Hydrogenated oil trans fatty acid content on-line monitoring and closed-loop control system

Abstract

This invention relates to an online monitoring and closed-loop control system for trans fatty acid content in hydrogenated oils, applied in the field of oil hydrogenation processing technology. The system includes a sampling pretreatment unit, a spectral detection unit, a control unit, and a hydrogenation execution unit. The sampling pretreatment unit processes high-temperature, high-pressure solid-containing materials, outputting uninterrupted test data. The spectral detection unit incorporates a quantitative prediction model for trans fatty acids and iodine value, outputting content data in real time. The control unit constructs a multi-parameter coupled collaborative control model with trans fatty acids as the first priority and iodine value as the second priority, outputting control commands. The hydrogenation execution unit adjusts hydrogen flow rate, stirring rate, and reaction temperature, forming a fully closed-loop feedback control chain. This invention achieves online real-time monitoring and precise closed-loop control of trans fatty acids, overcoming the industry bottleneck of traditional offline detection's lag and inability to suppress trans fatty acid formation, significantly improving product safety and process stability.

Description

Technical Field

[0001] This invention relates to the field of oil hydrogenation processing technology, specifically to an online monitoring and closed-loop control system for the trans fatty acid content of hydrogenated oils, which is particularly suitable for real-time monitoring and precise control of the trans fatty acid content in vegetable oils such as soybean oil, rapeseed oil, and palm oil during the hydrogenation reaction process. Background Technology

[0002] Hydrogenated oils are important raw materials in the food industry. Hydrogenation can improve the oxidative stability and plasticity of oils, thus enhancing their processing performance. However, the hydrogenation process inevitably produces trans fatty acids (TFAs). High intake of trans fatty acids increases the risk of cardiovascular disease. Therefore, various countries have established strict limits on the trans fatty acid content in oils.

[0003] In existing technologies, the control of hydrogenation reactions in hydrogenated oils mainly relies on offline sampling and detection. Operators periodically take samples from the reactor and send them to a laboratory for analysis using methods such as gas chromatography. Based on the test results, process parameters such as reaction temperature, hydrogen pressure, and stirring rate are manually adjusted. This offline detection method has the following drawbacks:

[0004] First, the detection process is severely delayed. From sampling and submission to obtaining test results, it usually takes tens of minutes or even hours. However, the rate of trans fatty acid formation during the hydrogenation reaction is relatively fast. By the time the test results are returned, the reaction may have already deviated from the target range, making timely intervention impossible and leading to unstable product quality or excessive trans fatty acid levels.

[0005] Secondly, there is a lack of closed-loop control mechanisms. In recent years, near-infrared spectroscopy has been attempted for the rapid detection of oil components, and existing near-infrared analyzers can be used for online monitoring of oil hydrogenation processes. However, these instruments only provide detection data output and do not form a linkage with the control system of the hydrogenation reactor. Operators still need to manually adjust process parameters based on the detection data, thus failing to achieve true closed-loop automatic control.

[0006] Furthermore, the complex operating conditions inside the hydrogenation reactor pose a technical obstacle to online detection. Hydrogenation reactions typically occur under high temperature (150-250℃) and high pressure (0.3-1.5MPa) conditions. The reactants contain solid catalyst particles (such as nickel-based catalysts with a particle size of approximately 5-50μm), and bubbles are generated during the reaction. Conventional near-infrared detection devices are difficult to install directly in such environments; the sampling pipeline is easily clogged by solid particles, and bubbles severely interfere with the spectral signal, leading to distorted detection data.

[0007] More fundamentally, the existing control logic for hydrogenation reactions has a fundamental flaw: the industry has consistently used iodine value as the sole core control target for hydrogenation reactions, treating trans fatty acids only as a compliance indicator for product release, without incorporating them into the real-time closed-loop control system for hydrogenation reactions. In hydrogenation reactions, there is a strong coupling contradiction between the acceptable range of iodine value and the amount of trans fatty acid formation—lowering reaction temperature and pressure can inhibit trans fatty acid formation, but it leads to a slower decrease in iodine value and a longer reaction time; increasing hydrogen flow rate and stirring speed can accelerate the attainment of iodine value, but it may promote trans fatty acid formation. This control logic, which targets only iodine value, cannot simultaneously guarantee the core functionalities of oils such as plasticity and melting point while inhibiting trans fatty acid formation at the source of the reaction. This is the core technical bottleneck that the hydrogenated oil processing industry has failed to resolve for decades, and it is the fundamental reason why the risk of excessive trans fatty acids in commercially available hydrogenated oils remains high.

