A centrifugal impurity removal rotating speed cooperative regulation method for pear juice based on a fuzzy PID algorithm

Abstract

This invention relates to the field of industrial equipment automatic control technology, specifically to a method for coordinated speed adjustment in pear juice centrifugation for impurity removal based on a fuzzy PID algorithm. The method includes: acquiring multi-dimensional parameters of the feed pipe section; calculating the equivalent flow resistance viscosity coefficient using Hagen-Poiseuille's law; calculating the turbidity deviation and deviation change rate based on a fixed process target turbidity and real-time discharge turbidity; using the equivalent flow resistance viscosity coefficient as a feedforward correction factor; performing fuzzy inference and defuzzification processing in conjunction with the turbidity deviation and deviation change rate; outputting the compensation amounts for the proportional gain and integral gain of the PID controller; generating speed adjustment commands through incremental proportional-integral calculations; and performing safe speed coordinated adjustment in conjunction with the centrifuge current and actual speed. This method achieves adaptive matching between centrifuge speed and operating conditions, improving the stability and impurity removal effect of centrifugation.

Description

Technical Field

[0001] This invention relates to the field of automatic control technology for industrial equipment. Specifically, it relates to a method for coordinated adjustment of centrifugal speed for removing impurities in pear juice based on a fuzzy PID algorithm. Background Technology

[0002] Pear syrup, a traditional Chinese medicine and food product, typically involves several continuous processes in its large-scale industrial production, including pressing fresh pears, primary filtration, centrifugation for impurity removal, ultrafiltration, and high-temperature boiling to concentrate it into a syrup. Among these, centrifugation for impurity removal during pear juice pressing is a crucial pre-process determining the clarity of the intermediate juice. The core control medium for this crucial process is a disc centrifuge, which generates corresponding centrifugal force by adjusting the spindle speed, achieving efficient separation of insoluble solids such as suspended fruit fibers and large-molecule pectin polymers from the clarified juice.

[0003] In existing technologies, in the pre-processing impurity removal stage of pear paste processing, fuzzy PID control algorithms are typically used to control and adjust the centrifuge speed during the impurity removal process of pear juice processing. This algorithm takes the turbidity of the pear juice meeting the standard as the core adjustment target. Its conventional implementation process is as follows: real-time acquisition of turbidity data at the centrifuge outlet, calculation of the deviation value and deviation change rate between the actual turbidity and the fixed process target turbidity, importing the two parameters into an offline solidified fuzzy rule matrix to perform fuzzy inference, outputting the speed adjustment control quantity, and then generating frequency converter drive commands through the PID controller to complete the closed-loop speed adjustment.

[0004] However, the control input of existing fuzzy PID control algorithms is limited to the result feedback signal related to the turbidity of the output. The core control parameters and fuzzy inference rules are fixed values ​​calibrated under offline ideal working conditions. They cannot respond to the real-time changes in the dynamic characteristics of the controlled object during centrifugation, which causes the centrifuge speed adjustment to be unable to match the real-time changes in the material-liquid separation characteristics. The output adjustment strength and response speed are not adapted to the current working conditions, which directly affects the clarity control effect of the pear juice product.

[0005] Specifically, under ideal conditions where the characteristics of fresh pear raw materials are stable, the algorithm can meet basic production needs and achieve relatively accurate and stable speed regulation. However, under non-ideal conditions where the characteristics of fresh pear raw materials fluctuate, it is prone to problems such as regulation lag or overshoot oscillation. For example, when processing thick feed liquid with high pectin content, the algorithm's regulation is insufficient and the response is severely lagging under the same turbidity deviation, and the turbidity cannot return to the qualified range for a long time, resulting in continuous production of unqualified feed liquid. When processing thin feed liquid with low pectin content, the algorithm's regulation is excessive under the same turbidity deviation, which easily leads to centrifuge speed overshoot, causing repeated turbidity oscillations, and ultimately resulting in poor consistency of pear juice clarity between the same batch and different batches. Summary of the Invention

[0006] To solve the problem that the existing fuzzy PID control algorithm has fixed core parameters and only relies on the turbidity feedback of the juice material, and cannot adapt to the fluctuations of the characteristics of the pear juice material, resulting in the mismatch between the strength and response speed of the rotational speed regulation and the actual working conditions, the present invention proposes a method for collaborative regulation of the rotational speed of centrifugal impurity removal of pear juice based on the fuzzy PID algorithm, and the method includes: During the process of centrifugal impurity removal of pear juice, taking any sampling moment as the current sampling moment, collecting the pressure difference of the feed pipe section of the centrifuge, the flow rate of the pear juice material in the feed pipe section, the turbidity of the discharge pipe section, the process-set target turbidity, as well as the current and actual rotational speed of the centrifuge; Based on the pressure difference and the flow rate, through the analysis of the fluid resistance characteristics of the Hagen-Poiseuille law, combined with the physical parameters of the feed pipe section, calculating the equivalent flow resistance viscosity coefficient of the pear juice material; calculating the turbidity deviation based on the process-set target turbidity and the turbidity of the discharge pipe section, and calculating the change rate of the turbidity deviation in combination with the turbidity deviation of the previous sampling moment; Taking the equivalent flow resistance viscosity coefficient as the feedforward correction factor, and jointly using the turbidity deviation and the change rate of the turbidity deviation as input variables to perform fuzzy inference and defuzzification processing, and converting them into the compensation amounts of the proportional gain and the integral gain of the PID controller; performing incremental proportional integral operation according to the compensation amounts of the proportional gain and the integral gain, generating the rotational speed regulation instruction of the centrifuge, and performing collaborative regulation in combination with the current and actual rotational speed of the centrifuge.

[0007] This technical solution combines fluid dynamics and electromechanical control. First, it collects multi-dimensional operation parameters covering the feed fluid characteristics, the impurity removal result of the discharge, and the electromechanical operation state, avoiding the one-sidedness of single turbidity feedback data. Subsequently, through the Hagen-Poiseuille law, it restores the true equivalent flow resistance viscosity coefficient of the fluid in the pipe section, realizes the low-cost online perception of the rheological characteristics of the material, and provides a basis for the subsequent control for the precondition working conditions. Then, taking the real-time viscosity as the feedforward correction factor, and jointly using the turbidity deviation and the change rate as the input variables of fuzzy inference, it performs online adaptive compensation for the core proportional and integral gains of the PID controller, making the regulation strength and response speed of the algorithm accurately match the separation characteristics of the current material, and solving the problems of regulation lag and overshoot oscillation caused by fixed parameters. Finally, the generated rotational speed regulation instruction is combined with the real current and rotational speed of the equipment for safety collaborative verification, realizing a complete collaborative control process from the upstream fluid characteristic prediction to the midstream adaptive parameter setting and then to the downstream hardware safety drive, ensuring that the system outputs a centrifugal force adapted to the current material characteristics without breaking through the safety boundary of the equipment, realizing the adaptive matching of the centrifuge rotational speed and complex working conditions, and improving the stability of the centrifugal impurity removal process of pear juice and the consistency of the finished product quality.

[0008] Furthermore, based on pressure difference and flow rate, and through the fluid resistance characteristic analysis of Hagen-Poiseuille's law, combined with the physical parameters of the feed pipe section, the equivalent flow resistance viscosity coefficient of the pear juice liquid is calculated. This includes: obtaining the inner diameter and length of the feed pipe section; constructing the Hagen-Poiseuille's law formula based on the flow rate, pressure difference, inner diameter, and length of the pipe at the current sampling time; and calculating the equivalent flow resistance viscosity coefficient at the current sampling time.

[0009] This technical solution extracts deterministic geometric parameters from inside the pipe section, combines them with real-time monitored pressure difference and flow rate fluid state indicators, and substitutes them into a classical fluid dynamics model for calculation. This avoids the hardware dependence on expensive external viscometers and utilizes the pipe itself as a fluid resistance sensing element, achieving low-cost, real-time online evaluation of the true rheological state of pear juice.

[0010] Furthermore, the equivalent flow resistance viscosity coefficient is used as a feedforward correction factor, and together with the turbidity deviation and the rate of change of turbidity deviation, it is used as input variables to perform fuzzy inference and defuzzification processing, transforming them into compensation amounts for the proportional gain and integral gain of the PID controller. This includes: using the turbidity deviation, the rate of change of turbidity deviation, and the equivalent flow resistance viscosity coefficient at the current sampling time as input variables of the fuzzy inference model, and importing them into a pre-constructed offline three-input fuzzy inference model; the fuzzy inference model has a built-in membership function that maps the exact input value to the degree of fuzzy subset membership; based on the equivalent flow resistance viscosity coefficient, adaptive weighted correction is performed on the membership function of the rate of change of turbidity deviation, and three-input fuzzy inference is performed to obtain the fuzzy output; the centroid method is used to defuzzify the fuzzy output to obtain the compensation amounts for the proportional gain and integral gain of the PID controller, respectively.

