Collagen peptide mechanical seal water circulation system

Abstract

The invention relates to the technical field of fluid circulation of collagen peptide production equipment, in particular to a collagen peptide mechanical seal water circulation system which comprises an intelligent control unit, a cooling unit and an ultrasonic descaling unit. Converting the vibration spectrum into a peptide molecule aggregation tendency coefficient, correcting a viscosity predicted value to generate a flow control weight, allocating resources to a cooling unit according to a hydrolysis risk thermodynamic diagram, directionally cooling a high-risk area, activating an ultrasonic descaling unit and reducing the flow when the aggregation coefficient continuously rises and exceeds a limit, the system generates cooperative control quantity through rolling of a three-dimensional dynamic weight matrix, synchronously adjusts the rotating speed of a water pump, the opening degree of a cooling valve and ultrasonic power, solves the problems of unstable flowing, peptide chain hydrolysis and aggregation scaling, and guarantees production stability and product quality.

Description

Technical Field

[0001] This invention relates to the field of fluid circulation technology for collagen peptide production equipment, and more specifically, to a collagen peptide machine sealing water circulation system. Background Technology

[0002] Fluid circulation in collagen peptide production equipment is a crucial technology. In modern industrial production of collagen peptides, this technology is key to ensuring production continuity and product quality. Through stable mechanical seal water circulation, it can provide cooling and lubrication for mechanical seal components, preventing damage to the sealing surface due to overheating caused by friction. It can also prevent external impurities from entering the circulation system and contaminating the collagen peptide solution, while maintaining the stability of solution flow and reducing damage to peptide molecular structure. It is an important support for promoting the upgrading of collagen peptide production from extensive management to refined assurance, and is widely applicable to collagen peptide production lines in the food and health product fields. However, existing collagen peptide mechanical sealing and circulation technologies suffer from core problems such as a lack of multi-factor synergistic regulation and insufficient adaptation to peptide molecule characteristics. This problem stems from three practical limitations: First, the flow rate is not controlled by incorporating the rheological properties of the collagen peptide solution; relying solely on a fixed pump speed cannot address viscosity fluctuations caused by changes in shear rate, easily leading to uncontrolled turbulent flow or sluggish flow. Second, the correlation between peptide chain hydrolysis and temperature and pH is ignored; relying solely on cooling is insufficient to accurately suppress peptide chain dissociation caused by localized high temperatures on the sealing surface, affecting product quality. Third, there is a lack of mechanisms for predicting and intervening in peptide molecule aggregation; aggregation trends are not identified through vibration signals, making it easy for aggregates to adhere to the sealing surface, forming scale, exacerbating equipment wear and clogging pipelines. These problems can have a cascading effect: Inaccurate flow control leads to unstable solution flow, which may damage the peptide molecular structure due to excessive shearing, or increase local residence time due to slow flow. Insufficient hydrolysis inhibition can cause peptide chain breakage, reducing the uniformity of product molecular weight. Aggregation and scaling can shorten the life of mechanical seals, increase equipment maintenance costs, and may also interrupt production due to pipeline blockage. Ultimately, existing technologies cannot meet the requirements of collagen peptide production for rheological adaptation, precise temperature control, and anti-aggregation and blockage mechanical seal water circulation. There is an urgent need for an intelligent circulation system that can couple rheological and thermodynamic properties and dynamically respond to changes in peptide molecules. To solve this technical problem, we have provided a collagen peptide mechanical seal water circulation system. Summary of the Invention

[0003] The purpose of this invention is to provide a collagen peptide machine-sealed water circulation system to solve the problems mentioned in the background art.

[0004] 1. Since the flow rate is not controlled by the rheological properties of the peptide solution, a fixed rotation speed is difficult to cope with the unstable flow caused by viscosity fluctuations. Therefore, this case establishes a rheological sub-model through an intelligent control unit, generates flow control weights based on the viscosity prediction value, and increases the pump speed at low viscosity to adapt to the rheological properties of the peptide solution and maintain stable flow.

[0005] 2. Due to the lack of intervention to predict peptide aggregation, aggregates are prone to scale formation and damage to equipment. Therefore, this case uses the vibration spectrum to convert the aggregation tendency coefficient. When the coefficient continuously rises for an extended period, the ultrasonic descaling unit is activated and the flow rate is reduced. This can intervene in aggregation in a timely manner and reduce scale formation and equipment wear.

[0006] To achieve the above objectives, a collagen peptide mechanical seal water circulation system is provided, including an intelligent control unit that achieves rheological-thermodynamic coupling optimization control through the following steps: real-time acquisition of fluid pressure pulsation data, mechanical seal axial vibration spectrum, sealing surface temperature gradient distribution, solution concentration, and pH value; A rheological sub-model was constructed based on fluid pressure pulsation data and solution concentration to establish a viscosity response relationship library of collagen peptide solution under dynamic shear rate. A thermodynamic sub-model was constructed by embedding collagen peptide chain hydrolysis activation energy parameters based on the temperature gradient distribution of the sealing surface and pH value. The axial vibration spectrum of the mechanical seal is converted into a peptide aggregation tendency coefficient, which is used as a dynamic correction factor input to the rheological sub-model to generate a corrected real-time viscosity prediction value. Flow control weights are generated based on the real-time viscosity prediction value. When the real-time viscosity prediction value is lower than a set threshold, the water pump speed is increased to control priority. At the same time, temperature control weights are generated based on the hydrolysis risk distribution map output by the thermodynamic sub-model. If the area of ​​the high-temperature region exceeds a preset percentage, a preset proportion of control resources is allocated to the cooling unit. The peptide aggregation tendency coefficient is converted into an anti-clogging compensation command. When the peptide aggregation tendency coefficient rises continuously for more than a preset monitoring time, the ultrasonic descaling unit is activated and the flow setpoint is reduced simultaneously. Finally, a three-dimensional dynamic weight matrix is ​​formed based on flow control weight, temperature control weight, and anti-clogging compensation command. A collaborative control quantity is generated once every fixed time interval to synchronously adjust the water pump speed, cooling valve opening, and ultrasonic power, thereby minimizing system energy loss.

[0007] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. A rheological sub-model is constructed through an intelligent control unit. A viscosity response relationship library is established by combining fluid pressure pulsation data and solution concentration. The viscosity prediction value is corrected by the peptide molecule aggregation tendency coefficient and a flow control weight is generated. The pump speed is increased at low viscosity to achieve the technical effect of adapting to the rheological characteristics of peptide solution. This solves the problem that a fixed speed is difficult to cope with viscosity fluctuations and causes unstable flow. It has the advantages of avoiding excessive shearing that damages peptide molecules and maintaining stable solution flow.

[0008] 2. By embedding peptide chain hydrolysis activation energy parameters into the intelligent control unit, a thermodynamic sub-model is constructed. Based on the temperature gradient and pH value of the sealing surface, a hydrolysis risk heat map is generated. Resources are allocated to the cooling unit according to the proportion of high-temperature areas, and the cooling of high-risk areas is enhanced in a targeted manner. This achieves the technical effect of precisely inhibiting peptide chain hydrolysis, solving the problem of local high temperature and peptide chain dissociation caused by single cooling. It has the advantages of ensuring the uniformity of product molecular weight and improving the quality of collagen peptides.

[0009] 3. By analyzing the entropy value of the characteristic frequency band of the vibration spectrum, the vibration spectrum is converted into the peptide molecule aggregation tendency coefficient. When the coefficient continuously rises for more than the preset time, the ultrasonic descaling unit is activated and the flow rate is reduced simultaneously. This achieves the technical effect of timely intervention in molecular aggregation, solving the problems of easy scaling of aggregates, equipment damage and pipeline blockage. It has the advantages of reducing seal wear, reducing maintenance costs and avoiding production interruption. Attached Figure Description

[0010] Figure 1 This is an overall block diagram of the present invention.

[0011] The meanings of the labels in the diagram are as follows: 1. Intelligent control unit; 2. Cooling unit; 3. Ultrasonic descaling unit. Detailed Implementation

[0012] 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 embodiments of the present invention, and not all embodiments. 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.

[0013] This invention provides a collagen peptide mechanical water circulation system; please refer to [link / reference]. Figure 1 As shown, the intelligent control unit 1 achieves rheological-thermodynamic coupling optimization control through the following steps: real-time acquisition of fluid pressure pulsation data, mechanical seal axial vibration spectrum, sealing surface temperature gradient distribution, solution concentration and pH value; A rheological sub-model is constructed based on fluid pressure pulsation data and solution concentration to establish a library of viscosity response relationships of collagen peptide solutions under dynamic shear rates. Specifically, the construction of the rheological sub-model based on fluid pressure pulsation data and solution concentration includes: The dynamic shear rate variation trend of fluid is analyzed by pressure pulsation waveform feature extraction technology. Combined with solution concentration matching collagen peptide multi-level shear thinning response surface library, this surface library is calibrated by simulating peptide chain depolymerization kinetics experiments under different temperature gradients. The dynamic shear rate is divided into critical thinning region, stable flow region and turbulent flow control region, and viscosity transition thresholds for each region are established.

[0014] A viscosity response database of collagen peptide solutions under dynamic shear rates was established. Furthermore, a dynamic calibration technique for shear thinning critical points was adopted. When the amplitude decay rate of the pressure pulsation waveform exceeds the preset fluctuation tolerance, the system automatically switches to the response surface library sub-library corresponding to the concentration and temperature combination, and outputs the viscosity prediction benchmark value that matches the current shear rate in real time. At the same time, the viscosity transition threshold is corrected according to the proportion of high-frequency energy in the axial vibration spectrum of the mechanical seal.

