamic penalty weight, including:
[0025] Within the evaluation window, based on the conventionally set temperature error weighting coefficient, the process target set temperature, the estimated actual material temperature, the control increment amplitude from the controller output to the actuator, and the comprehensive dynamic penalty weight, the target fitness function is calculated using the following formula:
[0026]
[0027] In the formula, Represents the target fitness function. Represents the evaluation window length. Represents the sequence number of the current sampling period. This represents the temperature error weighting coefficient that is set in a conventional manner. The target temperature for the process is set. Representing the Lag-free estimate of the actual material temperature for each sampling period. The controller is in The control increment amplitude is output to the actuator in each sampling cycle. Representing the The comprehensive dynamic penalty weight is calculated synchronously in each sampling period.
[0028] This invention introduces dynamic penalty weights into the reconstructed function, which can apply extremely high penalty scores when the physical properties of materials deteriorate or the control actions are too drastic, forcibly guiding the particle swarm algorithm to converge toward the parameter domain of a stable response, and actively avoiding dangerous optimization regions that are prone to overcompensation of power.
[0029] Furthermore, the operating parameters of the fluidized bed equipment and the measured material temperature with hysteresis characteristics are obtained, including:
[0030] The programmable logic controller (PLC) is used to synchronously collect the inlet air temperature, inlet air volume, current pipeline spray rate, and measured material temperature with hysteresis characteristics of the fluidized bed bottom equipment at a fixed sampling period.
[0031] Furthermore, a state-space model is constructed using the extended Kalman filter algorithm to deduce the hysteresis-free true material temperature and transient evaporation rate, including:
[0032] The actual material temperature and transient evaporation rate are constructed as a two-dimensional column vector of the state-space model, and the measured material temperature is used as the observation vector.
[0033] The prior state estimate is derived by deriving the prediction equation of the extended Kalman filter algorithm, the Kalman gain matrix is calculated, and the state is updated by combining the measured material temperature. The output is the real material temperature without hysteresis and the real-time transient evaporation rate.
[0034] Furthermore, the running parameters are substituted into the reconstructed target fitness function to trigger the particle swarm optimization algorithm for iterative search, obtaining the optimal set of control parameters, including:
[0035] The system calls the real-time proportioning information from the background formula database and substitutes the running parameters into the reconstructed target fitness function.
[0036] The iterative program of the particle swarm optimization algorithm is triggered, and multiple iterations are performed within the constraint boundary to search for the optimal set of control parameters that minimizes the objective fitness function. The optimal set of control parameters includes proportional control parameters, integral control parameters, and derivative control parameters.
[0037] Furthermore, the optimal control parameter set is overwritten into the underlying control closed-loop module for closed-loop control execution and dynamic parameter distribution, including:
[0038] The proportional control parameters, integral control parameters, and derivative control parameters are overwritten into the control closed-loop module;
[0039] The control closed-loop module calculates the adjustment signal based on the actual temperature deviation, drives the proportional-integral electric regulating valve to change the flow rate of the steam heat exchanger, and drives the frequency converter to change the output speed of the centrifugal fan.
[0040] Furthermore, the real-time dynamic viscosity is obtained by the host computer system through interpolation based on the mass flow rate ratio of each feed pump and a pre-stored fluid viscosity empirical data table. The cumulative total mass of the material in the current bed is obtained by adding the initial bottom material mass to the cumulative solid content of the sprayed liquid accumulated by the system integration.
[0041] The beneficial effects are:
[0042] This invention overcomes the limitations of traditional control algorithms that use completely fixed evaluation dimensions for different controlled polymer materials. It constructs a two-dimensional physical characteristic system encompassing the material's evaporation hysteresis coefficient and transient thermal shock intensity, cleverly transforming this into dynamic weights for the penalty boundary of the constraint control algorithm, thus reshaping the core evaluation logic of control optimization. Simultaneously, the introduction of a nonlinear state observer completely eliminates the thermal inertia error of the physical temperature sensor, achieving deep decoupling between the multi-dimensional perception of the controlled object's physical characteristics and the controller parameter search process. It eliminates the persistent problem of abnormal heating power compensation caused by external transient thermal shocks at the algorithm's source, significantly compressing the dynamic temperature range in the core working area of the fluidized bed. The highly constant reaction temperature field fundamentally ensures the compactness and uniformity of the multilayer composite membrane structure, effectively eliminating abnormal material agglomeration caused by local low temperatures and polymer cross-linking breakage caused by local high temperatures, significantly improving batch stability in the continuous production of controlled-release fertilizers in modern agriculture. Attached Figure Description
[0043] Figure 1 This is a flowchart of a PID control method for multi-element spray fluidized bed composite coating based on particle swarm optimization algorithm.