[0011] This technical solution takes the viscosity of the feed liquid, a core pre-variable that determines the separation efficiency, as the direct input of fuzzy inference and incorporates it into the core control logic. Through viscosity feedforward, the fuzzy inference process is adaptively corrected throughout the entire process, allowing the core gain parameter of the PID controller to change in real time with the characteristics of the feed liquid. This establishes a collaborative relationship between the prediction of the feeding condition and the closed-loop control of the discharge result, enabling the controller to output appropriate adjustment force and response speed for different viscosity conditions.

[0012] Furthermore, the pre-built offline three-input fuzzy inference model is constructed as follows: the input variables of the fuzzy inference model are determined to be turbidity deviation, the rate of change of turbidity deviation, and the equivalent flow resistance viscosity coefficient; the output of the fuzzy inference model is determined to be the compensation amount of the proportional gain and integral gain of the PID controller; fuzzy universes are set for the input variables and output variables respectively, and fuzzy subsets are divided for the fuzzy universes of the input variables and output variables respectively; based on the fuzzy subsets of the input variables and output variables, a three-input IF-THEN fuzzy rule matrix is ​​formulated, and the three-input fuzzy inference model is constructed.

[0013] Furthermore, the method for setting fuzzy domains for the input variables and output quantities is as follows: obtain the turbidity deviation, the rate of change of turbidity deviation, and the equivalent flow resistance viscosity coefficient at each sampling time under all operating conditions for pear juice centrifugation and purification; determine the fluctuation range of each of the turbidity deviation, the rate of change of turbidity deviation, and the equivalent flow resistance viscosity coefficient using the three-standard-deviation rule; set the fuzzy domains of the turbidity deviation, the rate of change of turbidity deviation, and the equivalent flow resistance viscosity coefficient to intervals that completely cover their respective fluctuation ranges; set the fuzzy domains of the proportional gain compensation and the integral gain compensation to intervals that completely cover the adjustment ranges of the PID controller for the proportional gain and the integral gain, respectively.

[0014] Furthermore, the method for dividing the fuzzy universes of the input variables and output quantities into fuzzy subsets is as follows: the fuzzy universes of the turbidity deviation, the rate of change of the turbidity deviation, the equivalent flow resistance viscosity coefficient, the compensation amount of the proportional gain, and the compensation amount of the integral gain are all divided into five fuzzy subsets: negative large, negative small, zero, positive small, and positive large.

[0015] Furthermore, an adaptive weighted correction is performed on the membership function of the turbidity deviation change rate, including: determining the membership weight based on the equivalent flow resistance viscosity coefficient, and using the membership weight to correct only the membership function of the turbidity deviation change rate; setting positive output upper limits for both the proportional gain compensation and the integral gain compensation; the equivalent flow resistance viscosity coefficient is positively correlated with both the membership weight and the positive output upper limit of the proportional gain compensation; if the equivalent flow resistance viscosity coefficient is not less than the preset benchmark equivalent flow resistance viscosity coefficient, the equivalent flow resistance viscosity coefficient is negatively correlated with the positive output upper limit of the integral gain compensation; if the equivalent flow resistance viscosity coefficient is less than the preset benchmark equivalent flow resistance viscosity coefficient, the equivalent flow resistance viscosity coefficient is positively correlated with the positive output upper limit of the integral gain compensation.

[0016] This technical solution establishes precise correction rules for different viscosity conditions: Under high viscosity conditions, it strengthens the predictive weight of the deviation change rate, relaxes the upper limit of the proportional gain to enhance the adjustment force, and at the same time constrains the upper limit of the integral gain to avoid integral saturation, thus solving the problem of lag in the adjustment of high viscosity liquids; Under low viscosity conditions, it weakens the predictive weight of the deviation change rate, narrows the upper limit of the proportional gain to avoid over-adjustment, and at the same time constrains the upper limit of the integral gain to avoid over-adjustment oscillation, thus solving the problem of over-adjustment oscillation in low viscosity liquids, achieving precise adaptation of adjustment force and response speed under all operating conditions.

[0017] Further, incremental proportional-integral (PI) calculations are performed based on the compensation amounts of the proportional gain and integral gain to generate centrifuge speed control commands. This includes: superimposing the compensation amounts of the proportional gain and integral gain with the preset base proportional gain and base integral gain of the PID controller, respectively, to obtain the adaptive proportional gain and adaptive integral gain at the current sampling time; calculating the proportional adjustment increment based on the adaptive proportional gain at the current sampling time and the difference between the turbidity deviation at the current sampling time and the turbidity deviation at the previous sampling time; calculating the integral adjustment increment based on the adaptive integral gain at the current sampling time and the turbidity deviation at the current sampling time; and superimposing the proportional adjustment increment and the integral adjustment increment to obtain the centrifuge speed control command at the current sampling time.

[0018] This technical solution uses the preset base proportional gain and base integral gain of the PID controller as the origin, and dynamically superimposes the compensation amount determined after fuzzy inference. It calculates and integrates the independent adjustment increments of the proportional and integral values. This incremental superposition operation not only preserves the original stable control benchmark of the system, but also provides targeted speed increment compensation for fluctuations in the viscosity characteristics of the liquid, thus realizing adaptive and coordinated control and adjustment of the centrifuge speed.

[0019] Furthermore, the coordinated adjustment is performed by combining the centrifuge's current and actual speed, including: obtaining the direction of the centrifuge's speed adjustment command at the current sampling moment; constructing a segmented safety bypass logic based on the direction of the speed adjustment command; calculating a safety factor by combining the centrifuge's current at the current sampling moment with a preset equipment safety current warning threshold; multiplying the speed adjustment command at the current sampling moment with the safety factor to obtain the speed adjustment increment after safety limiting; superimposing the speed adjustment increment after safety limiting with the centrifuge's actual speed at the current sampling moment to generate a coordinated speed command for the centrifuge at the current sampling moment; and driving the centrifuge's spindle motor to perform coordinated speed adjustment based on the coordinated speed command.

[0020] This technical solution does not execute the speed adjustment command immediately upon receiving it. Instead, it adds a safety bypass logic that determines the command direction and interacts with the hardware load. By using a safety factor to limit the speed adjustment command, it ensures that every drive command received by the centrifuge's spindle motor is within the equipment's rated safety range, thus balancing adjustment flexibility with equipment operational safety.

[0021] Furthermore, a segmented safety bypass logic is constructed based on the direction of the speed adjustment command. The safety factor is calculated by combining the current of the centrifuge at the current sampling moment with the preset equipment safety current warning threshold. This includes: when the direction of the speed adjustment command is a positive acceleration command, the safety factor is calculated based on the preset load reduction attenuation penalty coefficient and the magnitude by which the current of the centrifuge at the current sampling moment exceeds the equipment safety current warning threshold, using an exponential attenuation rule negatively correlated with the magnitude; when the direction of the speed adjustment command is a negative deceleration command or a zero value maintenance command, the safety bypass logic is executed, the safety factor is set to 1, and no attenuation processing is performed on the speed adjustment command.

[0022] This technical solution starts from the actual load state of the centrifuge and bypasses the deceleration or zero-value maintenance commands that reduce or maintain the current equipment load, ensuring that the control system can quickly guide the equipment back to or maintain a stable operating state. At the same time, for acceleration commands with potential overload risks, a smooth amplitude-limiting attenuation intervention is implemented, which not only retains the adjustment flexibility of the control algorithm under normal operating conditions, but also effectively avoids the risk of motor overload that may be caused by blind speed increase under abnormal operating conditions.