[0015] In the collagen peptide mechanical sealing water circulation system, the prerequisite for the intelligent control unit 1 to achieve rheological-thermodynamic coupled optimization control is to first accurately collect and understand the core monitoring parameters, and then construct a rheological sub-model based on the key parameters to understand the flow characteristics of the collagen peptide solution. First, it is necessary to clarify the meaning and function of each core parameter. Among them, fluid pressure pulsation data is the quantitative data of the periodic or non-periodic small fluctuations in pressure over time when the collagen peptide solution flows in the circulation system. It is collected by a high-frequency pressure sensor installed on the inner wall of the pipe. Its function is to reflect the dynamic state of fluid flow. The amplitude and frequency of pressure pulsation are positively correlated with the shear intensity of the fluid, providing a direct basis for subsequent analysis of the dynamic shear rate. The mechanical seal axial vibration spectrum is the spectrum of vibration energy distribution at different frequencies when the mechanical seal assembly vibrates along the axial direction, i.e., perpendicular to the sealing surface. It is collected by a piezoelectric acceleration sensor on the sealing end cap. Its function is to indirectly reflect the aggregation state of peptide molecules, because molecular aggregation changes the force distribution on the sealing surface, thus affecting the high-frequency energy of the vibration spectrum. The volume ratio and temperature gradient distribution of the sealing surface are the temperature differences and trends at different locations on the contact surface of the dynamic and static rings of the mechanical seal. These are acquired through an embedded infrared temperature sensor array and are used to assess the risk of peptide chain hydrolysis. Excessive temperature gradients can lead to localized overheating, accelerating the dissociation of the β-sheet structure of collagen peptides. Solution concentration is the mass fraction of collagen peptides per unit volume of circulating solution, detected in real-time by an online refractometer. Its function is to correlate with the rheological properties of the solution; different concentrations result in different strengths of intermolecular interactions between peptide molecules, leading to significant differences in the viscosity's response to shear rate. pH value is an indicator of the solution's acidity or alkalinity, acquired through an online pH electrode. Its function is to regulate the energy barrier of the peptide chain hydrolysis reaction. A pH deviation from neutral (7.0±0.5) will accelerate or slow down hydrolysis, affecting solution stability and viscosity characteristics. Based on these parameters, the intelligent control unit 1 first constructs a rheological sub-model using fluid pressure pulsation data and solution concentration to understand the viscosity change law of the collagen peptide solution under dynamic shear, providing a basis for subsequent flow control. The specific implementation method is as follows: The first step in constructing a rheological sub-model is to analyze the dynamic shear rate variation trend of the fluid using pressure pulsation waveform feature extraction technology. This technology is analyzed as follows: The first step is raw data preprocessing. First, 10,000 raw pressure data points collected by the pressure sensor within 10 seconds are subjected to a 50Hz low-pass filter to remove environmental interference from pipe vibration. Then, a peak detection algorithm is used to extract the peak Pmax and trough Pmin of each pressure pulsation cycle, and the pulsation amplitude ΔP = Pmax - Pmin is calculated. Simultaneously, the time interval T of each cycle is recorded. Based on the pressure pulsation-shear rate correlation model calibrated through pre-tested pipe flow field simulation experiments, the formula is γ = k × ΔP / T, where k is the pipe characteristic coefficient. Based on the system pipe diameter of 50mm and the collagen peptide solution density of 1.05g / cm³, it is pre-set to 0.02s⁻¹ / (MPa·ms). Substituting ΔP and T of each cycle into the formula, the instantaneous shear rate γt is obtained. In the above case, γt = 0.02 × 0.04 / 20 = 4 × 10⁻¹ 4 s⁻¹, a 5-period moving average is performed on the instantaneous shear rate over 50 consecutive periods to generate a dynamic shear rate trend curve. If the curve changes from 3×10⁻ 4 s⁻¹ gradually increases to 5×10⁻ 4 s⁻¹ indicates that the shear force on the fluid is continuously increasing, and vice versa. Through this process, easily collected pressure pulsation parameters can be transformed into dynamic shear rates that reflect the internal stress state of the fluid, laying the foundation for subsequent viscosity analysis. After obtaining the trend of dynamic shear rate changes, it is necessary to combine the solution concentration with a multi-level shear thinning response surface library of collagen peptides. This surface library is the core reference of the model, and its construction and matching process must strictly adhere to the experimental data: the construction of the surface library requires simulating peptide chain depolymerization kinetics experiments under different temperature gradients and different solution concentrations in the laboratory. The viscosity values ​​corresponding to different shear rates under each condition are measured using a rotational rheometer. A three-dimensional response surface is constructed with shear rate γ as the X-axis, solution concentration C as the Y-axis, and viscosity η as the Z-axis. Each temperature gradient corresponds to a sub-surface. All sub-surfaces are integrated to form a complete surface library. The library also synchronously stores auxiliary data such as peptide molecule depolymerization rate and molecular chain length distribution under each condition. The matching process is as follows: The current solution concentration Ccurrent = 8% is obtained from an online refractometer. Sub-surfaces corresponding to concentrations of 7.5% to 8.5% within ±0.5% of the concentration deviation are selected from the surface library. The dynamic shear rate trend curve is analyzed and compared point by point with the selected sub-surfaces to find the surface point that best matches the current shear rate γcurrent = 100s⁻¹ and concentration Ccurrent = 8%. The Z-axis viscosity value of this point is the preliminary matching result. At the same time, combined with the current average temperature gradient of the sealing surface of 35℃, the matching result is temperature compensated by the correction formula ηcorr = ηmatch × (1 + 0.01 × (Tcurrent - Tref)) (Tref is the sub-surface reference temperature of 30℃) to ensure that the matching result fits the actual operating conditions of the system. When constructing this surface library, it has been calibrated by simulating peptide chain depolymerization kinetics experiments under different temperature gradients. The dynamic shear rate is divided into the critical thinning region, the stable flow region, and the turbulent runoff control region, and the viscosity transition threshold of each region is established. The division basis and region definition need to be determined based on experimental data statistics. Measurements using a rotational rheometer revealed that when the shear rate γ ≤ 50 s⁻¹, the viscosity decreases by less than 5% with increasing shear rate. At this point, the peptide chains do not undergo significant depolymerization, and the fluid flow is stable. This region is defined as the stable flow region, where the viscosity range varies with concentration. For example, at an 8% concentration, the viscosity remains stable at 50-80 mPa·s. When 50 s⁻¹ < γ ≤ 300 s⁻¹, the viscosity suddenly decreases by more than 5% with increasing shear rate, and the decreasing trend gradually flattens. This is a shear thinning phenomenon, where the peptide chains depolymerize under shearing, reducing intermolecular entanglement. This is defined as the critical thinning region, with a viscosity range of 30%–80% of the lower limit of the stable flow region. For example, at an 8% concentration, the viscosity decreases from 80 mPa·s to 24 mPa·s. When γ > 300 s⁻¹, the viscosity decreases abruptly again by more than 10%, and the fluid flow in the pipe... Significant turbulent noise is observed, with excessive depolymerization and even structural damage of peptide molecular chains. This is defined as the turbulent uncontrolled region, where the viscosity is below 30% of the lower limit of the critical thinning region. For example, at an 8% concentration, the viscosity is below 24 mPa·s. The viscosity transition threshold is the critical viscosity value that divides the region. Each concentration corresponds to two sets of thresholds: one is the transition threshold η1 between the stable flow region and the critical thinning region (80 mPa·s at 8% concentration), and the other is the transition threshold η2 between the critical thinning region and the turbulent uncontrolled region (24 mPa·s at 8% concentration). These thresholds are determined by averaging the viscosity abrupt change points from five sets of parallel experiments and stored in the region division module of the surface library. This module is used to determine the flow region in which the fluid is located in real time. To further improve the model accuracy, a viscosity response relationship library for collagen peptide solutions under dynamic shear rates needs to be established. This relationship library is a refinement and expansion of the response surface library, as detailed below: Within each flow range, experimental data were supplemented with high precision at shear rate intervals of 1 s⁻¹, concentration intervals of 0.2%, and temperature intervals of 1 °C, forming a four-dimensional correlation table of shear rate-concentration-temperature-viscosity to ensure that the data density meets the real-time prediction requirements. Secondly, a specific viscosity response equation was established for each range. In the stable flow range, the linear equation η=a×γ+b was used, where a and b are functions of concentration and temperature; at 8% concentration and 35 °C, a=-0.16 and b=88. In the critical thinning range, the power-law equation η=k×γⁿ was used (n is the flow behavior exponent; n<1; at 8% concentration and 35 °C, k=1200 and n=-0.4). In the turbulent flow control range, the exponential equation η=η0×e^(-c×γ) was used, where η0 is the initial viscosity and c is the attenuation coefficient; at 8% concentration and 35 °C, η0=24 mPa·s and c=5×10⁻⁻⁻⁴. 4 The equation parameters are obtained by fitting experimental data using the least squares method. Finally, an outlier correction module is added. When the deviation between the measured viscosity and the predicted value under a certain condition combination exceeds 10%, the system automatically calls three adjacent sets of data for interpolation correction to ensure the accuracy of the relational database and provide fine data support for subsequent viscosity prediction. During model operation, a dynamic calibration technique for the shear thinning critical point is also required. When the amplitude decay rate of the pressure pulsation waveform exceeds the preset fluctuation tolerance, the system automatically switches to the response surface library sub-library corresponding to the concentration and temperature combination. The shear thinning critical point is the critical shear rate value of the fluid entering the critical thinning region from the stable flow region. The amplitude decay rate of the pressure pulsation waveform is the ratio of the difference between the amplitudes ΔP1 and ΔP2 of two consecutive cycles to ΔP1 (α=|ΔP2-ΔP1| / ΔP1×100%). The preset fluctuation tolerance is 15%. When α>15%, it indicates that the pressure pulsation is unstable and the shear rate may be close to the critical point. It is necessary to switch to the high-precision sub-library. The switching process is as follows: First, collect data at the current concentration of 8% and temperature of 35℃. Then, locate a sub-library in the surface library with a concentration deviation of ±0.2% and a temperature deviation of ±1℃, corresponding to concentrations of 7.8%–8.2% and temperatures of 34–36℃. Next, switch the viscosity prediction benchmark from the original surface library to this sub-library to ensure prediction accuracy near the critical point. After switching sub-libraries, a viscosity prediction benchmark value matching the current shear rate needs to be output in real time. The specific process is as follows: Extract the instantaneous shear rate γnow = 65s⁻¹ at the current moment. Search the sub-database for the interval 64-66s⁻¹ containing 65s⁻¹. Read the viscosity range of this interval as 65-68 mPa·s, calculate the mean as 66.5 mPa·s, and apply the power-law equation for the critical thinning region η = 1200 × γ^(-0.4) to calculate the predicted value ηpred = 1200 × 65^(-0.4) ≈ 66.2 mPa·s. Take a 50% weighted average of the mean and the predicted value to obtain the final viscosity prediction. The baseline value is measured as (66.5 + 66.2) / 2 ≈ 66.35 mPa·s. The above steps are repeated every 100 ms to update the baseline value, ensuring dynamic tracking of operating conditions. Simultaneously, to address the impact of peptide aggregation on viscosity characteristics, the viscosity transition threshold needs to be corrected based on the proportion of high-frequency energy in the mechanical seal's axial vibration spectrum. The proportion of high-frequency energy β is calculated, and the 800-1200 Hz characteristic frequency band strongly correlated with the Brownian motion of peptide molecules is extracted from the vibration spectrum. The total energy Ehigh of this frequency band and the full frequency band are calculated. The ratio of the total spectral energy Etotal, β = Ehigh / Etotal × 100%, for example, Ehigh = 0.5 J, Etotal = 2.5 J, β = 20%, then the threshold is adjusted according to the correction formula: η1' = η1 × (1 + 0.5 × β / 100%) (η1 is the threshold between the stable region and the thinned region, with a correction coefficient of 0.5, because aggregation has a greater impact on this threshold), η2' = η2 × (1 + 0.3 × β / 100%) (η2 is the threshold between the thinned region and the runaway region, with a correction coefficient of 0.5). With a positive coefficient of 0.3, for example, if the original η1=80mPa·s and η2=24mPa·s, after correction, η1'=80×(1+0.5×20 / 100%)=88mPa·s and η2'=24×(1+0.3×20 / 100%)=25.44mPa·s, this dynamic correction makes the threshold fit the molecular aggregation state, avoids misjudgment caused by changes in rheological properties due to aggregation, and ensures that the rheological sub-model always accurately reflects the actual flow characteristics of collagen peptide solution.