[0044] Figure 2 This is a comparative schematic diagram of the two-dimensional cloud map of the cross-sectional temperature field distribution of the present invention.
[0045] Figure 3 This is a schematic diagram of the two-dimensional clustering identification distribution of the physical properties of the multi-component coating material of the present invention.
[0046] Figure 4 This is a schematic diagram of the convergence of the adaptive optimization parameters based on dynamic penalty weights according to the present invention. Detailed Implementation
[0047] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0048] An embodiment of the PID control method for multi-element spray fluidized bed composite coating based on particle swarm optimization algorithm provided by this invention:
[0049] like Figure 1 As shown, the PID control method for multi-element spray fluidized bed composite coating based on particle swarm optimization includes the following steps:
[0050] S1: Obtain the operating parameters of the fluidized bed equipment and the measured material temperature with hysteresis characteristics. Construct a state-space model using the extended Kalman filter algorithm to deduce and output the real material temperature and transient evaporation rate without hysteresis.
[0051] In fluidized bed control systems, thermocouple temperature sensors exhibit a significant physical time constant lag between their output measurement signal and the actual material temperature due to the thermal resistance and heat capacity characteristics of their external metal protective sheath. During system operation, a programmable logic controller (PLC) synchronously collects the inlet air temperature, inlet air volume, current pipeline spray rate, and the measured material temperature (with lag characteristics) of the fluidized bed's bottom-layer equipment at a fixed sampling period. To obtain the true internal state, this scheme constructs a nonlinear state-space model based on the law of conservation of energy. The actual material temperature and transient evaporation rate are constructed as two-dimensional column vectors in the state-space model, and the measured material temperature is used as the observation vector.
[0052] The system then uses the prediction equation of the extended Kalman filter algorithm, combined with the input heat flux calculated from the inlet air temperature and air volume, and the mass input calculated from the liquid injection rate in the pipeline, to deduce the prior state estimate at the current moment; then it calculates the Kalman gain matrix and updates the state based on the measured material temperature. Taking a certain sampling analysis as an example, if the sensor's measured temperature remains at 323 Kelvin due to the heat capacity of the casing and changes slowly, the algorithm performs algebraic matrix compensation based on the current maximum inlet air heat flux. After calculating the matrix gain, it can predict and output the true material temperature of 333 Kelvin without hysteresis and the real-time transient evaporation rate value in advance, ensuring that the data obtained by the downstream calculation module truly reflects the instantaneous physical and thermodynamic state of the fluidized bed.
[0053] By constructing a nonlinear state-space observation mechanism, the physical thermal inertia artifacts caused by hardware sensors can be completely eliminated, providing high-fidelity prerequisite control variables for the control system.
[0054] S2, based on the actual material temperature and transient evaporation rate, combined with the physical properties of the coating material, calculates the material evaporation hysteresis coefficient and transient thermal shock intensity respectively.
[0055] In multi-component mixed spraying conditions, the kinematic viscosity of the mixed solution in the current pipeline directly determines the internal resistance of solvent molecules diffusing from the interior of the droplets to the gas-liquid interface. Based on the kinematic viscosity of the mixed solution in the current pipeline, the actual material temperature, and the physical boiling point constant of the specific solvent, this system calculates the material evaporation hysteresis coefficient using the following formula:
[0056]
[0057] In the formula, The evaporation hysteresis coefficient of the representative material. Represents the real-time dynamic viscosity of the mixed solution. This represents the basic dynamic viscosity of the pure solvent under standard atmospheric pressure. The absolute thermodynamic temperature value that represents the actual temperature of the material. The absolute thermodynamic temperature represents the physical boiling point constant. The first dimensionless ratio term in the equation characterizes the drag amplification effect caused by viscosity, while the subsequent dimensionless compensation term characterizes the compensating ability of thermal motion to overcome drag.