[0023] The present invention has the following effects: This invention, by synchronously collecting multi-dimensional operating parameters and based on the online detection of the viscosity coefficient of the equivalent flow resistance of the liquid feed, incorporates the core pre-variable of liquid feed viscosity as a direct input to the fuzzy inference control system. It uses viscosity feedforward to correct the core gain parameter of the fuzzy PID controller, solving the defects of fixed parameters failing to adapt to fluctuations in liquid feed characteristics, and mismatches between adjustment intensity, response speed, and operating conditions. Simultaneously, before issuing speed adjustment commands to the hardware, a segmented safety limiting mechanism is introduced, achieving deep coordination between feed fluid characteristic prediction, closed-loop control of discharge turbidity, and equipment safety boundaries. This enables adaptive matching of centrifuge speed with complex operating conditions, improving the stability of the centrifugation and impurity removal process, the finished product turbidity compliance rate, and batch-to-batch quality consistency. Attached Figure Description

[0024] Figure 1 This is a schematic diagram of the method flow of the present invention; Figure 2 This is a schematic diagram comparing the turbidity control effect of the present invention with that of the existing classic fuzzy PID algorithm. Detailed Implementation

[0025] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0026] This invention provides a method for coordinated adjustment of centrifugal speed for removing impurities from pear juice based on a fuzzy PID algorithm, such as... Figure 1 As shown, it includes: S1: Obtain multidimensional data of pear juice during the centrifugation and impurity removal process, and calculate the equivalent flow resistance viscosity coefficient based on the fluid resistance characteristics analysis of pressure difference and flow rate.

[0027] When performing the centrifugation and impurity removal process for pear juice, it is important to consider that in actual industrial production lines, due to differences in the maturity, variety, storage period, and preceding juicing and enzymatic hydrolysis processes of different batches of fresh pears, the content of soluble pectin and macromolecular substances in the liquid will fluctuate drastically, resulting in a non-linear abrupt change in the apparent viscosity of the fluid.

[0028] In order to obtain the dynamic evolution of material rheological properties, this step aims to establish a pre-processing layer for fluid characteristic analysis. By using the existing basic instruments on the feed pipe section, the complex non-Newtonian fluid rheological measurements are reduced to a direct proportional mapping relationship between pressure difference and flow rate, and the equivalent flow resistance viscosity coefficient reflecting the internal shear resistance of the fluid is extracted.

[0029] In this scheme, the flow of pear juice in the feed pipe section is analyzed by Hagen-Poiseuille's law to obtain the equivalent flow resistance viscosity coefficient, which is used to characterize the flow resistance characteristics of the liquid in the pipe section. It reflects the viscosity characteristics of the liquid itself, and also takes into account the combined effects of the shear characteristics of pear juice as a non-Newtonian fluid and the local resistance of the pipe section. It can be used as an effective feedforward index to characterize the separation characteristics of the liquid.

[0030] Specifically, during the centrifugation and impurity removal process of pear juice, any sampling time is taken as the current sampling time. The pressure difference of the feed pipe section of the centrifuge, the flow rate of the pear juice in the feed pipe section, the turbidity of the discharge pipe section, the fixed process setting target turbidity, and the current and actual speed of the centrifuge are collected. Among them, the sampling frequency of pressure and flow rate data is uniformly set to 1Hz, and the high-frequency hydraulic fluctuation interference caused by the operation of the feed pump is filtered out by the median filtering algorithm built into the general programmable logic controller to ensure that the data involved in the calculation is smooth and reliable.

[0031] Obtain the inner diameter and length of the feed pipe section. Based on the flow rate, pressure difference, pipe inner diameter, and pipe length at the current sampling time, calculate the equivalent flow resistance viscosity coefficient at the current sampling time using the fluid resistance characteristics analysis of Hagen-Poiseuille's law.

[0032] Specifically, it satisfies the following relationship:

[0033] In this relation, The equivalent flow resistance viscosity coefficient at the current sampling time reflects the shear resistance state caused by the enrichment of macromolecules such as cytokines in the current batch of pear juice. The larger the value, the more viscous the liquid and the greater the sedimentation resistance of large particles. The smaller the value, the thinner the liquid. The pressure difference at the current sampling moment reflects the energy loss of the fluid at both ends of the pressure measuring tube section. The larger this value is, the greater the resistance to fluid transport. The inner diameter of the feed pipe section is obtained in advance. The length of the feed pipe section. This represents the flow rate of pear juice in the feed pipe section at the current sampling time. Pi is a constant.

[0034] In this formula, the molecule part This represents the combined physical quantity of the work done by the fluid to overcome the pressure drop as it flows through the pipe and the pipe's cross-sectional area. It reflects the energy potential driving the fluid flow; a larger value indicates a stronger propulsive force applied by the on-site pumping system. The denominator... It represents the frictional distribution characteristics related to the actual volume of fluid passing through the pipe and the pipe length.

[0035] These two factors are combined through division because they take into account the physical principle in classical Hagen-Poiseuille's law that when a fluid undergoes laminar flow in a straight pipe, its friction loss is proportional to both the flow rate and the fluid viscosity. When the pressure difference... The larger the volume, the greater the simultaneous flow rate. The smaller the value, the greater the driving force of the system, but the actual volume of liquid flowing through it is extremely small, resulting in a lower calculated equivalent flow resistance viscosity coefficient. An increase in pressure difference corresponds to a viscous liquid at the current sampling time, leading to increased friction within the pipe; conversely, if the pressure difference is smaller and the flow rate is larger, the denominator dominates, resulting in a higher final equivalent flow resistance viscosity coefficient. The smaller the value, the more likely it is that the liquid is thin and the flow within the tube is smooth at the current sampling time. To prevent arithmetic overflow, in extreme flow conditions... When the value is 0, a minimal constant term is added to the denominator to prevent it from being zero; it is usually set to 0. .

[0036] This relationship is based on a modified derivation of the Hagen-Poiseuille law in classical fluid mechanics, which describes the relationship between flow rate and pressure drop for incompressible Newtonian fluids moving in laminar flow in a circular pipe. In the current scenario, its modification is used to inversely deduce the equivalent viscosity, cleverly reducing the complexity of non-Newtonian fluid rheological measurements to a direct proportional mapping between pressure difference and flow rate. This allows for online prediction of operating conditions using the pipe's built-in pressure difference and flow rate data, without requiring additional hardware.

[0037] The obtained equivalent flow resistance viscosity coefficient will serve as a feedforward factor, providing accurate predictive data for subsequent feedforward adjustments and forming the foundation for such adjustments. The viscosity of pear juice fluctuates due to batch changes or fluctuations in the preceding enzymatic hydrolysis process. If adjustments are made only after turbidity deviations occur, it will lead to lag and the production of substandard juice. By calculating the equivalent flow resistance viscosity coefficient in real time, changes in juice viscosity can be detected in advance, providing early warnings to the control system. This avoids the lag inherent in passive responses, provides a preliminary basis for subsequent parameter corrections, ensures that adjustments precede deviation changes, and improves system response speed.

[0038] In a specific example, the current pear juice centrifugal impurity removal production line is processing a batch of highly ripe fresh pear juice stored in cold storage, and the inner diameter of the feed pipe section is being measured. Length of feed pipe section At the current sampling moment, the pressure difference between the two ends of the feed pipe section is measured. The flow rate of the filtered pear juice liquid Calculate the equivalent flow resistance viscosity coefficient: ,pass This reflects the viscous state of the fruit juice material at the current sampling time, which has a high pectin content.

[0039] S2: Construct fuzzy PID feedforward adaptive correction rules based on equivalent flow resistance viscosity coefficient.

[0040] After obtaining the equivalent flow resistance viscosity coefficient that characterizes the viscosity of the liquid, given that high-viscosity fluids generate strong viscous resistance in centrifugal separation, hindering the sedimentation of suspended fruit pulp particles and macromolecules, if the control system still uses the fixed PID parameters calibrated under ideal conditions, it will lead to insufficient adjustment and lag response under high viscosity conditions, and excessive adjustment and overshoot oscillation under low viscosity conditions.

[0041] Therefore, this step aims to construct a feedforward adaptive correction mechanism based on the laws of fluid mechanics. By utilizing the positive correlation between resistance and viscosity in Stokes' sedimentation law, an adaptation rule for viscosity and PID gain parameters is established, so that the adjustment intensity and response speed of the algorithm can accurately match the separation characteristics of the current liquid, thus solving the problem of mismatch between fixed parameters and operating conditions.

[0042] This step is the execution phase of feedforward regulation. Its core lies in using the equivalent flow resistance viscosity coefficient to pre-correct control parameters, achieving precise matching between operating conditions and regulation intensity. In actual industrial scenarios, liquids of different viscosities have completely different requirements for centrifugal separation: high-viscosity liquids require greater regulation intensity and weaker integral action to avoid integral saturation, while low-viscosity liquids require even less regulation intensity and weaker integral action to avoid overshoot.