[0016] A thermodynamic sub-model was constructed by embedding collagen peptide chain hydrolysis activation energy parameters based on the temperature gradient distribution and pH value of the sealing surface. Specifically, this sub-model includes: The temperature gradient distribution of the sealing surface is input into the peptide chain hydrolysis spatial probability mapping algorithm. A hydrolysis risk heat map is generated based on the dissociation activation energy parameter of the collagen peptide β-sheet structure. The pH value is dynamically corrected by adjusting the hydrolysis reaction energy barrier height to modify the color distribution of the heat map. When the local temperature continuously exceeds the critical dissociation temperature and the pH deviates from the neutral range, a high-risk block is generated and marked as a priority cooling target.

[0017] After completing the construction of the rheological sub-model to understand the flow characteristics of collagen peptide solution, in order to further avoid the impact of peptide chain hydrolysis on the system's operational stability (peptide chain hydrolysis can lead to abnormal changes in solution viscosity, scaling on mechanical seal surfaces, and even blockage of circulation pipelines), the intelligent control unit 1 needs to construct a thermodynamic sub-model based on the temperature gradient distribution and pH value of the sealing surface, embedding the activation energy parameters of collagen peptide chain hydrolysis. This allows for precise temperature control by quantifying the hydrolysis risk. The specific implementation method is as follows: First, accurate data on the temperature gradient distribution of the sealing surface needs to be obtained. This data is collected by an embedded infrared temperature sensor array installed on the contact surface of the dynamic and static rings of the mechanical seal. The sensors are arranged with a grid spacing of 5mm×5mm, covering a sealing surface with a diameter of 50mm, for a total of 100 collection points. The sampling frequency is 10Hz, the temperature measurement range is 20-150℃, and the accuracy is ±0.1℃. The collected raw data needs to be filtered by a 3-point moving average to eliminate instantaneous temperature fluctuations caused by mechanical vibration. Then, the temperature value of each sensor is associated with the spatial coordinates (radial r, circumferential θ) of the sealing surface through coordinate mapping to form a coordinate-temperature... The two-dimensional temperature gradient distribution matrix is ​​obtained. After acquiring the pre-processed temperature gradient distribution of the sealing surface, it is input into a peptide chain hydrolysis spatial probability mapping algorithm. This algorithm is a mathematical modeling tool based on the thermodynamic Arrhenius equation and combined with the structural characteristics of collagen peptides to transform the spatial temperature distribution into a peptide chain hydrolysis probability distribution. Its core is to achieve spatial visualization of risk by quantifying the driving effect of temperature at different spatial locations on peptide chain hydrolysis. The key to applying this algorithm is to first clarify the dissociation activation energy parameter of the β-sheet structure of collagen peptides. This parameter refers to the β-sheet structure in the collagen peptide molecule, which is the main characteristic of collagen peptides. Secondary structure, maintained by hydrogen bonds between peptide chains, accounts for 35%–40% of the molecular secondary structure. Its stability directly determines whether the peptide chain is easily hydrolyzed: when the β-sheet structure is intact, the peptide bonds are enclosed within the structure, making hydrolysis difficult; once the β-sheet structure dissociates and the peptide bonds are exposed, the hydrolysis rate increases by 3–5 times. The minimum energy required for dissociation needs to be accurately determined in the laboratory using differential scanning calorimetry (DSC): collagen peptide samples are placed in a DSC instrument under different temperatures (20–100℃) and different pH conditions (2–12) to monitor the heat flow changes during β-sheet dissociation, combined with Arrhenius... The equation k = Ae^(-Ea / (RT)) is used, where k is the dissociation rate constant of the β-sheet structure, A is the pre-exponential factor, Ea is the dissociation activation energy, R is the gas constant, and T is the absolute temperature. Fitting calculations yield Ea values ​​under different operating conditions. For example, at pH=7 (neutral) and 30℃, Ea = 82 kJ / mol; at pH=5 (weakly acidic) and 30℃, Ea decreases to 65 kJ / mol; and at pH=9 (weakly alkaline) and 30℃, Ea decreases to 70 kJ / mol. These data are stored in the algorithm parameter library according to the temperature-pH-Ea correlation for real-time calculation. Based on the above parameters, the peptide chain hydrolysis spatial probability mapping algorithm generates a hydrolysis risk heatmap according to the following steps: The process involves mesh generation and temperature matching. The sealing surface is divided into 100 mesh units (5mm × 5mm) according to the sensor acquisition grid. Each mesh unit corresponds to a unique temperature value T, extracted from the temperature gradient distribution matrix. The corresponding pH value is then collected in real-time by an online pH electrode. The pH deviation between the solution near the sealing surface and the system's bulk pH is ≤0.1. Based on the T and pH of the mesh unit, the corresponding Ea value is retrieved from the parameter library and substituted into the Arrhenius equation to calculate the β-sheet dissociation rate constant k. Here, A is taken as a fixed value calibrated in the laboratory, 1.2 × 10¹² s⁻¹, R = 8.314 J / (mol·K), T = 58 + 273.15 = 331.15 K, resulting in k = 1.2 × 10¹² × e^( -80000 / (8.314×331.15))≈1.2×10¹²×e^(-29.0)≈1.5×10⁻²s⁻¹. Based on the residence time t of the peptide chain in the grid cell, calculated from the circulation flow rate, when the circulation flow rate is 5L / min, t≈0.6s. The hydrolysis probability of each grid cell is calculated using the hydrolysis probability P=1-e^(-k×t). Substituting k=1.5×10⁻²s⁻¹ and t=0.6s, we get P=1-e^(-0.009)≈0.0089 (i.e., 0.89%). The color range for the hydrolysis probability is set as follows: P<1% is light green (low risk), 1%~5% is dark green (relatively low risk), 5%~15% is yellow (medium risk), and 15%~30% is... Orange (higher risk), >30% is red (higher risk). The P-value of each grid cell is matched to its corresponding color level. Then, the color levels of all grid cells are stitched together according to the spatial coordinates of the sealing surface to generate a two-dimensional hydrolysis risk heatmap. The horizontal axis of the heatmap represents the circumferential angle of the sealing surface, and the vertical axis represents the radial distance. This visually displays the differences in hydrolysis risk at different locations. For example, a grid cell at the edge of the sealing surface has a high frictional temperature (T=65℃) and a hydrolysis probability P=25%, corresponding to the orange color level, while the central area has T=50℃ and a P=0.5%, corresponding to the light green color level. After generating the initial hydrolysis risk heatmap, the hydrolysis reaction barrier height needs to be adjusted by pH value to dynamically correct the color level distribution of the heatmap. The hydrolysis reaction barrier refers to the barrier that a peptide chain needs to overcome to undergo a hydrolysis reaction. The higher the energy barrier, the more difficult the hydrolysis reaction is to occur. pH affects the energy barrier height by altering the ionization state of peptide bonds: when pH is in the neutral range (7.0 ± 0.5), the amino group (-NH2) and carboxyl group (-COOH) in the peptide bond are in an electrically neutral state, maintaining the stability of the hydrogen bonds in the β-sheet structure, resulting in the highest energy barrier for hydrolysis. When pH deviates from the neutral range (pH < 6.5 or pH > 7.5), the amino group becomes protonated (-NH3⁺) or the carboxyl group becomes deprotonated (-COO⁻), weakening the hydrogen bond forces and making the β-sheet structure more prone to dissociation. Consequently, the energy barrier for hydrolysis decreases. For example, at pH = 5.5, the energy barrier is 22% lower than under neutral conditions, and at pH = 8.5, the energy barrier is 18% lower. The specific correction process is as follows: First, obtain the current system pH value. Then, query the preset pH-barrier correction coefficient table in the algorithm. Based on the positive correlation between the barrier and the hydrolysis probability, multiply the original hydrolysis probability P by the reciprocal of the correction coefficient to obtain the corrected hydrolysis probability P'. Finally, adjust the color level of the grid cell according to P' and update the color level display of the grid cell in the heatmap to ensure that the heatmap can accurately reflect the impact of pH value on hydrolysis risk and avoid misjudgment of risk due to ignoring the pH factor. After completing the pH correction of the heatmap, the system will automatically identify high-risk blocks and mark priority cooling targets. First, set the critical dissociation temperature, which refers to the temperature threshold at which the β-sheet structure of collagen peptides begins to dissociate in large quantities. This temperature is determined by laboratory testing and is 60℃ at neutral pH. For every 0.5 deviation of pH from neutral, the critical dissociation temperature is set. The critical dissociation temperature decreases by 1℃. Then, all grid cells in the heat map are traversed. If a grid cell simultaneously meets the three conditions of local temperature exceeding the critical dissociation temperature for three consecutive sampling cycles and pH value deviating from the neutral range and hydrolysis probability P'>15%, the area where the grid cell is located is determined to be a high-risk block. The system will select the block with a white dashed line on the heat map, record its center coordinates, and transmit the coordinate information to the partition solenoid valve control module of cooling unit 2, marking it as a priority cooling target. Cooling unit 2 will immediately increase the opening of the solenoid valve corresponding to the area from the normal 30% to 60%, directionally enhance the coolant injection flow, quickly reduce the temperature of the area, suppress β-sheet structure dissociation and peptide chain hydrolysis, and ensure the stable operation of the mechanical sealing surface.