[0008] Therefore, there is an urgent need in this field for a hydrogenation reaction closed-loop control system that can overcome the obstacles of detecting solid materials under high temperature and high pressure, realize online real-time detection of trans fatty acids, prioritize the safety control of trans fatty acids, and take into account the functionality of oils. Summary of the Invention

[0009] In view of the above-mentioned prior art, the present invention aims to solve the problems of delayed detection of trans fatty acids in the hydrogenation reaction process and the inability to achieve closed-loop precise control in the prior art, overcome the interference of high temperature and high pressure solid materials on online spectral detection, and more importantly, solve the fundamental technical bottleneck of the industry that the prior art uses iodine value as the only control target and cannot inhibit the generation of trans fatty acids from the source.

[0010] To achieve the above objectives, the present invention provides an online monitoring and closed-loop control system for the trans fatty acid content of hydrogenated oils, including a sampling pretreatment unit, a spectral detection unit, a control unit, and a hydrogenation execution unit, and is configured in conjunction with a hydrogenation reaction device for hydrogenated oils. The hydrogenation reaction device includes a hydrogenation reactor, a hydrogen supply pipeline, and a stirring assembly.

[0011] The core of this invention lies in constructing a multi-parameter coupled collaborative control model with trans fatty acid content as the primary constraint and iodine value as a co-target, and introducing a mechanism to dynamically adjust the weights of process parameters based on real-time deviations, thereby achieving multi-objective optimization closed-loop control of the hydrogenation reaction process.

[0012] The sampling end of the sampling pretreatment unit extends into the reaction material chamber of the hydrogenation reactor to perform online impurity removal, defoaming, and constant temperature and pressure stabilization treatment on the high-temperature and high-pressure solid-containing materials during the hydrogenation reaction process, and outputs uninterrupted test material; the sampling pretreatment unit includes a high-temperature and high-pressure resistant in-situ sampling probe that extends into the hydrogenation reactor, as well as an online impurity removal component, a defoaming component, and a constant temperature and pressure stabilization component that are sequentially sealed and connected to form an integrated online pretreatment pipeline assembly;

[0013] The spectral detection unit is an explosion-proof near-infrared spectral detection unit adapted to the high-risk combustible atmosphere of the hydrogenation reaction. It is hermetically connected to the output end of the sampling and pretreatment unit. It is internally equipped with a quantitative prediction model for trans fatty acids and iodine value in hydrogenated oils and fats, and is used for real-time spectral collection and analysis of interference-free待测 materials, and synchronously outputs real-time content data of trans fatty acids and real-time iodine value data;

[0014] The control unit is internally equipped with a trans fatty acid target closed-loop regulation module. The input end of the target closed-loop regulation module is communicatively connected to the output end of the spectral detection unit, and is used to receive real-time content data of trans fatty acids and real-time iodine value data. Taking the trans fatty acid target threshold pre-stored in the system locally or set through the human-machine interface as the first-priority control target and the iodine value target threshold as the second-priority control target, a multi-parameter coupling and collaborative regulation model is constructed, and a hydrogenation process regulation instruction is output;

[0015] The hydrogenation execution unit includes an electronically controlled proportional regulating valve arranged on the hydrogen supply pipeline, a variable-frequency speed-regulating motor传动连接 to the stirring component, and a temperature regulation component配套 with the hydrogenation reaction kettle. The controlled end of the hydrogenation execution unit is communicatively connected to the output end of the control unit, and is used to receive the hydrogenation process regulation instruction and adjust the process parameters of the hydrogenation reaction in real time, so as to form a closed-loop feedback control link with multi-parameter coordination of flow-pressure-temperature-stirring for the whole process of the hydrogenation reaction of hydrogenated oils and fats, and realize the dynamic stability of process parameters and the convergent control of product quality indicators.

[0016] As a further improvement of the present invention, the online impurity removal component is an online precision filtration component with a filtration accuracy of 5-10μm, adapted to the precision purification of high-temperature and high-viscosity materials containing solid catalyst particles; the defoaming component is an ultrasonic defoaming component; the multi-parameter coupling and collaborative regulation model presets a coupling influence weight matrix of hydrogenation process parameters on the formation of trans fatty acids. The hydrogenation process parameters include reaction temperature, hydrogen pressure, hydrogen flow rate and stirring rate; the model calculates and outputs a multi-parameter linkage adjustment combination that makes the deviation converge based on the deviation between the real-time content of trans fatty acids and the trans fatty acid target threshold, and the deviation between the real-time iodine value data and the iodine value target threshold, and outputs the corresponding hydrogenation process regulation instruction; the target closed-loop regulation module is also internally equipped with a trans fatty acid generation trend prediction sub-module based on hydrogenation reaction kinetics, and based on real-time detection data and current process parameters,推演预测 the change trend of trans fatty acid content within the next 10-30 minutes, and提前输出 a pre-regulation instruction.