[0043] This step uses real-time changes in the feedforward factor to pre-correct the upper limits of proportional gain, integral gain, and membership weights of the turbidity deviation change rate. Parameter tuning can be completed without waiting for the turbidity deviation to expand, achieving rapid linkage from operating condition changes to feedforward triggering and then to parameter correction. Compared to traditional control, this feedforward correction allows parameters to adapt to the operating conditions in advance. When turbidity deviation occurs, the control system already has the ability to adjust to the current operating conditions, shortening the deviation convergence time and reducing the production of substandard liquid, meeting the core needs of industrial production.

[0044] The baseline equivalent flow resistance viscosity coefficient for pear juice impurity removal under ideal operating conditions is obtained in advance. The upper limit of the positive output of the proportional gain compensation of the PID controller. The upper limit of the positive output of the integral gain compensation amount Turbidity deviation change rate benchmark membership weight All the above parameters are fixed values ​​calibrated based on ideal operating conditions during the centrifuge equipment commissioning phase.

[0045] First, the relative viscosity coefficient at the current sampling time is calculated through dimensionless processing, serving as the basis for calculating all correction rules. The formula is as follows:

[0046] In the formula, The relative viscosity coefficient at the current sampling time characterizes the degree of deviation of the current liquid viscosity from the ideal reference condition. The equivalent flow resistance viscosity coefficient at the current sampling time; The viscosity coefficient is the benchmark equivalent flow resistance.

[0047] With relative viscosity coefficient To delineate the working conditions: High viscosity operating conditions; Low viscosity operating conditions.

[0048] For high-viscosity conditions, based on Stokes's law of sedimentation, the settling resistance of suspended particles increases with increasing liquid viscosity, and the particle settling resistance is positively correlated with fluid viscosity. To counteract this nonlinear physical resistance, a gain scheduling strategy from nonlinear robust control is employed. The correction objectives are: to increase the adjustment magnitude, enhance trend prediction, prevent integral saturation, and avoid adjustment lag under high-viscosity conditions. Specifically, the natural logarithm is used to ensure that the correction magnitude increases smoothly with increasing viscosity, avoiding abrupt oscillations, and negative exponential decay is used. Used to constrain integral gain and prevent integral saturation.

[0049] The corresponding formula is as follows: Formula for correcting the upper limit of the positive output of the proportional gain compensation:

[0050] In the formula, The upper limit of the positive output of the compensation amount of the proportional gain at the current sampling time; The upper limit of the positive output of the proportional gain compensation reference for offline calibration. Using the natural logarithm function ensures that the correction magnitude increases smoothly with increasing viscosity, avoiding abrupt oscillations, thus relaxing the upper limit and enhancing the adjustment power. This represents the relative viscosity coefficient at the current sampling time.

[0051] Turbidity deviation change rate membership weighted correction formula:

[0052] In the formula, The membership weight is used to calculate the rate of change of turbidity deviation at the current sampling time. The membership weight of the turbidity deviation change rate benchmark. The relative viscosity coefficient at the current sampling time. It is a logarithmic function, which increases the weight and strengthens trend prediction.

[0053] Formula for correcting the upper limit of the positive output of the integral gain compensation:

[0054] In the formula, This is the upper limit of the positive output of the compensation amount for the integral gain at the current sampling time; The upper limit of the positive output for the compensation amount of the offline calibrated integral gain; It is a natural exponential function, used to ensure that the upper limit of the integral tightens smoothly as viscosity increases, thus avoiding integral saturation.

[0055] For low-viscosity conditions, based on the characteristic of Stokes' sedimentation law that the settling resistance of suspended particles decreases as the viscosity of the liquid decreases, the correction objective is to: reduce the adjustment intensity, weaken the anticipatory adjustment, prevent over-adjustment oscillation, and avoid excessive adjustment under low-viscosity conditions. The corresponding formula is as follows: Formula for correcting the upper limit of the positive output of the proportional gain compensation:

[0056] In the formula, The upper limit of the positive output of the compensation amount of the proportional gain at the current sampling time; The upper limit of the positive output of the proportional gain compensation reference for offline calibration. It is a natural exponential function, due to the low viscosity conditions. <1 indicates that the index is always negative, and the multiplier factor is less than 1. This ensures that as the viscosity decreases, the upper limit of the proportional gain output is smoothly compressed. This is used to ensure that the correction amplitude narrows smoothly as the viscosity decreases, avoiding sudden oscillations. This achieves the narrowing of the upper limit, reduces the adjustment force, and avoids speed overshoot in low-viscosity liquids.

[0057] Turbidity deviation change rate membership weighted correction formula:

[0058] In the formula, The weighted average of the membership degree of the rate of change of turbidity deviation at the current sampling time. The membership weight of the turbidity deviation change rate benchmark. It is a natural exponential function, used to ensure that the weight decays smoothly as viscosity decreases, avoiding overshoot caused by premature adjustment, thus reducing the weight and weakening premature adjustment.

[0059] Formula for correcting the upper limit of the positive output of the integral gain compensation:

[0060] In the formula, This is the upper limit of the positive output of the compensation amount for the integral gain at the current sampling time; The upper limit of the positive output for the compensation amount of the offline calibrated integral gain; As a natural exponential function, it has low resistance to material-liquid separation and fast system response under low viscosity conditions, and the turbidity deviation converges quickly. By tightening the upper limit of the integral gain, it avoids speed overshoot and turbidity oscillation caused by excessive integral action.

[0061] This step establishes quantitative adaptation rules for viscosity and PID control characteristics, enabling pre-coordination from feed condition prediction to closed-loop control parameters, and providing adaptive correction constraints for subsequent three-input fuzzy inference.

[0062] Based on the data from step S1, the parameters calibrated by the centrifuge during the equipment commissioning phase under ideal operating conditions are as follows: , , , Given the equivalent flow resistance viscosity coefficient at the current sampling time. ; Calculate the relative viscosity coefficient This falls under a high-viscosity operating condition. The system automatically triggers the high-viscosity operating condition correction rule and calculates the following: The upper limit of the positive output of the proportional gain compensation amount The threshold has been relaxed compared to the benchmark value; Turbidity deviation change rate membership degree weighted weight This is an improvement over the benchmark value. Integral gain compensation upper limit of positive output It is tighter than the benchmark value; The result obtained here , and This provides corrective constraints to adapt to high-viscosity conditions for subsequent three-input fuzzy inference.

[0063] S3: Calculate the turbidity deviation and rate of change based on the fixed process target turbidity, and perform three-input fuzzy inference in combination with viscosity feedforward correction rules to convert it into the gain compensation amount of the PID controller.

[0064] After establishing a feedforward correction rule that conforms to the rheological characteristics of the liquid at the current sampling time, the traditional PID controller with constant parameters cannot cope with the time-varying and complex error evolution due to the obvious dynamic hysteresis and high nonlinearity of the pipeline fluid in the entire pear juice processing process.

[0065] Therefore, this step utilizes a three-input fuzzy inference system with viscosity feedforward correction to tune the proportional and integral gain parameters of the turbidity deviation online, and finally calculates the output speed adjustment command, thereby establishing a complete and highly responsive adaptive negative feedback control chain.

[0066] This step is the integration of feedforward regulation and closed-loop control. It combines the parameters corrected by feedforward with the turbidity deviation and deviation change rate of the closed-loop feedback to achieve synergy between proactive regulation and precise correction.

[0067] Simple feedforward regulation cannot cope with random fluctuations in turbidity deviation, such as a sudden increase in impurities in the liquid. Simple closed-loop control has a hysteresis defect. This step introduces the equivalent flow resistance viscosity coefficient as a feedforward factor into the fuzzy inference model and combines it with the deviation data from the closed-loop feedback to achieve dual control of feedforward prediction and closed-loop correction.

[0068] Feedforward correction ensures that parameters are adapted to the current operating conditions, while closed-loop feedback ensures that deviations are accurately converged. The two work together to achieve precise matching of adjustment intensity, response speed, operating conditions, and deviations. Finally, the output is a PID gain compensation amount adapted to the current operating conditions, providing a precise basis for the generation of subsequent speed adjustment commands. This ensures that the system can still operate stably under complex operating conditions, meeting the stability requirements of industrial production.

[0069] S31: Obtain the turbidity deviation, the rate of change of turbidity deviation, and the equivalent flow resistance viscosity coefficient at each sampling time under all operating conditions for pear juice centrifugation and impurity removal. Using the three-standard-deviation rule of statistics, set the fluctuation range of each variable and construct the fuzzy domain of the corresponding variable.