[0018] The axial vibration spectrum of the mechanical seal is converted into a peptide aggregation tendency coefficient. Specifically, the vibration spectrum characteristic frequency band entropy analysis method is used to extract the energy distribution of characteristic frequency bands in the axial vibration spectrum of the mechanical seal that are strongly correlated with the Brownian motion of peptide molecules. The energy entropy value within the characteristic frequency band is calculated and compared with a preset aggregation threshold to obtain the comparison result. The aggregation tendency coefficient increases in a negative correlation ratio according to the comparison result. At the same time, it is cross-validated with the real-time viscosity prediction value output by the rheological sub-model.

[0019] Based on the construction of a rheological sub-model to understand the flow characteristics of collagen peptide solutions and a thermodynamic sub-model to avoid the risk of peptide chain hydrolysis, peptide molecule aggregation can lead to abnormally high solution viscosity, increased frictional resistance of mechanical seal surfaces, and even pipeline blockage. Therefore, it is necessary to convert the axial vibration spectrum of the mechanical seal into a peptide molecule aggregation tendency coefficient to quantify the degree of aggregation in real time, providing a basis for subsequent anti-clogging control. Specifically, this is achieved using the entropy value analysis method of the characteristic frequency band of the vibration spectrum, as follows: First, accurate acquisition of the axial vibration spectrum of the mechanical seal is required. This spectrum is obtained by a piezoelectric accelerometer installed on the sealing end cap. The sensor has a range of ±5g, a frequency response range of 0-2kHz, and a sampling frequency of 1kHz to ensure complete capture of high-frequency vibration signals related to peptide molecule motion. The acquired raw spectrum data needs to be filtered by a 200Hz high-pass filter to remove low-frequency vibration interference from the pipeline. Then, a fast Fourier transform is used to convert the time-domain signal into a frequency-domain energy distribution spectrum, laying the foundation for subsequent feature frequency band extraction. Next, feature frequency bands strongly correlated with the Brownian motion of peptide molecules are extracted. The Brownian motion of peptide molecules refers to the random, irregular motion of collagen peptide molecules in a circulating solution due to the influence of molecular thermal motion. The intensity of this motion is directly related to the molecular aggregation state. When peptide molecules are not aggregated, the intermolecular interactions are weak, and the Brownian motion is intense, which can produce irregular micro-impacts on the mechanical seal surface, leading to specific... The vibrational energy distribution within a fixed frequency band is dispersed. When peptide molecules aggregate, they form larger aggregates, hindering Brownian motion, weakening the impact, and increasing regularity. Consequently, the vibrational energy distribution within the corresponding frequency band tends to concentrate. Previous experiments have verified that this characteristic frequency band is concentrated between 800 and 1200 Hz. Within this band, the vibrational energy generated by the Brownian motion of peptide molecules accounts for over 60% and is unaffected by vibrations from other components such as motors and pumps. Therefore, 800–1200 Hz is set as the characteristic analysis frequency band, and this band is evenly divided into 20 sub-bands to facilitate subsequent refined energy distribution calculations. The energy entropy value within the characteristic frequency band is then calculated. Entropy is a physical quantity that measures the degree of disorder in energy distribution. The more dispersed the energy distribution (corresponding to vigorous Brownian motion and no aggregation of peptide molecules), the higher the entropy value; the more concentrated the energy distribution (corresponding to aggregation of peptide molecules and weakened Brownian motion), the lower the entropy value. The specific calculation process involves three steps: The energy of each sub-band is calculated. The vibration energy Ei within each 20Hz sub-band is obtained through integration. The normalized energy percentage pi is calculated using the formula pi = Ei / ΣEi, where ΣEi is the sum of the 20 sub-band energies. The energy entropy value H is calculated by substituting the data into the Shannon entropy formula: H = -Σ(pi × lnpi), where i ranges from 1 to 20 and ln is the natural logarithm. The calculated energy entropy value is then compared with a preset aggregation threshold, which is determined through laboratory calibration experiments. At different levels of peptide aggregation, the aggregate size was measured using a laser particle size analyzer. Aggregates <100nm were considered non-aggregated, 100-500nm were considered slightly aggregated, and >500nm were considered heavily aggregated. Simultaneously, energy entropy values ​​for corresponding characteristic frequency bands were collected. Statistical analysis showed that the entropy value was stable at 0.8-1.2 for non-aggregated aggregates, decreased to 0.6-0.8 for slightly aggregated aggregates, and was below 0.6 for heavily aggregated aggregates. Therefore, two preset aggregation thresholds were set: 0.8 (the critical value distinguishing between non-aggregated and slightly aggregated aggregates) and 0.6 (the critical value distinguishing between slightly aggregated and heavily aggregated aggregates). During comparison... If the calculated energy entropy value H ≥ 0.8, it is determined to be a non-aggregated state; if 0.6 ≤ H < 0.8, it is determined to be a slightly aggregated state; and if H < 0.6, it is determined to be a heavily aggregated state. This provides a clear state basis for the subsequent calculation of the aggregation tendency coefficient. Based on the above comparison results, the aggregation tendency coefficient is calculated according to the negative correlation ratio. An increasing negative correlation ratio indicates that the lower the energy entropy value, the more severe the aggregation, and the larger the aggregation tendency coefficient. The coefficient value range is set to 0-1, where 0 represents theoretically no aggregation and 1 represents extreme heavy aggregation. The specific calculation process uses piecewise linear mapping to ensure accuracy. When H ≥ 0.8 (no aggregation), the aggregation tendency coefficient k = 0.2, setting a baseline value to avoid the coefficient being 0, which would lead to no response in subsequent anti-clogging control. When 0.6 ≤ H < 0.8 (mild aggregation), k = 0.2 + (0.8 - H) × 1.5, where 0.8 - H is the degree to which the entropy value deviates from the no-aggregation threshold, and 1.5 is the experimentally calibrated incremental slope to ensure that the coefficient increases linearly with the degree of aggregation. When H < 0.6 (severe aggregation), k = 0.5 + (0.6 - H) × 2.0, increasing the incremental slope to 2.0 to accelerate the coefficient growth to match the risk level of severe aggregation. At the same time, the upper limit of the coefficient is set to 0.9 to avoid the coefficient reaching 1, which would lead to excessive ultrasonic descaling power and damage to the mechanical seal surface. This ensures that the coefficient can accurately reflect the degree of aggregation without exceeding the safe execution range of subsequent control commands. To further ensure the accuracy of the aggregation tendency coefficient, it is necessary to correlate it with the rheological sub-model input. The real-time viscosity prediction values ​​are cross-validated. Peptide aggregation leads to increased intermolecular entanglement in the solution, which in turn increases viscosity. Therefore, there is a positive correlation between the two. If the aggregation tendency coefficient k > 0.5 (slight or higher aggregation), but the real-time viscosity prediction value output by the rheological sub-model does not exceed the upper limit of viscosity in the stable flow region at the corresponding concentration, it is determined that there may be vibration sensor signal interference or local aggregation, which does not affect the overall viscosity. Vibration spectrum data needs to be collected again and the entropy value is recalculated. If the results of two consecutive calculations still show k > 0.5 and the viscosity is normal, the sensor self-check program is triggered. Conversely, if the viscosity prediction value exceeds the upper limit of the stable flow region but k < 0.5, it is necessary to check whether the concentration matching of the rheological sub-model is accurate. Through this two-way cross-validation, the misjudgment of aggregation state caused by single parameter error is eliminated, ensuring that the aggregation tendency coefficient can truly and reliably guide the generation of subsequent anti-clogging compensation commands.

[0020] As a dynamic correction factor input to the rheological sub-model, it generates a corrected real-time viscosity prediction value, specifically achieved through an aggregation tendency-viscosity coupling compensation mechanism: When the aggregation tendency coefficient exceeds the dynamic calibration threshold, negative offset compensation of the viscosity prediction value is triggered. The offset compensation amount is adaptively adjusted according to the mapping relationship between the energy entropy value of the characteristic frequency band and the solution concentration, and the offset viscosity prediction value is synchronously fed back to the thermodynamic sub-model to solve the peptide chain hydrolysis rate in the high temperature region.