[0058] Calculation Example: Assume the current dynamic viscosity of the mixed solution is 1.2 mPa·s, the base dynamic viscosity of the pure solvent is 0.8 mPa·s, and the actual material thermodynamic temperature observed in step S1 is 333.15 Kelvin, corresponding to a solvent physical boiling point of 353.15 Kelvin. First, calculate the absolute difference: 333.15 minus 353.15 equals 20 Kelvin. Divide the difference by the boiling point parameter: 20 divided by 353.15 equals 0.0566. Next, calculate the thermodynamic compensation within the parentheses: 1 plus 0.0566 equals 1.0566, taking its reciprocal (negative 1) equals 0.946. Then calculate the first viscosity ratio: 1.2 divided by 0.8 equals 1.5. Finally, multiply the two results: 1.5 multiplied by 0.946 equals 1.419. This result accurately measures the physical nature of how high viscosity fluids amplify the evaporation hysteresis phenomenon by a factor of 1.419.
[0059] Some readily film-forming polymers undergo explosive volatilization upon contact with particles, severely disrupting the stability of the temperature field within the fluidized bed. This system calculates the transient thermal shock intensity based on the transient evaporation rate, the latent heat of vaporization of the solvent, and the comprehensive specific heat capacity of the solid phase material within the fluidized bed, using the following formula:
[0060]
[0061] In the formula, Represents transient thermal shock intensity. Represents the transient evaporation rate. This represents the control sampling cycle time. The latent heat of vaporization of the solvent. The overall specific heat capacity of a solid material. Represents the cumulative total mass. This represents the maximum temperature fluctuation tolerance boundary value. In the formula, the numerator calculates the total actual heat consumption within a single cycle, the denominator calculates the total maximum buffered heat of the system, and the product of each term ensures the uniformity of energy dimensions at the joule level. Finally, a natural exponential function is used for normalization mapping to prevent divergence.
[0062] Calculation Example: Assume the current transient evaporation rate is 0.05 kg / s, the sampling cycle is 1 second, the latent heat of vaporization is 846,000 joules / kg, the specific heat capacity inside the bed is 1500 joules / kg Kelvin, the total mass is 100 kg, and the tolerance boundary is 5 Kelvin. First, calculate the numerator of the total heat consumption: 0.05 multiplied by 1 multiplied by 846,000 equals 42,300 joules. Next, calculate the denominator of the maximum buffer: 1500 multiplied by 100 multiplied by 5 equals 750,000 joules. Then, calculate the ratio and take the negative: -42,300 divided by 750,000 equals -0.0564. Substituting into the natural exponential function, the exponent value of -0.0564 is approximately 0.945. Finally, subtracting this value from 1 equals 0.055, indicating that the system is currently experiencing a heat loss and damage load of 5.5%.
[0063] By constructing mathematical relationships for the physical properties of materials, complex nonlinear thermodynamic mass transfer processes can be accurately converted into dynamic engineering indicators that can be directly called by control algorithms.
[0064] S3 utilizes the material evaporation hysteresis coefficient and transient thermal shock intensity to calculate the comprehensive dynamic penalty weight in parallel, and uses the comprehensive dynamic penalty weight to reconstruct the target fitness function of the particle swarm algorithm.
[0065] In conventional algorithm optimization, the fitness function is usually composed only of the error integral, completely ignoring the problem of physical property deterioration under multiple material conditions. This scheme multiplies a preset pre-basic allocation coefficient by the material evaporation hysteresis coefficient to obtain a pre-product, and multiplies a preset post-basic allocation coefficient by the transient thermal shock intensity to obtain a post-product. Adding these two products yields the comprehensive dynamic penalty weight. To intuitively illustrate the calculation logic of this weight, assuming the pre-basic allocation coefficient calibrated for the current heating actuator is 0.4 and the post-basic allocation coefficient is 0.6, combined with the values calculated in step S2, the pre-product result is 0.4 multiplied by 1.419 equals 0.5676, and the post-product result is 0.6 multiplied by 0.055 equals 0.033. Adding these two together yields a comprehensive penalty weight of 0.6006. This value comprehensively measures the combined superposition of the current system's response hysteresis attribute and thermal field destructive attribute.