[0070] For example, the turbidity deviation at the current sampling time The rate of change of turbidity deviation at the current sampling time Turbidity deviation reflects the absolute physical distance between the centrifuge's separation effect at the current moment and the fixed qualified standard. The larger the positive value, the more turbid the effluent is and the further it deviates from the target. The turbidity of the discharge pipe section at the current sampling time. Set a target turbidity for a fixed process; The rate of change of turbidity deviation This is the turbidity deviation read from the register at the previous sampling time. The actual turbidity of the discharge pipe section... Turbidity higher than the fixed process target At that time, a large positive error is generated. If at the same time A large positive error indicates a large and rapidly deteriorating error, potentially indicating a sudden influx of high-impurity clumps into the feed end and insufficient centrifugal separation. In this case, the fuzzy rule matrix will deduce the need for a large proportional gain compensation to instantly increase the rotational speed; conversely, if... A negative value indicates that the current separation is excessive. A negative value indicates that the trend of excessive separation is intensifying. The fuzzy rule will output a negative gain compensation to reduce the system's response sensitivity and avoid overshoot.

[0071] In this embodiment, the fuzzy universe and fuzzy subset division of each variable are as follows: The fuzzy universe of discourse for turbidity deviation E corresponds to [-3,3]NTU, which is divided into five fuzzy subsets: negative large (NB), negative small (NS), zero (ZE), positive small (PS), and positive large (PB). The fuzzy universe of discourse for the rate of change of turbidity deviation EC is [-2,2] NTU / s, which is divided into five fuzzy subsets: negative large (NB), negative small (NS), zero (ZE), positive small (PS), and positive large (PB). Equivalent flow resistance viscosity coefficient The fuzzy universe of discourse corresponds to the range [0.005, 0.05] Pa. s is divided into five fuzzy subsets: negative large (NB, corresponding to extremely low viscosity), negative small (NS, corresponding to low viscosity), zero (ZE, corresponding to reference viscosity), positive small (PS, corresponding to medium and high viscosity), and positive large (PB, corresponding to high viscosity). proportional gain compensation The fuzzy domain range corresponds to [-5,5], and is divided into 5 fuzzy subsets: negative large (NB), negative small (NS), zero (ZE), positive small (PS), and positive large (PB). Compensation amount of integral gain The fuzzy domain range corresponds to [-2,2], and is divided into 5 fuzzy subsets: negative large (NB), negative small (NS), zero (ZE), positive small (PS), and positive large (PB).

[0072] Based on the aforementioned fuzzy subsets of input and output variables, a three-input IF-THEN fuzzy rule matrix is ​​formulated, and a three-input fuzzy inference model is pre-built offline. The core rule logic is as follows: when When E is positive (high viscosity), and EC is positive, Output is positive. The output is small to significantly improve the adjustment power and avoid lag, while constraining the integral gain to prevent saturation; when When E is negative (very low viscosity), E is positive (small), and EC is positive (small), Output positive small, The output is small to narrow the adjustment range and avoid overshoot; when When the viscosity is zero (reference viscosity), E is zero, and EC is zero, Output zero. The output is zero, maintaining stable operation of the baseline parameters; the remaining rules are formulated according to the core logic that viscosity is positively correlated with the upper limit of the positive output of the proportional gain and negatively correlated with the upper limit of the positive output of the integral gain, forming a complete matrix of 125 three-input fuzzy rules.

[0073] S32: The turbidity deviation, the rate of change of turbidity deviation, and the equivalent flow resistance viscosity coefficient at the current sampling time are used as input variables and imported into the three-input fuzzy inference model that has been built offline in advance.

[0074] For the membership function, this scheme adopts the mature triangular membership function already available in the field of industrial fuzzy control. Based on the fuzzy universe of discourse of each input / output variable, the parameters of the triangular membership function corresponding to each fuzzy subset are calibrated offline. The specific process is as follows: Determine the fuzzy universe of discourse for each variable and divide the fuzzy subsets (e.g., NB, NS, ZE, PS, PB); calibrate the three vertex parameters (left boundary) of the triangular membership function for each fuzzy subset. Peak point Right boundary The calibration of vertex parameters follows the principles of "covering the universe of discourse, overlapping adjacent subsets, and adapting to industrial control precision." It is determined offline based on historical operating data, without the need for complex calculations. This is a routine operation in industrial fuzzy control. The calibrated triangular membership function is then embedded into the PLC controller as the basic module for fuzzy inference and called in real time.

[0075] The mathematical expression for the membership function:

[0076] In the formula, It refers to the precise numerical value of the input / output variable, which in this invention can be the turbidity deviation. Turbidity deviation change rate Equivalent flow resistance viscosity coefficient It can also be the compensation amount for the output proportional gain. Compensation amount of integral gain , The left boundary parameters of the corresponding fuzzy subsets are calibrated offline, such as turbidity deviation. The left boundary of the "positive small PS" fuzzy subset, , The peak point parameters corresponding to the fuzzy subset are typical representative values ​​of that fuzzy subset obtained through offline calibration, such as turbidity deviation. The peak point of the "positive small PS" fuzzy subset, ; The right boundary parameters of the corresponding fuzzy subsets are calibrated offline, such as turbidity deviation. The right boundary of the "positive small PS" subset, ; It is the membership function value, or simply membership, with a value range of [0,1]. It is the core parameter that maps the precise value of the input / output to the degree of belonging to the corresponding fuzzy subset, and can be calculated in real time using the formula above.

[0077] Membership functions are the core foundation of fuzzy inference and the only bridge connecting precise measurement data from industrial sites with fuzzy rule language, enabling the transformation from precise values ​​to fuzzy language: Turbidity deviation, viscosity coefficient, and other parameters collected from industrial sites are continuous precise values, such as turbidity deviation of 1.14 NTU and viscosity of 0.0221 Pa. Certain turbidity deviations, such as s, cannot be directly identified by fuzzy rules. However, by calculating the membership function, the precise value can be transformed into the degree to which it belongs to a certain fuzzy subset. For example, a turbidity deviation of 1.14 NTU, after calculation, has a membership degree of 0.76, meaning it belongs to the "positive small PS" subset by 76%, allowing fuzzy inference to execute normally. The value of the membership degree directly corresponds to the trigger weight of the fuzzy rule. The closer the membership degree is to 1, the higher the trigger strength of the rule for the corresponding fuzzy subset, and the greater the impact on the gain compensation of the final output. Conversely, the closer the membership degree is to 0, the lower the trigger strength of the corresponding rule, and the smaller the impact.

[0078] S33: Combine the viscosity feedforward correction rule to perform adaptive weighted correction on the membership function, perform three-input fuzzy inference to obtain fuzzy output, and use the centroid method to defuzzify the fuzzy output to obtain the compensation amount of the proportional gain and integral gain of the PID controller respectively.

[0079] Weighted by membership degree of the rate of change of turbidity deviation The original membership degree can be directly calculated from the membership function. Multiplication enables adaptive correction of membership degrees, thereby adjusting the triggering intensity of fuzzy rules and allowing fuzzy inference to adapt to the current material viscosity conditions. This is also the core implementation of feedforward regulation in the fuzzy inference process.

[0080] First, during the equipment debugging phase, based on the fuzzy universe and fuzzy subsets of each variable, the corresponding fuzzy subset is calibrated. Three vertex parameters, for example, turbidity deviation The fuzzy subset corresponding to the "positive small PS" And solidify the parameters and the above mathematical expressions into In the fuzzy inference module of the controller.

[0081] Then, at each sampling time, Turbidity deviation was collected Turbidity deviation change rate Equivalent flow resistance viscosity coefficient After obtaining the precise value, the triangular membership function of the corresponding variable is automatically invoked to calculate the membership value corresponding to each fuzzy subset. For example, Substituting the membership function of the "positive small PS" fuzzy subset, we can calculate... ; Next, a weighted correction is performed, adjusting only the rate of change of turbidity deviation. The membership value is corrected to reduce turbidity deviation. Equivalent flow resistance viscosity coefficient The membership values ​​remain unchanged to ensure the stability of the quality judgment criteria. The specific correction method is as follows: , Turbidity deviation change rate The corrected membership value, It is the rate of change of turbidity deviation. The membership value, The membership weight is the weighted weight of the rate of change of turbidity deviation at the current sampling time.