[0021] After obtaining the peptide aggregation tendency coefficient through the entropy value analysis of the characteristic frequency band of the vibration spectrum, this coefficient is not only used to determine the aggregation state, but also needs to serve as a dynamic correction factor to feed back into the rheological sub-model. Because peptide aggregation changes the actual rheological properties of the solution, relying solely on the original viscosity prediction value can easily lead to flow control deviations. For example, if aggregation causes the actual viscosity to be higher than the predicted value, adjusting the pump speed according to the predicted value will cause insufficient shear. Therefore, an aggregation tendency-viscosity coupling compensation mechanism is needed to generate a corrected real-time viscosity prediction value, ensuring that the output of the rheological sub-model closely matches the actual state of the solution. The specific implementation method is as follows: First, the setting logic of the dynamic calibration threshold needs to be clarified. This threshold is the critical threshold that triggers viscosity compensation. The degree of aggregation needs to be calibrated by considering the influence of aggregation on viscosity under different solution concentrations. Experiments have shown that when the aggregation tendency coefficient is below 0.3, the peptide aggregate particle size is <100nm, and the impact on the overall solution viscosity is less than 5%, requiring no compensation. When the coefficient is ≥0.3, the aggregate particle size increases with the coefficient, and the viscosity increase exceeds 5%, requiring compensation. Therefore, the dynamic calibration threshold is uniformly set to 0.3. The system will compare the current aggregation tendency coefficient with this threshold in real time. If the coefficient is ≥0.3 for two consecutive sampling periods, negative offset compensation of the viscosity prediction value will be triggered. This negative offset compensation does not reduce the actual viscosity value, but rather corrects the prediction bias of the rheological sub-model. Because the rheological sub-model is built based on the assumption of uniform dispersion of peptide molecules, and aggregation can cause the actual viscosity to be higher than the model's predicted value, negative offset compensation is achieved by superimposing a negative offset on the predicted value to make the corrected predicted value closer to the actual viscosity. For example, if the original model prediction is 60 mPa·s, but the actual viscosity due to aggregation is 70 mPa·s, the predicted value needs to be corrected to 68 mPa·s through negative offset compensation, rather than reducing the actual viscosity. This avoids flow inaccuracies caused by adjusting the flow rate based on the lowered original prediction. The core of the offset compensation calculation is an adaptive adjustment based on the mapping relationship between the energy entropy value of the characteristic frequency band and the solution concentration. This mapping relationship was constructed through extensive laboratory experiments and stored as a three-dimensional mapping table, with the energy entropy value H as the X-axis, the solution concentration C as the Y-axis, and the compensation amount Δη as the Z-axis. The specific adjustment process is as follows: The energy entropy value H (0.7) of the current characteristic frequency band was obtained from the previous vibration spectrum analysis, and the real-time solution concentration C (8%) was obtained from the online refractometer. The second step was to locate the mapping interval. In the three-dimensional mapping table, the intersection interval of H=0.7±0.05 and C=8%±0.2% was found. Multiple sets of experimental data were pre-stored in this interval to ensure that accurate compensation amount can be output under any combination of H and C. Moreover, the absolute value of the compensation amount increases as H decreases (aggregation degree increases) and C increases (intermolecular interaction increases). This law is consistent with reality. The more severe the aggregation and the higher the concentration, the stronger the effect of peptide molecule aggregates on viscosity. A larger negative offset is needed to correct the prediction. After obtaining the offset compensation, a corrected real-time viscosity prediction value is generated. First, the original viscosity prediction value ηoriginal output by the rheological submodel is retrieved. Then, the correction value is calculated using the formula ηcorrected = ηoriginal + Δη. Substituting the data, we get ηcorrected = 75.5 + (-7.5) = 68 mPa·s. To ensure the reasonableness of the correction value, upper and lower limits need to be set for verification. The correction value must not be lower than the lower limit of viscosity in the critical thinning zone at the current concentration, nor higher than the upper limit of viscosity in the stable flow zone. If the correction value exceeds the range, it should be adjusted according to the closest boundary value to avoid over-correction that causes the viscosity prediction to deviate from reality. Within the flow range, the corrected real-time viscosity prediction needs to be simultaneously fed back to the thermodynamic sub-model to calculate the peptide chain hydrolysis rate in the high-temperature region. Since viscosity changes affect the flow velocity of the solution at the sealing surface, thus altering the residence time of the peptide chains in the high-temperature region, the thermodynamic sub-model, when calculating the hydrolysis rate, needs to substitute the corrected ηcorrected into the residence time calculation formula t = μ × L / (ρ × v × d²), where μ is the corrected viscosity, L is the length of the high-temperature region at the sealing surface, ρ is the solution density, v is the average flow velocity, and d is the pipe diameter. The peptide chain residence time t is then recalculated, and the hydrolysis rate constant k is updated using the Arrhenius equation to finally obtain the corrected ηcorrected. A more accurate hydrolysis risk thermogram is obtained. For example, before correction, ηoriginal=75.5mPa·s, t=0.5s, k=1.4×10⁻²s⁻¹, and after correction, ηcorrected=68mPa·s, t=0.45s, k=1.3×10⁻²s⁻¹. The hydrolysis rate decreases slightly as the residence time shortens. The color level of the corresponding high-temperature region in the thermogram is slightly adjusted from dark green to light green, making the hydrolysis risk assessment more consistent with the actual flow state of the solution and avoiding misjudgment of risk due to viscosity prediction deviation. At the same time, it realizes the bidirectional coupling of rheology and thermodynamic model, so that the control logic of the whole system forms a closed loop.

[0022] Based on real-time viscosity prediction values, flow control weights are generated. When the real-time viscosity prediction value is lower than a set threshold, the pump speed is increased to control priority. Specifically, a viscosity-driven weighted grading strategy is implemented. When the corrected real-time viscosity prediction value is lower than the lower limit of the shear thinning critical region, the flow control weight is automatically increased to the highest level, forcing the water pump speed to increase to the anti-shear instability range. If the viscosity prediction value is accompanied by an abnormal increase in the aggregation tendency coefficient, the aggregation tendency-viscosity coupling compensation mechanism is activated simultaneously for secondary correction to ensure that the flow control weight is decoupled from the molecular aggregation state.

[0023] After obtaining the corrected real-time viscosity prediction value through the aggregation tendency-viscosity coupling compensation mechanism, the intelligent control unit 1 needs to use this as the core to generate flow control weights. This is because the corrected viscosity directly reflects the current flow stability of the solution. If the viscosity is too low, it is easy to cause uncontrolled turbulence in the pipeline, leading to increased wear on the mechanical seal surface. If it is too high, it will increase the pump load and cause energy waste. Therefore, a viscosity-driven weighted classification strategy is needed to achieve precise flow control. The specific implementation method is as follows: First, it is necessary to clarify the definition and range of the shear-thinning critical region. This region is the flow range in which collagen peptide solutions, under dynamic shear, begin to undergo significant depolymerization of peptide molecules and the viscosity decreases significantly with increasing shear rate. Its delineation is based on the results calibrated using a multi-stage shear-thinning response surface library in the rheological sub-model mentioned earlier, and needs to be dynamically adjusted in conjunction with the solution concentration to ensure that the range matches the peptide molecule depolymerization kinetics. The lower limit of the shear-thinning critical region is the lowest viscosity value within this range. When the corrected real-time viscosity prediction value is lower than this lower limit, it indicates that the solution is subjected to excessive shear, and the peptide molecules are excessively depolymerized, approaching the turbulent flow control region (shear rate > 300 s⁻¹). It is necessary to urgently increase the flow control weight to stabilize the flow state. When the correction is detected... When the real-time viscosity prediction value is below the lower limit of the shear thinning critical region (24 mPa·s) for two consecutive sampling periods, the flow control weight will automatically increase from the normal level (level 3, out of 5 levels, with level 1 being the lowest and level 5 the highest) to the highest level (level 5). At this level, the flow control has the highest control priority in the system, temporarily disabling some resources for temperature control and anti-clogging control to prioritize flow stability. At the same time, the system will forcibly increase the pump speed to the anti-shear instability range. The anti-shear instability range refers to the range of pump speeds that can cause the dynamic shear rate of the fluid in the pipeline to fall back to the shear thinning critical region. The range value is determined by the speed-shear rate correlation model, which is based on a pipeline diameter of 50 mm and a fluid density of 1.05 g / cm³. ³ Calibration is performed using the formula n = k × γ, where k is the rotational speed coefficient. At an 8% concentration, k = 30 r / min⁻¹ / s⁻¹. For example, if the shear rate needs to be reduced from the current 320 s⁻¹ (corresponding to a viscosity of 22 mPa·s) to the lower limit of the critical zone (50 s⁻¹), the pump speed needs to be adjusted to 50 × 30 = 1500 r / min. The upper limit of the anti-shear instability range is set to ensure that the shear rate does not exceed 1.6 times the lower limit of the critical zone (i.e., 80 s⁻¹), corresponding to a rotational speed of 1800 r / min. Therefore, at an 8% concentration, the anti-shear instability range is 1500-1800 r / min. The rotational speed is adjusted in increments of 50 r / min to avoid sudden increases in speed that could cause pressure pulsations, until the shear rate stabilizes between 50-80 r / min. At this point, the solution viscosity will rise to 24-35 mPa·s, escaping the risk of turbulent flow control. If, while the predicted viscosity is below the lower limit of the shear thinning critical region, the peptide aggregation tendency coefficient shows an abnormal increase (abnormal increase means the coefficient is ≥0.5 for two consecutive sampling periods, far exceeding the dynamic calibration threshold of 0.3), it indicates that the aggregate particle size has exceeded 300 nm. Although the current viscosity is low, aggregation will lead to local viscosity inhomogeneity in the solution. If the rotation speed is only adjusted according to low viscosity, it will easily aggravate the collision and breakage of aggregates. Therefore, it is necessary to simultaneously activate the aggregation tendency-viscosity coupling compensation mechanism for secondary correction. The core of the secondary correction is to further superimpose the aggregation-shear coupling compensation amount on the basis of the original compensation to avoid the flow control weight being interfered with by the molecular aggregation state. The specific process is as follows: First, confirm the abnormally elevated aggregation tendency coefficient, and re-extract the characteristic frequency band energy entropy value H and the real-time solution concentration C. Then, call the extended mapping table of the aggregation tendency-viscosity coupling compensation mechanism. This table adds the aggregation tendency coefficient dimension to the original three-dimensional mapping table, forming a four-dimensional compensation matrix. Locate the intersection interval of H=0.55±0.05, C=8%±0.2%, and aggregation tendency coefficient=0.6±0.05. Calculate the aggregation-shear coupling compensation amount Δη'=+4mPa·s through trilinear interpolation. The positive sign indicates that a positive correction is added on the basis of the original negative compensation to offset the local viscosity increase caused by aggregation. Update the corrected real-time viscosity prediction value. The formula is ηcorrected'=ηcorrected+Δη'. If the original correction value is 22mPa·s, substituting it, we get ηcorrected'=22+4=26mPa·s. This value is still lower than the lower limit of the critical zone (24mPa·s), so the flow control weight remains the maximum. For high-level applications, the target rotation speed in the anti-shear instability range needs to be adjusted based on the new viscosity value. Using the shear rate formula, 26 mPa·s corresponds to a shear rate of approximately 45 s⁻¹. Therefore, the target rotation speed needs to be adjusted from 1500 r / min to 1600 r / min (45 × 30 × 1.18, where 1.18 is the aggregation correction coefficient). This ensures that increasing the rotation speed pulls the overall shear rate back to the critical zone without exacerbating aggregate breakage due to excessive speed. To verify the effectiveness of the secondary correction, the actual viscosity of the solution is collected using an online viscosity sensor. If the deviation between the actual viscosity and ηcorrected' is ≤5%, the secondary correction takes effect, and flow control is executed according to the new parameters. If the deviation is >5%, the above steps are repeated to recalculate the compensation amount until the deviation meets the requirements. Ultimately, this achieves decoupling of flow control weights from molecular aggregation states. The weights are determined solely by the overall flow stability of the solution and the corrected viscosity, while the local effects of aggregation are mitigated through secondary correction, preventing inaccurate flow control due to misjudgment of aggregation.