[0017] As a further improvement of the present invention, the targeted closed-loop control module achieves the first priority and second priority control objectives by constructing a dual-objective collaborative optimization objective function with constraints. The constraint is that the real-time trans fatty acid content does not exceed the trans fatty acid target threshold, and the optimization objective is to stably control the iodine value within the iodine value target threshold range, thereby achieving synergistic optimization of the hydrogenation functionality of the oil and the safety management of trans fatty acids. The targeted closed-loop control module also has a built-in dynamic weight adjustment mechanism. The dynamic weight adjustment mechanism automatically adjusts the weight coefficients of each process parameter in the coupling influence weight matrix according to the deviation between the real-time trans fatty acid content and the trans fatty acid target threshold: when the deviation is less than the preset approach threshold, the weight coefficients of reaction temperature and / or hydrogen pressure are increased first; when the deviation is greater than the preset deviation threshold, the weight coefficients of hydrogen flow rate and / or stirring rate are increased first. The approach threshold and deviation threshold are deviation judgment critical values ​​determined by a preset self-tuning algorithm or a manual preset method.

[0018] As a further improvement of the present invention, the quantitative prediction model for trans fatty acids and iodine value of hydrogenated oils is a dynamic generalization prediction model based on transfer learning. It uses hydrogenated oil samples with different raw material oils, different catalytic systems, and different hydrogenation conditions as the basic dataset. The distribution shift caused by the difference in operating conditions is eliminated by the domain adaptive algorithm to overcome the influence of spectral drift and sample distribution differences at the hydrogenation site, and is adapted to the hydrogenation reaction scenarios of various raw material oils such as soybean oil, rapeseed oil, and palm oil.

[0019] As a further improvement of the present invention, the spectral detection unit has a built-in spectral preprocessing module and a spectral signal quality self-diagnosis module. The spectral preprocessing module is equipped with a multivariate scattering correction algorithm combined with a wavelet transform denoising algorithm to perform baseline correction, scattering interference elimination, and noise filtering on the acquired raw near-infrared spectrum, outputting standardized spectral data and inputting it into a quantitative prediction model. The spectral signal quality self-diagnosis module monitors the signal-to-noise ratio and baseline drift of the near-infrared spectrum in real time. When the signal-to-noise ratio is lower than a preset threshold or the baseline drift exceeds a preset range, it sends a signal to the control unit. The control unit then instructs the isothermal and pressure-regulating component in the sampling preprocessing unit to perform a parameter calibration process. After calibration, the spectral data is reacquired until the comprehensive spectral quality index meets the preset admission criteria.

[0020] As a further improvement of the present invention, the control unit also incorporates a catalyst management module, which includes a catalyst activity self-learning submodule, a catalyst type adaptive submodule, and a historical data traceability submodule; wherein:

[0021] The catalyst activity self-learning submodule records the change curve of trans fatty acid formation rate during each hydrogenation reaction, compares and analyzes it with historical reaction data, and dynamically updates the catalyst activity correction factor in the multi-parameter coupled synergistic regulation model to offset the formation deviation caused by catalyst batch differences and activity decay in real time.

[0022] The catalyst type adaptive submodule automatically switches the set of control parameters in the multi-parameter coupled synergistic regulation model according to the type of catalyst used in the current hydrogenation reaction, so as to adapt to the differences in hydrogenation characteristics of nickel-based catalysts, noble metal catalysts or catalysts with different supports.

[0023] The historical data traceability submodule records the reaction temperature, hydrogen pressure, hydrogen flow rate, stirring rate, real-time trans fatty acid content, and iodine value of each hydrogenation reaction, which is used for full-chain traceability of product quality.