[0082] Subsequently, participate in fuzzy reasoning: the corrected ,as well as and The uncorrected membership values ​​of each are substituted into the three-input fuzzy rule matrix to trigger the corresponding rule. for , for , for The fuzzy output is used in the calculation of the fuzzy output, and finally the fuzziness is defuzzified using the centroid method to obtain the compensation amount of the proportional gain of the PID controller. Compensation amount for integral gain .

[0083] It should be noted that the reason for only weighting the membership function of the turbidity deviation change rate is as follows: From a process compliance perspective, turbidity deviation is the absolute difference between the actual output turbidity and the fixed process qualification standard. It is the core benchmark for determining whether a product meets food safety and corporate quality standards. If its membership function is weighted and modified, it will change the logic for determining whether the product is qualified, which violates the quality control requirements of the target turbidity set by the fixed process. Therefore, the membership function of turbidity deviation must remain fixed to ensure the uniqueness and stability of the quality judgment benchmark.

[0084] From a control theory perspective: turbidity deviation serves as the steady-state error benchmark for closed-loop control, belonging to the state feedback quantity. Its core function is to measure the absolute difference between the current control effect and the acceptable standard, requiring the benchmark stability to be maintained. The turbidity deviation change rate, on the other hand, represents the dynamic trend of error change, belonging to the trend prediction quantity. The viscosity of the feed liquid directly determines the inertia, hysteresis, and dynamic response speed of the centrifugal separation system. Higher viscosity results in greater system hysteresis, necessitating a higher prediction weight for the error change rate to compensate for hysteresis in advance. Conversely, lower viscosity leads to a faster system response, requiring a lower prediction weight for the error change rate to avoid overshoot caused by premature adjustment. Therefore, adaptive weighting of only the membership degree of the turbidity deviation change rate ensures both the stability of the quality benchmark and adaptive adaptation to the system's dynamic characteristics.

[0085] This step achieves an adaptive transformation from rigid error to flexible control parameters by performing three-input fuzzy inference with viscosity feedforward correction and centroid defuzzification, thus enabling real-time tuning of PID control gain to counteract time-varying operating conditions.

[0086] Continuing from the data in step S2, the target turbidity is known to be fixed in the process settings. The actual turbidity of the discharge pipe section was collected by an online turbidity meter at the discharge end. Calculate the turbidity deviation at the current sampling time. Assuming the turbidity deviation at the previous sampling time rate of change of turbidity deviation Equivalent flow resistance viscosity coefficient Combining the feedforward correction rule, the membership function of the turbidity deviation change rate is corrected using a weighted correction formula. The three input variables are then imported into the offline-constructed three-input fuzzy inference model, and the model is calculated using a weighted correction of the membership function. The turbidity deviation is mapped to a "positive small (PS)" fuzzy subset. The rate of change is mapped to a fuzzy subset of "positive small (PS)". The viscosity coefficient is mapped to a fuzzy subset of “positive (PB)”.

[0087] The three-input fuzzy rule matrix is ​​set to include at least the following rules: when the equivalent flow resistance viscosity coefficient is positive, the turbidity deviation is positive, and the rate of change of the turbidity deviation is positive, the proportional gain compensation output is positive and the integral gain compensation output is negative; when the equivalent flow resistance viscosity coefficient is negative, the turbidity deviation is positive, and the rate of change of the turbidity deviation is positive, the proportional gain compensation output is negative and the integral gain compensation output is negative; when the equivalent flow resistance viscosity coefficient is zero, the turbidity deviation is zero, and the rate of change of the turbidity deviation is zero, the proportional gain compensation output is zero and the integral gain compensation output is zero.

[0088] After defuzzification calculation using the center-of-gravity method, the system accurately outputs the compensation amount for the proportional gain. Compensation amount of integral gain Simultaneously, the positive output upper limit of the compensation amount of the proportional gain calculated in step S2. The upper limit of the positive output of the integral gain compensation. The upper limit is set, and a limiting check is performed to confirm that the output value is within a safe range. This indicates that under the current high viscosity conditions, slight deviation, and stable divergence trend, the system amplifies the proportional gain compensation by using viscosity as a feedforward input variable for fuzzy inference, thereby increasing the adjustment strength to avoid lag. At the same time, it appropriately controls the integral gain to prevent integral saturation, fully adapting to the separation characteristics of the current high viscosity liquid.

[0089] S4: Perform incremental proportional-integral calculations to generate speed regulation commands based on the compensation amounts of the proportional gain and integral gain.

[0090] After obtaining the PID parameter compensation values ​​that have been tuned in real time through fuzzy inference, this step aims to convert these control coefficients into physical speed commands that can directly drive the centrifuge mechanical spindle frequency converter, thus establishing a complete and highly responsive adaptive negative feedback control chain.

[0091] This step is the execution stage of feedforward regulation, which converts the feedforward-corrected PID gain parameters into speed commands that can directly drive the equipment, thus enabling the advanced control effect of feedforward regulation to be executed.

[0092] This step uses incremental proportional-integral (PI) calculations to combine the feedforward-corrected proportional and integral gains with the turbidity deviation and deviation change rate from the closed-loop feedback to generate precise speed adjustment commands. Feedforward correction ensures that the commands are adapted to the current liquid viscosity, while incremental calculations ensure the smoothness of the commands, avoiding equipment shocks and turbidity oscillations caused by sudden speed changes. This achieves a complete process from feedforward prediction to parameter correction and command implementation.

[0093] Specifically, the compensation amounts of the proportional gain and integral gain are superimposed with the preset base proportional gain and base integral gain of the PID controller to obtain the adaptive proportional gain and adaptive integral gain at the current sampling time. Based on the adaptive proportional gain at the current sampling time, and combined with the difference between the turbidity deviation at the current sampling time and the turbidity deviation at the previous sampling time, the proportional adjustment increment is calculated. Based on the adaptive integral gain at the current sampling time, and combined with the turbidity deviation at the current sampling time, the integral adjustment increment is calculated. The proportional adjustment increment and the integral adjustment increment are superimposed to obtain the centrifuge speed adjustment command at the current sampling time.

[0094] Specifically, it satisfies the following relationship:

[0095] In this relation, This indicates the speed adjustment command, reflecting the absolute action amount that the system sends to the frequency converter at the current moment, requiring the centrifuge spindle speed to increase or decrease. A positive value indicates an acceleration requirement, and a negative value indicates a deceleration requirement. The preset base proportional gain for the PID controller, The preset integral gain for the PID controller, This is the compensation amount for the proportional gain at the current sampling time. This is the compensation amount for the integral gain at the current sampling time. This represents the turbidity deviation at the current sampling time. This represents the turbidity deviation at the previous sampling time. This constitutes the corrected adaptive proportional gain. This constitutes the modified adaptive integral gain.

[0096] In this formula, the first term The first term represents the proportional adjustment increment, reflecting the system's instantaneous damping response to changes in turbidity deviation. A larger value indicates a rapidly worsening error, requiring the system to instantly apply a large proportional gain to counteract the torque and suppress divergence. The second term... The integral adjustment increment represents the steady-state elimination thrust of the system against the current actual absolute deviation.

[0097] This relationship is a discrete-time incremental PI control model. The first term reflects the system's instantaneous damping response to the trend of deviation change, and the second term reflects the system's thrust in eliminating steady-state issues. By adding these two independent physical control actions, when the feed turbidity suddenly deteriorates, causing a sharp increase in turbidity deviation, The ratio becomes extremely large, and the first proportional effect instantly takes over, rapidly increasing the speed command output to suppress deviation divergence; if, at this time, due to the high viscosity of the feed liquid and the large separation resistance, the turbidity fails to reach the standard for a long time, leading to... When the value remains positive, the second integral action begins to exert its force steadily, continuously replenishing the steady-state driving force after the proportional action has decayed. The combined effect of these two actions results in the final calculated speed adjustment command. An increase corresponds to a situation where the centrifuge's impurity removal efficiency is low at the current sampling time, and there is an urgent need to quickly inject kinetic energy to improve the centrifugal settling torque; conversely, when the deviation turns negative, various linkages generate negative adjustment commands, and the system steadily and smoothly implements frequency reduction operation.

[0098] This step achieves a high-precision closed-loop mapping of PID parameters to actuator control quantities by arithmetically superimposing the proportional and integral terms based on the error increment, resulting in smooth output, rapid response, and no steady-state error accumulation in the time domain for generating speed commands.

[0099] Continuing from the data in step S3, the turbidity deviation at the current sampling time is known. Turbidity deviation at the previous sampling time The compensation amount of the proportional gain output by fuzzy inference Compensation amount of integral gain Obtain the base proportional gain set by the PID controller. Basic integral gain ; Adaptive proportional gain: Adaptive integral gain: Proportional adjustment increment: Integral adjustment increment: The two control quantities are superimposed to obtain the speed regulation command output at the current sampling time. .