[0024] Simultaneously, temperature control weights are generated based on the hydrolysis risk distribution map output by the thermodynamic sub-model. If the area ratio of the high-temperature region exceeds a preset percentage, a preset proportion of control resources is allocated to cooling unit 2. Specifically, the risk area ratio weight allocation method is implemented: When the area ratio of high-temperature regions in the hydrolysis risk heat map exceeds the preset risk tolerance, the temperature control weight is increased proportionally according to the area ratio, and the corresponding proportion of control resources are dynamically allocated to the solenoid valves of each zone of cooling unit 2. At the same time, the highest risk block is located based on the peptide chain hydrolysis spatial probability mapping algorithm, and the coolant injection flow rate in that area is increased in a targeted manner.

[0025] After generating a hydrolysis risk thermogram and completing dynamic pH correction using the thermodynamic sub-model, the area ratio of the high-temperature region directly determines the control priority of cooling unit 2. If the high-temperature region is too large, conventional cooling intensity alone cannot quickly suppress peptide chain hydrolysis, and may even cause wear on the mechanical sealing surface due to local overheating. Therefore, it is necessary to use a risk region area ratio weighting method to accurately allocate control resources to the solenoid valves of each zone of cooling unit 2 to ensure that cooling efficiency matches the risk level. The specific implementation method is as follows: First, it's necessary to clarify the definitions of the high-temperature region and the preset risk tolerance. The high-temperature region is a set of grid cells in the hydrolysis risk thermogram where the temperature consistently exceeds the critical dissociation temperature of the collagen peptide β-sheet structure and the hydrolysis probability P' > 10%. Each grid cell has an area of ​​25 mm² (5 mm × 5 mm). The area of ​​the high-temperature region is the sum of the areas of these grid cells. The preset risk tolerance is the maximum allowable area percentage of the high-temperature region in the system. Based on the mechanical seal's heat dissipation capacity and the peptide chain hydrolysis inhibition requirements, it is calibrated to 20%. That is, when the proportion of the high-temperature region to the total sealing surface area exceeds 20%, it is considered a risk exceeding the limit, and dynamic allocation of control resources needs to be initiated. When the proportion of the high-temperature region exceeds the preset risk tolerance, the excess area percentage is first calculated: Exceedance percentage = (Actual percentage - Preset risk tolerance) / Preset risk tolerance × 100% = (22.9% ~ 20%) / 20% × 100% = 14.5%. Then, the temperature control weight is increased according to the formula: Temperature Control Weight = Base Weight × (1 + Exceedance Ratio). The base weight is the temperature control priority under normal operating conditions. Substituting the data, we get the temperature control weight = 3 × (1 + 14.5%) ≈ 3.43. Rounding up to level 4 ensures that the weight increase is positively correlated with the degree of risk exceeding the limit. The greater the excess, the higher the weight, and the more abundant the available control resources. The core step is to dynamically allocate the corresponding proportion of control resources to the solenoid valves of each zone in cooling unit 2. The solenoid valves of cooling unit 2 are divided into 5 areas according to the spatial coordinates of the sealing surface: 1 central area and 4 annular areas. Each area corresponds to an independent solenoid valve, controlling the coolant injection within that area. The initial control resource ratio for each solenoid valve, i.e., the coolant flow allocation ratio, is 20%, and the total resource ratio is 100%. The specific allocation process is as follows: This coefficient is positively correlated with the number of high-temperature grid cells in the corresponding region. The formula is: Distribution coefficient for a region = Number of high-temperature grid cells in that region / Total number of high-temperature grid cells. For example, if the total number of high-temperature grid cells is 18 (corresponding to 450 mm²), with 6 high-temperature grid cells in ring area 2, 5 in ring area 3, 3 in the central area, 2 in ring area 1, and 2 in ring area 4, then the distribution coefficient for ring area 2 is approximately 6 / 18 = 0.333, for ring area 3 approximately 0.278, for the central area approximately 0.167, for ring area 1 approximately 0.111, and for ring area 4 approximately 0.111. The resource proportion of each region is then determined by: Resource proportion of a region = Initial proportion × (1... Calculated using the allocation coefficient × excess ratio, the resource percentage for Ring Zone 2 is 20% × (1 + 0.333 × 14.5%) ≈ 20% × 1.048 ≈ 20.96%, Ring Zone 3 ≈ 20% × (1 + 0.278 × 14.5%) ≈ 20% × 1.040 ≈ 20.80%, Central Zone ≈ 20% × (1 + 0.167 × 14.5%) ≈ 20% × 1.024 ≈ 20.48%, and Ring Zones 1 and 4 ≈ 20% × (1 + 0.111 × 14.5%) ≈ 20% × 1.016 ≈ 20.32%. The sum of the resource percentages for each zone needs to be adjusted to 100%. Flow allocation and cooling unit are then implemented. 2. The total coolant flow rate is 10 L / min. The actual flow rates are calculated based on the resource proportions of each zone: Ring Zone 2 = 10 × 21.0% = 2.1 L / min, Ring Zone 3 = 10 × 20.8% = 2.08 L / min, Central Zone = 10 × 20.5% = 2.05 L / min, Ring Zone 1 = 10 × 20.3% = 2.03 L / min, Ring Zone 4 = 10 × 20.4% = 2.04 L / min. The system uses PWM signals to control the opening of the solenoid valves in each zone, achieving dynamic allocation of control resources. This ensures that areas with dense high-temperature grid cells receive more coolant, improving local heat dissipation efficiency. While controlling resources, the system locates the highest-risk blocks based on the peptide chain hydrolysis spatial probability mapping algorithm mentioned earlier. This algorithm recalculates the comprehensive hydrolysis risk value of each high-temperature grid unit. The comprehensive value = the magnitude of temperature deviation from the critical dissociation temperature × hydrolysis probability × pH deviation coefficient. The system then selects the grid unit cluster with the highest comprehensive value, forming the highest-risk block centered on this unit. The system records the center coordinates of this block and sends a directional enhancement command to the solenoid valve in the corresponding area. Based on the allocated flow rate, the injection flow rate is increased by an additional 15%. For example, the original allocated flow rate for annular zone 2 is 2.1 L / min, and after directional enhancement, it becomes 2.1 × (1 + 15%) = 2.At a flow rate of 415 L / min, the spray angle of the solenoid valve in this area is controlled to focus the coolant spray on the most critical area. The spray angle is adjusted from 30° to 15° via the solenoid valve's built-in guide vanes, achieving precise drip-like cooling and rapidly reducing the temperature in this area. The goal is to lower the temperature below the critical dissociation temperature within 30 seconds, preventing continuous hydrolysis of peptide chains in the high-risk area and avoiding energy waste caused by excessive overall flow. This results in both efficient and precise cooling control.

[0026] The peptide aggregation tendency coefficient is converted into an anti-clogging compensation command. When the peptide aggregation tendency coefficient rises continuously for more than the preset monitoring time, the ultrasonic descaling unit 3 is activated and the flow rate set value is reduced simultaneously. This is specifically achieved through an aggregation tendency time-domain cumulative triggering mechanism. When the aggregation tendency coefficient of peptide molecules continuously increases beyond the preset monitoring time, an ultrasonic frequency sweep signal with an intensity positively correlated with the aggregation tendency coefficient is generated. At the same time, based on the viscosity offset output by the aggregation tendency-viscosity coupling compensation mechanism, the flow rate setting of the circulation pipeline is reduced proportionally to weaken the molecular collision kinetic energy. Furthermore, the frequency band of the ultrasonic frequency sweep signal matches the entropy value analysis results of the characteristic frequency band.