[0066] Subsequently, within the evaluation window, this system reconstructs the target fitness function based on the conventionally set temperature error weighting coefficient, the process target set temperature, the estimated actual material temperature, the control increment amplitude from the controller output to the actuator, and the comprehensive dynamic penalty weight, using the following relationship:
[0067]
[0068] In the formula, Represents the target fitness function. Represents the evaluation window length. Represents the temperature error weighting coefficient. The target temperature for the process is set. This represents an estimated value. This represents the control increment magnitude. This represents the overall dynamic penalty weight.
[0069] To demonstrate the fitness extrapolation within a single sampling period, the process target was set to 338 Kelvin, the estimated real material temperature without hysteresis was 333 Kelvin, and the temperature error coefficient was set to 1.0 by default. During this period, the system attempted to output a valve control increment of 10 physical percentage points, with a comprehensive penalty weight of 0.6006 as calculated previously. First, the error penalty term was calculated: 1.0 multiplied by the absolute value 338 minus 333 equals 5. Next, the action penalty term was calculated: 0.6006 multiplied by 10 equals 6.006. Finally, the two were added together, resulting in a fitness function value of 11.006 for this period. When dealing with high-viscosity materials that cause the weight to increase, the fitness function will impose a very high penalty on drastic actions of the control mechanism, forcing the particle swarm optimization algorithm to converge towards the parameter domain of a stable response.
[0070] By reconstructing the objective function boundary using dynamic weights that incorporate material properties, the control algorithm can be scientifically guided to proactively avoid dangerous optimization regions that are prone to overcompensation of power.
[0071] S4 substitutes the running parameters into the reconstructed target fitness function to trigger the particle swarm optimization algorithm for iterative search, obtains the optimal control parameter set, and overwrites the optimal control parameter set into the underlying control closed-loop module for closed-loop control execution and dynamic parameter distribution, thereby realizing proportional-integral-derivative control of multi-element spraying fluidized bed composite coating based on particle swarm algorithm.
[0072] In a continuous and stable industrial operation, the underlying control bus maintains high-frequency, uninterrupted acquisition of field temperature signals and fan feedback signals. After acquiring the various characteristic data decoupled from the aforementioned pre-processing, the host computer control software calls the synchronous feed ratio information from the background formula database in real time. The system substitutes all operating parameters, including the status of the air intake heat source and the real-time ratio status, into the reconstructed target fitness function. Subsequently, the main control program automatically triggers the core iterative process of the particle swarm optimization algorithm. The algorithm performs thousands of high-speed iterations based on the preset particle swarm size and search space constraint boundaries. Individual particles continuously exchange local and global position information, continuously approaching the mathematical convergence pole that minimizes the reconstructed target fitness function. Finally, it searches for and outputs the optimal control parameter set under the current harsh physical conditions. This parameter set comprehensively covers proportional control parameters, integral control parameters, and derivative control parameters that can suppress temperature fluctuations most quickly.
[0073] After completing global optimization and convergence, the communication gateway immediately overwrites and sends the optimal control parameter set to the underlying hardware control closed-loop module, forcibly replacing the original fixed parameters. The underlying control closed-loop module strictly calculates precise analog adjustment commands based on the current real thermodynamic temperature deviation signal and the new parameters. This drives the proportional-integral electric regulating valve to smoothly change the flux of the steam heat exchanger, while simultaneously driving the frequency converter to precisely change the output speed of the centrifugal fan. Because this set of commands deeply integrates a thermal shock penalty mechanism, the mechanical opening and closing action of the electric valve exhibits a gradual operating mode of first making a small probe and then smoothly closing, completely eliminating the situation where large opening would lead to extreme heat overload. As the formulation of multi-element spray coating materials is constantly changed, this system executes the entire physical algorithm derivation and parameter overwriting process in a millisecond-level response cycle, thereby ensuring that all kinds of coating materials can crosslink and cure into films in the optimal thermodynamic temperature field.