[0100] This means that in order to eliminate the turbidity deviation of 1.14 NTU and adapt to the separation characteristics of the current high-viscosity liquid, the PID controller decides to add an execution command of 15.504 r / min to the current speed of the centrifuge. Compared with the fixed parameter scheme, this improves the adjustment power and avoids the adjustment lag under high viscosity conditions.

[0101] S5: Calculate the safety factor based on the direction of the speed adjustment command and the current of the centrifuge, and combine it with the actual speed to generate a coordinated speed command to complete the coordinated adjustment of the speed for centrifuging and impurity removal of pear juice.

[0102] After receiving the speed adjustment command at the current sampling moment, it is further considered that when the centrifuge is running at high speed, occasional abnormal conditions such as material clumping impact and sudden flow changes may easily cause the centrifuge's main shaft drive motor to overload.

[0103] If an acceleration command is issued directly without conditions, there is a risk of triggering the equipment's overcurrent protection shutdown. This step strictly introduces a forced safety limit load reduction logic based on the actual operating load, and outputs the final coordinated speed command to solidify the equipment's safe operation boundary.

[0104] This step is a collaborative safeguard between feedforward regulation and equipment safety. Its core significance lies in ensuring the safety of feedforward regulation operations, preventing equipment overload caused by acceleration commands from feedforward correction, and balancing regulation effectiveness with equipment safety. Through segmented safety bypass logic, the speed regulation command after feedforward correction is subject to safety limits. When the equipment load is safe, the feedforward regulation command is fully released to ensure the feedforward effect is implemented; when the equipment load approaches overload, the acceleration command is attenuated to prevent equipment failure.

[0105] Specifically, the direction of the centrifuge's speed adjustment command at the current sampling moment is obtained, a segmented safety bypass logic is constructed based on the direction of the speed adjustment command, and a safety factor is calculated by combining the centrifuge's current at the current sampling moment with the preset equipment safety current warning threshold. When the speed adjustment command is a positive acceleration command, the safety factor is calculated based on the preset load reduction attenuation penalty coefficient and the magnitude of the centrifuge current exceeding the equipment safety current warning threshold at the current sampling time, using an exponential attenuation rule negatively correlated with the magnitude. When the speed adjustment command is a negative deceleration command or a zero value maintenance command, the safety bypass logic is executed, the safety factor is set to 1, and no attenuation processing is performed on the speed adjustment command.

[0106] For cases where a positive acceleration command is triggered, the safety factor satisfies the following relationship:

[0107] In this formula, This represents the safety margin, reflecting the proportion of acceleration commands allowed by the system within the limits of the underlying hardware. The closer this value is to 1, the more ample the safety margin, and the commands can be sent almost without loss. The closer it is to 0, the more severely the device is overloaded, and the acceleration commands will be blocked. The preset load reduction attenuation penalty coefficient is used to adjust the severity of the thresholding logic triggering. This represents the current of the centrifuge at the current sampling moment. The safe current warning threshold for the centrifuge. This is a function to find the maximum value. It is a natural exponential function.

[0108] In this formula, The current overload difference indicates the degree of urgency by which the electrical actuator approaches its rated limit. The calculation extracts the positive overload amplitude that only takes effect when the actual current exceeds the safety threshold. This formula establishes a nonlinear suppression strategy using a negatively correlated natural exponential function, when the actual load current... Exceeding the warning threshold The greater the magnitude, the greater the penalty. The linear amplification is rapid due to the external exponential function. The negative mapping characteristic of this property causes the safety factor to decay exponentially and approach zero. This means that at the current sampling moment, the centrifuge is on the verge of overload. At this time, no matter how serious the deviation of the turbidity control index at the front end is or how urgent the acceleration request issued by the algorithm is, the underlying system will intercept any acceleration command that increases the load on the equipment to ensure that the centrifuge spindle operates within the safety threshold. Conversely, if the actual operating current is stable and far below the warning threshold, the overload amplitude is reduced to zero by the max function. This means the device is within a safe range, and the calculated acceleration commands will be fully issued and executed. For cases triggering negative acceleration commands, the safety factor meets the following requirements. .

[0109] This relationship adopts the exponential penalty function model / barrier function method in optimization theory. In industrial electromechanical protection scenarios, when the physical load approaches the hardware limit, the output is controlled by exponential decay, which is a classic soft safety interception mechanism to ensure the electrical safety of equipment.

[0110] Then, the speed adjustment command at the current sampling time is multiplied by the safety factor to obtain the speed adjustment increment after safety limit; the speed adjustment increment after safety limit is superimposed with the actual speed of the centrifuge at the current sampling time to generate the coordinated speed command of the centrifuge at the current sampling time, and the centrifuge spindle motor is driven to perform coordinated speed adjustment based on the coordinated speed command.

[0111] Specifically, the following relationship is satisfied:

[0112] In the relation, This refers to the coordinated rotation speed command of the centrifuge at the current sampling moment. This represents the current actual speed of the drive motor. This is the centrifuge speed adjustment command at the current sampling time. This is for the safety factor.

[0113] when When the value is less than or equal to 0, the safety bypass logic is set to not perform attenuation, i.e. When the load current approaches or exceeds the limit, the acceleration gain will rapidly decay exponentially until it reaches zero. This mechanism strictly limits the system to only performing frequency reduction or maintaining the original speed protection actions under abnormal heavy load conditions. Moreover, the speed reduction command will be issued quickly and without hindrance when overloaded, avoiding the dangerous logic of using overload fault parameters to amplify the control gain, and ultimately outputting a safe coordinated speed command.

[0114] This step multiplies the speed adjustment command with the safety factor and then weights and superimposes it with the actual speed in the final stage. This achieves the highest priority limit constraint of the rigid electrical load at the bottom layer on the flexible control command at the top layer. It achieves the synergistic protection effect of forcibly terminating the overload speed-up of the equipment under extreme heavy load and sudden working conditions, and preventing major safety shutdown accidents.

[0115] Continuing from the output data of step S4, the basic speed adjustment command is calculated and issued based on the current sampling time. If the value is greater than 0, the command direction is determined to be a positive acceleration command, thus immediately triggering the overload penalty attenuation calculation link. The actual operating speed of the centrifuge spindle at the current sampling moment is read in real time via the general-purpose inverter bus. and the synchronously transmitted current Current warning threshold descent attenuation penalty coefficient Since the current of 32A is far below the warning threshold of 40A, there is no overload. Substituting... The function returns an excess of 0; calculate the safety factor: Subsequently, the system performs a safety limiting calculation, and the speed adjustment increment after safety limiting is as follows: The final coordinated speed command issued: .

[0116] This means that, with sufficient safety margin in the equipment, the system fully releases acceleration commands adapted to high viscosity conditions, ensuring that the adjustment intensity and response speed are perfectly matched to the current liquid characteristics, quickly eliminating turbidity deviations, and avoiding the production of unqualified liquids.

[0117] like Figure 2As shown in the figure, this diagram visually demonstrates the comparison between the optimized fuzzy PID algorithm of this invention and the traditional fuzzy PID algorithm in controlling the turbidity of pear juice output under conditions of sudden changes in feed viscosity. Around the 10th second of centrifuge operation, a batch of high-pectin, viscous liquid suddenly rushed in. The traditional fuzzy PID algorithm, relying solely on post-event feedback and with fixed parameters, experienced a lag in centrifugal speed adjustment, causing the output turbidity to deviate from the process target of 20 NTU. The peak value instantly rose to over 45 NTU, and the system underwent severe overshoot and oscillation for over 20 seconds before gradually stabilizing. In contrast, the solution of this invention benefits from the feedforward correction mechanism of the equivalent flow resistance viscosity coefficient. The system can detect changes in the rheological properties of the material in advance and adaptively tune the control parameters. When encountering the same load impact, it not only effectively suppressed the turbidity peak to approximately 32 NTU, but also significantly improved the system convergence speed, accompanied only by extremely small, industrially realistic damped oscillations. It quickly returned to the set baseline and operated stably within a short time, solving the problems of adjustment lag and unstable oscillations that easily occur under complex and sudden changes in operating conditions.