[0027] After real-time viscosity correction is achieved through the aggregation tendency-viscosity coupling compensation mechanism and the aggregation state of peptide molecules is clearly defined, if the aggregation tendency coefficient shows a continuous upward trend, it means that the peptide aggregates in the solution are constantly increasing and growing larger. If not intervened in time, the aggregates are prone to adhere to the mechanical seal surface or the inner wall of the pipeline to form scale, which will not only aggravate seal wear, but may also reduce the pipeline diameter and lead to flow attenuation. Therefore, it is necessary to activate the ultrasonic descaling unit 3 and adjust the flow rate synchronously through the aggregation tendency time-domain accumulation triggering mechanism to actively resolve the aggregation risk. The specific implementation method is as follows: The logic for setting the preset monitoring duration is clearly defined. This duration is the minimum time window for determining whether an increase in the aggregation tendency coefficient indicates a continuous trend. A balance must be struck between preventing false triggers and ensuring timely response. If the duration is too short (<500ms), fluctuations in the coefficient caused by instantaneous sensor noise may be misinterpreted as a continuous increase. If it is too long (>2000ms), it will delay descaling, leading to excessive accumulation of aggregates. Based on collagen peptide aggregate growth kinetics experiments, it takes approximately 1500ms for aggregates to grow from a particle size of 100nm to 500nm. Therefore, the preset monitoring duration is set to 1000ms. The system will start a duration counter, beginning the count from the first detection of an increase in the aggregation tendency coefficient. If the coefficient shows an increasing trend (gradually rising from 0.3 to 0.55) for 10 consecutive periods, and the increase in adjacent periods is ≥0.02 (avoiding small increases being interpreted as a continuous increase), then the system is deemed to have met the trigger condition of a continuous increase exceeding the preset monitoring duration, and the system will immediately... This involves activating the aggregation tendency time-domain accumulation triggering mechanism. After activation, the first step is to generate an ultrasonic sweep signal. The core characteristics of this signal must precisely match the aggregation state. The intensity is positively correlated with the aggregation tendency coefficient, meaning the ultrasonic power increases linearly with the coefficient. The power baseline is set at 100W; for every 0.1 increase in the coefficient, the power increases by 30W. Simultaneously, a power upper limit of 300W is set to avoid excessive power damaging the silicon carbide coating on the mechanical seal surface without damaging the sealing components. Matching the ultrasonic sweep signal frequency band with the entropy analysis results of the characteristic frequency band ensures that the ultrasonic energy accurately acts on the peptide molecule aggregates. The entropy analysis of the characteristic frequency band determines that the frequency band strongly correlated with the Brownian motion of peptide molecules is 800–1200Hz. Therefore, the frequency band range of the sweep signal is set to 800–1200Hz, and the sweep step size is consistent with the width of the characteristic frequency band sub-band. The sweep period is 1s. The logic behind this matching design is: Aggregates exhibit the highest ultrasonic energy absorption efficiency in their Brownian motion-related frequency bands. A sweep frequency signal of 800-1200Hz can maximize energy transfer to the aggregates, breaking hydrogen bonds between them through vibration, thus achieving efficient descaling and avoiding energy waste in irrelevant frequency bands. Simultaneously with generating the ultrasonic sweep frequency signal, the flow rate setpoint in the circulation pipeline must be proportionally reduced based on the viscosity offset output by the aggregation tendency-viscosity coupling compensation mechanism. The viscosity offset is Δη output by this compensation mechanism; for example, Δη = -7.5 mPa·s. The negative sign indicates that the actual viscosity is higher than the model prediction due to aggregation. The absolute value of the offset, 7.5 mPa·s, reflects the degree of impact of aggregation on viscosity. The reduction ratio of the flow rate setpoint is positively correlated with the absolute value of the offset. The correlation logic is as follows: A larger viscosity offset indicates a greater number of aggregates and more frequent intermolecular collisions, requiring a more significant reduction in flow rate to weaken the kinetic energy of molecular collisions. Kinetic energy is positively correlated with the square of the flow velocity; reducing the flow rate reduces the flow velocity, thereby reducing the chance of aggregate collisions and aggregation. The specific calculation uses a linear mapping formula: Flow rate reduction ratio = (Absolute value of viscosity offset / Maximum allowable offset) × 100%, where the maximum allowable offset is 50 mPa·s (corresponding to the limiting state of aggregate particle size 500 nm). If the current absolute value of the viscosity offset is 7.5 mPa·s, then the flow rate reduction ratio = (7.5 / 50) × 100% = 15%. The system first retrieves the current flow rate setpoint baseline (e.g., 10 L / min) for the circulation pipeline, and then calculates the reduced flow rate setpoint = baseline value × (1 - reduction ratio) = 10L / min × (1 - 15%) = 8.5L / min. Simultaneously, the lower limit of the flow rate is set to 5L / min (to avoid dry friction on the mechanical seal surface due to insufficient lubrication caused by excessively low flow). If the calculated flow rate is lower than 5L / min, then 5L / min is used as the final set value. After adjusting the flow rate set value, the system controls the water pump speed through a PID controller to reduce the flow rate: using 8.5L / min as the target flow rate, the actual flow rate in the pipeline is collected in real time, and after calculating the deviation, the water pump speed is adjusted according to the proportional coefficient Kp = 2.0 and the integral coefficient Ki = 0.5. The speed reduction step is set to 50 r / min to avoid pressure pulsation caused by a sudden drop in speed. During this process, the ultrasonic descaling unit 3 and the flow rate adjustment must be synchronized. The ultrasonic signal is generated and the flow rate reduction command is issued simultaneously. The ultrasonic power is continuously output according to the set value until the aggregation tendency coefficient drops below 0.3. The flow rate is then stabilized at the reduced set value. The two work together to break up the existing aggregates with ultrasonic waves and reduce the chance of new aggregates forming by reducing the flow rate. This creates a dual anti-clogging effect of breaking up aggregates and preventing blockage. At the same time, the system collects the aggregation tendency coefficient and viscosity shift every 200ms. If the coefficient decreases for three consecutive cycles, the flow rate is gradually restored to the set value by 5% and the ultrasonic power is reduced to avoid excessive descaling and excessively low flow rate affecting the system's circulation efficiency. Ultimately, this achieves effective removal of aggregates and a stable balance in the solution flow state.

[0028] Finally, a three-dimensional dynamic weight matrix is ​​formed based on flow control weights, temperature control weights, and anti-clogging compensation commands. This matrix generates a coordinated control quantity every fixed time interval, synchronously adjusting the pump speed, cooling valve opening, and ultrasonic power to minimize system energy loss. Specifically, the formation of the three-dimensional dynamic weight matrix based on flow control weights, temperature control weights, and anti-clogging compensation commands, and the generation of the coordinated control quantity, employs a multi-objective constraint weight fusion technique. A three-dimensional matrix space is constructed with flow control weight as the X-axis, temperature control weight as the Y-axis, and anti-clogging compensation command as the Z-axis. The optimal control point is solved by Pareto front solution search algorithm every fixed period. The output is a set of coordinated commands for pump speed increase, cooling valve opening increase, and ultrasonic power amplitude. The command set is forced to meet the dual constraints of viscosity transition threshold in rheological sub-model and hydrolysis risk tolerance in thermodynamic sub-model.

[0029] After completing the calculation of flow control weights, the allocation of temperature control weights, and the generation of anti-clogging compensation commands, the three need to form a coordinated control logic. If only one weight is used for control, the system may suffer from neglecting one aspect while addressing another. Therefore, it is necessary to construct a three-dimensional dynamic weight matrix and generate coordinated control quantities in a rolling manner through multi-objective constraint weight fusion technology to ensure that the adjustments of water pump speed, cooling valve opening, and ultrasonic power are matched. The specific implementation method is as follows: First, the construction rules of the three-dimensional dynamic weight matrix need to be clarified. This matrix uses flow control weight as the X-axis, temperature control weight as the Y-axis, and anti-clogging compensation command as the Z-axis to build a three-dimensional space. The weight / command value range of each axis needs to be uniformly quantified to ensure the effectiveness of fusion: The X-axis (flow control weight) is converted into a normalized value of 0-1 according to the 5-level classification standard mentioned above (Level 1 = 0.2, Level 2 = 0.4, Level 3 = 0.6, Level 4 = 0.8, Level 5 = 1.0). The Z-axis (anti-clogging compensation command) is quantified based on the comprehensive quantification of ultrasonic power and flow reduction ratio, with the ultrasonic power being the proportion of maximum power. Multiplying by 0.6 and then multiplying the flow rate reduction ratio by 0.4, we obtain a normalized value of 0-1. This quantification method reflects both the intensity of ultrasonic descaling and the anti-clogging effect of flow rate adjustment, ensuring that the Z-axis accurately reflects the anti-clogging control strength. After matrix construction, the system updates and solves the problem on a fixed cycle of 500ms. This duration covers the parameter update cycle of each sub-model (flow rate, temperature, and anti-clogging parameters are updated every 100ms) and ensures timely control response, avoiding lag. At the beginning of each cycle, the system checks the flow control module and temperature control module... The anti-clogging control module reads the current normalized X, Y, and Z values ​​respectively, using them as the current state point in three-dimensional space. It then uses a Pareto front solution search algorithm to find the optimal control point. The Pareto front solution refers to the set of optimal solutions in multi-objective optimization where it is impossible to improve one objective without harming another. This algorithm iterates through a preset library of pump speed increase, cooling valve opening increase, and ultrasonic power amplitude command combinations, calculating the objective function value for each command combination. The objective function is calculated as: 0.4 × flow adaptability + 0.4 × temperature control efficiency + 0.2 × anti-clogging efficiency. The weight allocation is based on... Given the system requirements of prioritizing stable flow, followed by temperature control, and auxiliary anti-clogging, the combination of instructions with the highest objective function value and no other combination that can optimize any objective without reducing this value is selected as the cooperative instruction set corresponding to the optimal control point. For example, the optimal instruction selected in a certain period is a pump speed increase of +30 r / min, a cooling valve opening increase of +15%, and an ultrasonic power amplitude of 150W. This instruction can alleviate the problem of low viscosity by increasing the speed, control the risk of high temperature by increasing the cooling opening, and maintain the anti-clogging effect with 150W power, thus achieving a balance of multiple objectives.The key is that the output cooperative instruction set must strictly satisfy the dual constraints of the viscosity transition threshold in the rheological sub-model and the hydrolysis risk tolerance in the thermodynamic sub-model. This is the core guarantee to prevent instructions from exceeding the safety boundary. Regarding the rheological constraints, the system will substitute the pump speed increase into the rheological sub-model to calculate the viscosity value corresponding to the adjusted dynamic shear rate. If this value is lower than the lower limit of the shear thinning critical region (e.g., 24 mPa·s at 8% concentration) or higher than the upper limit of the stable flow region (80 mPa·s), the speed increase will be corrected according to the viscosity transition threshold. For thermodynamic constraints, the cooling valve opening increment is substituted into the thermodynamic sub-model to simulate the adjusted sealing surface temperature distribution. If the area ratio of the high-temperature region still exceeds the preset risk tolerance (20%), the cooling valve opening increment is increased. For example, after constraint verification, the above command is corrected to +25 r / min because the speed increase of +30 r / min causes the viscosity to approach the critical lower limit. The cooling valve opening increment of +15% can reduce the high-temperature ratio to 19%, satisfying the constraint. The ultrasonic power amplitude of 150W has been verified for anti-clogging efficiency and does not require correction. The final output is the corrected value. The collaborative instruction set consists of a 25 r / min increase in water pump speed, a 15% increase in cooling valve opening, and a 150W ultrasonic power amplitude. After generation, the instruction set is synchronously sent to the water pump drive module, the cooling unit 2 solenoid valve control module, and the ultrasonic descaling unit 3 drive module via the system bus. The water pump drive module adjusts the speed by a 25 r / min increase, the cooling unit 2 adjusts the corresponding zone solenoid valve by a 15% opening increase, and the ultrasonic descaling unit 3 outputs a sweep frequency signal with a 150W amplitude. The entire execution process is completed within 100ms, ensuring seamless connection with parameter updates in the next cycle. Simultaneously, the system records the collaborative instruction set and execution effect for each cycle and stores it in a historical database. This data is used to optimize the instruction combination library for the Pareto front solution search algorithm, gradually improving the accuracy of collaborative control and ultimately minimizing system energy loss. Experimental data shows that this technology reduces system energy consumption by 12%–15% compared to single-parameter control, while controlling peptide chain hydrolysis rate below 3% and aggregate scaling rate below 2%, fully meeting the stable operation requirements of the collagen peptide mechanical sealed water circulation system.