[0074] By constructing a complete and frequently updated underlying closed-loop feedback path, the mathematical theoretical advantages of high-dimensional algorithm solutions can be effectively transformed into direct control efficiency to maintain the continuous constant temperature operation of industrial production lines.
[0075] Figure 2 This is a comparative schematic diagram of two-dimensional cloud maps showing the temperature field distribution across the cross-section of the reactor. The image objectively illustrates the temperature distribution pattern across the reactor cross-section at the same operating moment. The image on the left, under conventional control, shows numerous high-temperature regions with extremely high grayscale values and low-temperature patches representing cold islands, indicating a severe imbalance in heat transfer. In contrast, the temperature field on the right, controlled using this scheme, exhibits a smooth and uniform transition without grayscale discontinuities, directly verifying the effectiveness of this scheme in maintaining global thermodynamic uniformity.
[0076] Figure 3 This is a schematic diagram of the two-dimensional clustering identification of the physical properties of multi-component coating materials. A two-dimensional Cartesian coordinate system is constructed with viscosity engineering index as the horizontal axis and thermal shock engineering index as the vertical axis. It clearly presents three physical data lattice clusters of slow volatilization, fast volatilization and mixing properties. The distribution boundary strictly distinguishes the coordinate intervals that are prone to causing response hysteresis and overshoot, proving the scientificity and authenticity of the feature system division.
[0077] Figure 4 This is a schematic diagram of the convergence of adaptive optimization parameters based on dynamic penalty weights in two dimensions. The multi-layered irregular closed contour lines reflect the complex fitness terrain under physical constraints. The white dots representing the first generation are distributed in a divergent state. The black particles of the final generation, after logical deduction, are highly concentrated in the optimal parameter region at the bottom of the contour line valley and accurately avoid the extreme value dead zone, further confirming the strong guiding role of the reconstructed fitness function in the direction of optimization action.
[0078] The above are all preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Therefore, all equivalent changes made in accordance with the structure, shape and principle of the present invention should be covered within the scope of protection of the present invention.
Claims
1. A PID control method for multi-element spraying fluidized bed composite coating based on particle swarm optimization algorithm, characterized in that, Includes the following steps: The operating parameters of the fluidized bed equipment and the measured material temperature with hysteresis characteristics are obtained. A state-space model is constructed by using the extended Kalman filter algorithm to deduce and output the real material temperature and transient evaporation rate without hysteresis. Based on the actual material temperature and transient evaporation rate, combined with the physical properties of the coating material, the material evaporation hysteresis coefficient and transient thermal shock intensity are calculated, including: Based on the kinematic viscosity of the mixed solution in the current pipeline, the actual material temperature, and the physical boiling point constant of the specific solvent, the material evaporation hysteresis coefficient is calculated. : This represents the real-time dynamic viscosity of the mixed solution at the current formulation ratio. This represents the basic dynamic viscosity of the pure solvent under standard atmospheric pressure. The absolute thermodynamic temperature value that represents the actual temperature of the material. The absolute thermodynamic temperature value representing the physical boiling point constant of a specific solvent; The transient thermal shock intensity is calculated based on the transient evaporation rate, the latent heat of vaporization of the solvent, and the comprehensive specific heat capacity of the solid materials inside the fluidized bed. : Represents the transient evaporation rate. This represents the control sampling cycle time of the host computer. The latent heat of vaporization of the solvent. This represents the overall specific heat capacity of the solid phase material inside the fluidized bed. This represents the total cumulative mass of materials currently within the bed. This represents the maximum allowable temperature fluctuation tolerance boundary value in the process specification. The comprehensive dynamic penalty weight is calculated in parallel using the material evaporation hysteresis coefficient and transient thermal shock intensity, including: The preset pre-basic allocation coefficient is multiplied by the material evaporation hysteresis coefficient to obtain the pre-product result, the preset post-basic allocation coefficient is multiplied by the transient thermal shock intensity to obtain the post-product result, and the pre-product result and the post-product result are added together to obtain the comprehensive dynamic penalty weight. The objective fitness function of the particle swarm optimization algorithm is reconstructed using a comprehensive dynamic penalty weight; The running parameters are substituted into the reconstructed target fitness function to trigger the particle swarm optimization algorithm for iterative search, obtain the optimal control parameter set, and overwrite the optimal control parameter set into the underlying control closed-loop module for closed-loop control execution and dynamic parameter distribution, thereby realizing proportional-integral-derivative control of multi-element spraying fluidized bed composite coating based on particle swarm algorithm.