[0118] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for coordinated adjustment of centrifugal speed for removing impurities in pear juice based on fuzzy PID algorithm, characterized in that, include: During the centrifugation and impurity removal process of pear juice, any sampling time is taken as the current sampling time. The pressure difference of the feed pipe section of the centrifuge, the flow rate of the pear juice in the feed pipe section, the turbidity of the discharge pipe section, the process set target turbidity, and the current and actual speed of the centrifuge are collected. Based on pressure difference and flow rate, the equivalent flow resistance viscosity coefficient of pear juice is calculated by analyzing the fluid resistance characteristics through Hagen-Poiseuille's law and combining the physical parameters of the feed pipe section. The turbidity deviation is calculated based on the target turbidity set in the process and the turbidity of the discharge pipe section, and the rate of change of turbidity deviation is calculated by combining the turbidity deviation at the previous sampling time. The equivalent flow resistance viscosity coefficient is used as a feedforward correction factor, and together with the turbidity deviation and the rate of change of turbidity deviation, it is used as an input variable to perform fuzzy inference and defuzzification processing, which is converted into the compensation amount of the proportional gain and integral gain of the PID controller. Based on the compensation amount of the proportional gain and the compensation amount of the integral gain, incremental proportional-integral operation is performed to generate the centrifuge speed adjustment command, and coordinated adjustment is performed in combination with the centrifuge current and actual speed.

2. The method for coordinated adjustment of centrifugal speed for removing impurities from pear juice as described in claim 1, characterized in that, Based on pressure difference and flow rate, and through the fluid resistance characteristic analysis of Hagen-Poiseuille's law, combined with the physical parameters of the feed pipe section, the equivalent flow resistance viscosity coefficient of the pear juice liquid is calculated, including: Obtain the inner diameter and length of the feed pipe section. Based on the flow rate, pressure difference, pipe inner diameter, and pipe length at the current sampling time, construct the Hagen-Poiseuille law formula and calculate the equivalent flow resistance viscosity coefficient at the current sampling time.

3. The method for coordinated adjustment of centrifugal speed for removing impurities from pear juice as described in claim 1, characterized in that, The equivalent flow resistance viscosity coefficient is used as a feedforward correction factor, and together with turbidity deviation and the rate of change of turbidity deviation, it is used as input variables to perform fuzzy inference and defuzzification processing, transforming it into the compensation amounts for the proportional gain and integral gain of the PID controller. This includes: using the turbidity deviation, the rate of change of turbidity deviation, and the equivalent flow resistance viscosity coefficient at the current sampling time as input variables of the fuzzy inference model, and importing them into a pre-constructed offline three-input fuzzy inference model; the fuzzy inference model has a built-in membership function that maps the exact input value to the degree of fuzzy subset membership; based on the equivalent flow resistance viscosity coefficient, adaptive weighted correction is performed on the membership function of the rate of change of turbidity deviation, and three-input fuzzy inference is performed to obtain the fuzzy output; the centroid method is used to defuzzify the fuzzy output to obtain the compensation amounts for the proportional gain and integral gain of the PID controller, respectively.

4. The method for coordinated adjustment of centrifugal speed for removing impurities from pear juice as described in claim 3, characterized in that, The pre-built offline three-input fuzzy inference model is constructed as follows: The input variables of the fuzzy inference model are determined to be turbidity deviation, the rate of change of turbidity deviation, and the equivalent flow resistance viscosity coefficient; the output variables of the fuzzy inference model are determined to be the compensation values ​​of the proportional gain and integral gain of the PID controller; fuzzy universes are defined for the input variables and the output variables respectively, and fuzzy subsets are divided for the fuzzy universes of the input variables and the output variables respectively. Based on fuzzy subsets of input variables and output quantities, a three-input IF-THEN fuzzy rule matrix is ​​formulated to construct a three-input fuzzy inference model.

5. The method for coordinated adjustment of centrifugal speed for removing impurities from pear juice as described in claim 4, characterized in that, The method for setting fuzzy universes for input variables and output quantities separately is as follows: The turbidity deviation, the rate of change of turbidity deviation, and the equivalent flow resistance viscosity coefficient were obtained at various sampling times under all operating conditions during the centrifugation and impurity removal of pear juice. The fluctuation range of each of the turbidity deviation, the rate of change of turbidity deviation, and the equivalent flow resistance viscosity coefficient was determined by the three-standard deviation rule. The fuzzy domains of turbidity deviation, the rate of change of turbidity deviation, and the equivalent flow resistance viscosity coefficient are each set to an interval that completely covers their respective fluctuation ranges; the fuzzy domains of the compensation amounts of proportional gain and integral gain are each set to an interval that completely covers the adjustment ranges of the PID controller for proportional gain and integral gain, respectively.

6. The method for coordinated adjustment of centrifugal speed for removing impurities from pear juice as described in claim 4, characterized in that, The method for dividing the fuzzy universes of discourse of input variables and output quantities into fuzzy subsets is as follows: The fuzzy universes of discourse for turbidity deviation, rate of change of turbidity deviation, equivalent flow resistance viscosity coefficient, compensation amount of proportional gain, and compensation amount of integral gain are all divided into five fuzzy subsets: negative large, negative small, zero, positive small, and positive large.

7. The method for coordinated adjustment of centrifugal speed for removing impurities from pear juice as described in claim 3, characterized in that, An adaptive weighted correction is performed on the membership function of the turbidity deviation change rate, including: determining the membership weight based on the equivalent flow resistance viscosity coefficient, and using the membership weight to correct only the membership function of the turbidity deviation change rate; setting positive output upper limits for both the proportional gain compensation and the integral gain compensation; the equivalent flow resistance viscosity coefficient is positively correlated with both the membership weight and the positive output upper limit of the proportional gain compensation; if the equivalent flow resistance viscosity coefficient is not less than the preset benchmark equivalent flow resistance viscosity coefficient, the equivalent flow resistance viscosity coefficient is negatively correlated with the positive output upper limit of the integral gain compensation; if the equivalent flow resistance viscosity coefficient is less than the preset benchmark equivalent flow resistance viscosity coefficient, the equivalent flow resistance viscosity coefficient is positively correlated with the positive output upper limit of the integral gain compensation.

8. The method for coordinated adjustment of centrifugal speed for removing impurities from pear juice as described in claim 4, characterized in that, Incremental proportional-integral (PI) calculations are performed based on the compensation amounts of the proportional gain and integral gain to generate centrifuge speed control commands. This includes: superimposing the compensation amounts of the proportional gain and integral gain with the preset base proportional gain and base integral gain of the PID controller, respectively, to obtain the adaptive proportional gain and adaptive integral gain at the current sampling time; calculating the proportional adjustment increment based on the adaptive proportional gain at the current sampling time and the difference between the turbidity deviation at the current sampling time and the turbidity deviation at the previous sampling time; calculating the integral adjustment increment based on the adaptive integral gain at the current sampling time and the turbidity deviation at the current sampling time; and superimposing the proportional adjustment increment and the integral adjustment increment to obtain the centrifuge speed control command at the current sampling time.

9. The method for coordinated adjustment of centrifugal speed for removing impurities from pear juice as described in claim 1, characterized in that, The process of coordinating speed regulation by combining the centrifuge's current and actual rotation speed includes: acquiring the direction of the centrifuge's speed regulation command at the current sampling moment; constructing a segmented safety bypass logic based on the direction of the speed regulation command; calculating a safety factor by combining the centrifuge's current at the current sampling moment with a preset equipment safety current warning threshold; multiplying the speed regulation command at the current sampling moment by the safety factor to obtain the speed regulation increment after safety limiting; superimposing the speed regulation increment after safety limiting with the actual rotation speed of the centrifuge at the current sampling moment to generate a coordinated speed command for the centrifuge at the current sampling moment; and driving the centrifuge's spindle motor to perform coordinated speed regulation based on the coordinated speed command.

10. The method for coordinated adjustment of centrifugal speed for removing impurities from pear juice as described in claim 9, characterized in that, A segmented safety bypass logic is constructed based on the direction of the speed adjustment command. The safety factor is calculated by combining the current of the centrifuge at the current sampling moment with the preset equipment safety current warning threshold. This includes: when the direction of the speed adjustment command is a positive acceleration command, the safety factor is calculated based on the preset load reduction attenuation penalty coefficient and the magnitude by which the current of the centrifuge at the current sampling moment exceeds the equipment safety current warning threshold, using an exponential attenuation rule that is negatively correlated with the magnitude; when the direction of the speed adjustment command is a negative deceleration command or a zero value maintenance command, the safety bypass logic is executed, the safety factor is set to 1, and no attenuation processing is performed on the speed adjustment command.

Images