[0030] In this invention, the intelligent control unit 1 collects multiple parameters in real time, constructs rheological and thermodynamic sub-models, converts the vibration spectrum into peptide molecule aggregation tendency coefficients, corrects the viscosity prediction value to generate flow control weights, allocates resources to the cooling unit 2 based on the hydrolysis risk thermogram, and directionally cools high-risk areas. When the aggregation coefficient continuously rises beyond the limit, the ultrasonic descaling unit 3 is activated and the flow rate is reduced. The system generates collaborative control quantities through a three-dimensional dynamic weight matrix, and synchronously adjusts the water pump speed, cooling valve opening, and ultrasonic power to solve problems such as unstable flow, peptide chain hydrolysis, aggregation, and scaling, thereby ensuring stable production and product quality.

[0031] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A collagen peptide mechanical water circulation system, characterized in that, The intelligent control unit (1) achieves rheological-thermodynamic coupling optimization control through the following steps: real-time acquisition of fluid pressure pulsation data, mechanical seal axial vibration spectrum, sealing surface temperature gradient distribution, solution concentration and pH value; A rheological sub-model was constructed based on fluid pressure pulsation data and solution concentration to establish a viscosity response relationship library of collagen peptide solution under dynamic shear rate. A thermodynamic sub-model was constructed by embedding collagen peptide chain hydrolysis activation energy parameters based on the temperature gradient distribution of the sealing surface and pH value. The mechanical seal axial vibration spectrum is converted into a peptide molecule aggregation tendency coefficient, which is used as a dynamic correction factor input into the rheological sub-model to generate a corrected real-time viscosity prediction value. The flow control weight is generated based on the real-time viscosity prediction value. When the real-time viscosity prediction value is lower than the set threshold, the pump speed is increased to control the priority. At the same time, the temperature control weight is generated based on the hydrolysis risk distribution map output by the thermodynamic sub-model. If the area ratio of the high temperature region exceeds the preset percentage, the preset proportion of control resources is allocated to the cooling unit (2). The peptide molecule aggregation tendency coefficient is converted into an anti-clogging compensation command. When the peptide molecule aggregation tendency coefficient rises continuously for more than the preset monitoring time, the ultrasonic descaling unit (3) is activated and the flow set value is reduced simultaneously. Finally, a three-dimensional dynamic weight matrix is ​​formed based on flow control weight, temperature control weight, and anti-clogging compensation command. A collaborative control quantity is generated once every fixed time interval to synchronously adjust the water pump speed, cooling valve opening, and ultrasonic power, thereby minimizing system energy loss.

2. A collagen peptide mechanical water circulation system according to claim 1, characterized in that: A rheological sub-model is constructed based on fluid pressure pulsation data and solution concentration, specifically including: The dynamic shear rate variation trend of fluid is analyzed by pressure pulsation waveform feature extraction technology. Combined with solution concentration matching collagen peptide multi-level shear thinning response surface library, this surface library is calibrated by simulating peptide chain depolymerization kinetics experiments under different temperature gradients. The dynamic shear rate is divided into critical thinning region, stable flow region and turbulent flow control region, and viscosity transition thresholds for each region are established.

3. A collagen peptide mechanical water circulation system according to claim 2, characterized in that: A viscosity response database of collagen peptide solutions under dynamic shear rates was established. Furthermore, a dynamic calibration technique for shear thinning critical points was adopted. When the amplitude decay rate of the pressure pulsation waveform exceeds the preset fluctuation tolerance, the system automatically switches to the response surface library sub-library corresponding to the concentration and temperature combination, and outputs the viscosity prediction benchmark value that matches the current shear rate in real time. At the same time, the viscosity transition threshold is corrected according to the proportion of high-frequency energy in the axial vibration spectrum of the mechanical seal.

4. A collagen peptide mechanical water circulation system according to claim 1, characterized in that: A thermodynamic sub-model was constructed based on the temperature gradient distribution of the sealing surface and the activation energy parameters of collagen peptide chain hydrolysis embedded in the pH value. Specifically, this includes: The temperature gradient distribution of the sealing surface is input into the peptide chain hydrolysis spatial probability mapping algorithm. A hydrolysis risk heat map is generated based on the dissociation activation energy parameter of the collagen peptide β-sheet structure. The pH value is dynamically corrected by adjusting the hydrolysis reaction energy barrier height to modify the color scale distribution of the heat map. When the local temperature continuously exceeds the critical dissociation temperature and the pH deviates from the neutral range, a high-risk block is generated and marked as a priority cooling target.

5. A collagen peptide mechanical water circulation system according to claim 1, characterized in that: The axial vibration spectrum of the mechanical seal is converted into a peptide molecule aggregation tendency coefficient. Specifically, the characteristic frequency band entropy analysis method of vibration spectrum is used to extract the energy distribution of characteristic frequency bands that are strongly correlated with the Brownian motion of peptide molecules in the axial vibration spectrum of the mechanical seal. The energy entropy value in the characteristic frequency band is calculated and compared with the preset aggregation threshold to obtain the comparison result. The aggregation tendency coefficient increases in a negative correlation ratio according to the comparison result. At the same time, it is cross-validated by the real-time viscosity prediction value output by the rheological sub-model.

6. A collagen peptide machine-sealed water circulation system according to claim 5, characterized in that: As a dynamic correction factor input to the rheological sub-model, it generates a corrected real-time viscosity prediction value, specifically achieved through an aggregation tendency-viscosity coupling compensation mechanism: When the aggregation tendency coefficient exceeds the dynamic calibration threshold, negative offset compensation of the viscosity prediction value is triggered. The offset compensation amount is adaptively adjusted according to the mapping relationship between the energy entropy value of the characteristic frequency band and the solution concentration, and the offset viscosity prediction value is synchronously fed back to the thermodynamic sub-model to solve the peptide chain hydrolysis rate in the high temperature region.

7. A collagen peptide machine-sealed water circulation system according to claim 6, characterized in that: Flow control weights are generated based on real-time viscosity prediction values, and the priority of pump speed control is increased. Specifically, a viscosity-driven weighted grading strategy is implemented. When the corrected real-time viscosity prediction value is lower than the lower limit of the shear thinning critical region, the flow control weight is automatically increased to the highest level, forcing the water pump speed to increase to the anti-shear instability range. If the viscosity prediction value is accompanied by an abnormal increase in the aggregation tendency coefficient, the aggregation tendency-viscosity coupling compensation mechanism is activated simultaneously for secondary correction to ensure that the flow control weight is decoupled from the molecular aggregation state.

8. A collagen peptide mechanical water circulation system according to claim 4, characterized in that: Based on the hydrolysis risk distribution map output by the thermodynamic sub-model, temperature control weights are generated and control resources are allocated to the cooling unit (2). Specifically, the risk area proportion weight allocation method is implemented: When the area ratio of high-temperature regions in the hydrolysis risk heatmap exceeds the preset risk tolerance, the temperature control weight is increased proportionally according to the area ratio, and the corresponding proportion of control resources are dynamically allocated to the solenoid valves of each zone of the cooling unit. At the same time, the highest risk block is located based on the peptide chain hydrolysis spatial probability mapping algorithm, and the coolant injection flow rate in that area is increased in a targeted manner.

9. A collagen peptide mechanical water circulation system according to claim 5, characterized in that: The peptide molecule aggregation tendency coefficient is converted into an anti-clogging compensation command and activated by the ultrasonic descaling unit (3), specifically through the aggregation tendency time-domain accumulation triggering mechanism: When the aggregation tendency coefficient of peptide molecules continuously increases beyond the preset monitoring time, an ultrasonic frequency sweep signal with an intensity positively correlated with the aggregation tendency coefficient is generated. At the same time, based on the viscosity offset output by the aggregation tendency-viscosity coupling compensation mechanism, the flow rate setting of the circulation pipeline is reduced proportionally to weaken the molecular collision kinetic energy. Furthermore, the frequency band of the ultrasonic frequency sweep signal matches the entropy value analysis results of the characteristic frequency band.

10. A collagen peptide mechanical water circulation system according to claim 9, characterized in that: A three-dimensional dynamic weight matrix is ​​formed based on flow control weights, temperature control weights, and anti-blocking compensation commands, and collaborative control quantities are generated on a rolling basis. Specifically, multi-objective constraint weight fusion technology is adopted. A three-dimensional matrix space is constructed with flow control weight as the X-axis, temperature control weight as the Y-axis, and anti-clogging compensation command as the Z-axis. The optimal control point is solved by Pareto front solution search algorithm every fixed period. The output is a set of coordinated commands for pump speed increase, cooling valve opening increase, and ultrasonic power amplitude. The command set is forced to meet the dual constraints of viscosity transition threshold in rheological sub-model and hydrolysis risk tolerance in thermodynamic sub-model.

Images