2. The PID control method for multi-element spraying fluidized bed composite coating based on particle swarm optimization algorithm according to claim 1, characterized in that, The target fitness function of the particle swarm optimization algorithm is reconstructed using a comprehensive dynamic penalty weight. This includes: within the evaluation window, based on the conventionally set temperature error weight coefficient, the process target set temperature, the estimated actual material temperature, the control increment amplitude from the controller output to the actuator, and the comprehensive dynamic penalty weight, the target fitness function is calculated using the following formula: In the formula, Represents the target fitness function. Represents the evaluation window length. Represents the sequence number of the current sampling period. This represents the temperature error weighting coefficient that is set in a conventional manner. The target temperature for the process is set. Representing the Lag-free estimate of the actual material temperature for each sampling period. The controller is in The control increment amplitude is output to the actuator in each sampling cycle. Representing the The comprehensive dynamic penalty weight is calculated synchronously in each sampling period.
3. The PID control method for multi-element spraying fluidized bed composite coating based on particle swarm optimization algorithm according to claim 2, characterized in that, Obtain the operating parameters of the fluidized bed equipment and the measured material temperature with hysteresis characteristics, including: The programmable logic controller (PLC) is used to synchronously collect the inlet air temperature, inlet air volume, current pipeline spray rate, and measured material temperature with hysteresis characteristics of the fluidized bed bottom equipment at a fixed sampling period.
4. The PID control method for multi-element spraying fluidized bed composite coating based on particle swarm optimization algorithm according to claim 3, characterized in that, A state-space model is constructed using the extended Kalman filter algorithm to deduce the hysteresis-free true material temperature and transient evaporation rate, including: The actual material temperature and transient evaporation rate are constructed as a two-dimensional column vector of the state-space model, and the measured material temperature is used as the observation vector. The prior state estimate is derived by deriving the prediction equation of the extended Kalman filter algorithm, the Kalman gain matrix is calculated, and the state is updated by combining the measured material temperature. The output is the real material temperature without hysteresis and the real-time transient evaporation rate.
5. The PID control method for multi-element spraying fluidized bed composite coating based on particle swarm optimization algorithm according to claim 4, characterized in that, Substituting the running parameters into the reconstructed target fitness function triggers the particle swarm optimization algorithm to perform an iterative search, obtaining the optimal set of control parameters, including: The system calls the real-time proportioning information from the background formula database and substitutes the running parameters into the reconstructed target fitness function. The iterative program of the particle swarm optimization algorithm is triggered, and multiple iterations are performed within the constraint boundary to search for the optimal set of control parameters that minimizes the objective fitness function. The optimal set of control parameters includes proportional control parameters, integral control parameters, and derivative control parameters.
6. The PID control method for multi-element spraying fluidized bed composite coating based on particle swarm optimization algorithm according to claim 5, characterized in that, The optimal control parameter set is overwritten into the underlying control closed-loop module for closed-loop control execution and dynamic parameter distribution, including: The proportional control parameters, integral control parameters, and derivative control parameters are overwritten into the control closed-loop module; The control closed-loop module calculates the adjustment signal based on the actual temperature deviation, drives the proportional-integral electric regulating valve to change the flow rate of the steam heat exchanger, and drives the frequency converter to change the output speed of the centrifugal fan.
7. The PID control method for multi-element spraying fluidized bed composite coating based on particle swarm optimization algorithm according to claim 6, characterized in that, The real-time dynamic viscosity is obtained by the host computer system through interpolation based on the mass flow rate ratio of each feed pump and a pre-stored fluid viscosity empirical data table. The cumulative total mass of the material in the current bed is obtained by adding the initial bottom material mass to the cumulative solid content of the sprayed liquid accumulated by the system integration.