[0024] As a further improvement of the present invention, the control unit also incorporates a multi-mode control switching logic. The multi-mode control switching logic automatically selects the control mode based on the deviation between the real-time trans fatty acid content and the target threshold of trans fatty acids: when the deviation is greater than a preset first threshold, a fast response control mode is adopted, prioritizing the adjustment of hydrogen flow rate and reaction temperature; when the deviation is less than a preset second threshold, a fine adjustment control mode is adopted, prioritizing the adjustment of stirring rate; when the deviation is between the first and second thresholds, a multi-parameter collaborative control mode is adopted, synchronously adjusting all process parameters; wherein, the first threshold is greater than the second threshold, and the first and second thresholds are preset judgment thresholds based on the graded allowable deviation range of trans fatty acids.

[0025] As a further improvement of the present invention, the system also includes an early warning and safety interlock unit, which is communicatively connected to the control unit and has preset graded safety triggering rules: when the real-time content of trans fatty acids exceeds the preset early warning threshold, an audible and visual alarm is automatically triggered; when the real-time content of trans fatty acids exceeds the preset shutdown safety threshold, or when the spectral signal quality fails to return to the qualified range after a preset time of continuous calibration and repair attempts, an emergency shutdown interlock of the hydrogenation reaction unit is automatically triggered.

[0026] As a further improvement of the present invention, the catalyst type adaptive submodule has a built-in catalyst type parameter library, which includes at least a nickel-based catalyst control parameter set, a noble metal catalyst control parameter set, and a supported catalyst control parameter set, and is used to automatically call the corresponding control parameter set according to the catalyst type used in the current hydrogenation reaction.

[0027] Compared with the prior art, the present invention has the following beneficial effects:

[0028] 1. By constructing a dual-objective synergistic control logic with trans fatty acid content as the first priority and iodine value as the second priority, an optimized objective function with constraints is built. Under the hard constraint of ensuring that trans fatty acid content does not exceed the limit, the iodine value is stably controlled within the target range, fundamentally solving the core industry bottleneck that existing technologies cannot simultaneously address the functionality of hydrogenated oils and food safety. The dynamic weight adjustment mechanism can automatically optimize the control strategy according to the deviation range, significantly improving control accuracy.

[0029] 2. By integrating the sampling pretreatment unit, the spectral detection unit, the control unit, and the hydrogenation execution unit, a fully closed-loop feedback control link with coordinated flow, pressure, temperature, and stirring parameters is formed. This solves the problems of lagging traditional offline detection and inability to adjust in a timely manner, and realizes real-time tracking and automatic correction of trans fatty acid content. The control response speed is significantly improved compared with manual adjustment.

[0030] 3. The sampling pretreatment unit uses an in-situ sampling probe in conjunction with online impurity removal, defoaming, and constant temperature and pressure stabilization components, which can effectively remove solid catalyst particles, eliminate bubbles, and stabilize the sample state, providing interference-free test material for near-infrared spectroscopy detection. This solves the problem of data distortion under high temperature and high pressure conditions containing solids, and significantly improves the accuracy and stability of the detection data.

[0031] 4. It has a built-in trend prediction submodule based on hydrogenation reaction kinetics, which can predict the trend of trans fatty acid changes in advance and output pre-regulation instructions. It can intervene in advance before the deviation actually occurs, thus completely avoiding the problem of trans fatty acid exceeding the standard caused by regulatory lag.

[0032] 5. The quantitative prediction model is built based on transfer learning and can be adapted to various raw oils such as soybean oil, rapeseed oil, and palm oil; the catalyst management module can learn the catalyst activity decay and automatically match the control parameters of different types of catalysts, which significantly improves the system's versatility and robustness under operating conditions.

[0033] 6. The multi-mode control switching logic can adapt to control requirements with different deviation ranges. The hierarchical safety interlocking mechanism can automatically alarm or shut down in case of abnormality. The historical data traceability module fully records key data throughout the entire process, realizing full-chain traceability of product quality and greatly improving the system's operational safety and compliance. Attached Figure Description

[0034] Figure 1 This is a system structure block diagram of the present invention. Detailed Implementation

[0035] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of protection of this invention.

[0036] Example 1: System Overall Structure and Sampling Preprocessing Unit

[0037] like Figure 1 As shown, this embodiment provides an online monitoring and closed-loop control system for the trans fatty acid content of hydrogenated oils, which is installed in conjunction with a hydrogenation reaction device for hydrogenated oils. The hydrogenation reaction device includes a hydrogenation reactor, a hydrogen supply pipeline, and a stirring assembly.

[0038] The specific structure of the sampling preprocessing unit:

[0039] The sampling pretreatment unit includes a high-temperature, high-pressure in-situ sampling probe that extends into the hydrogenation reactor. This probe is made of 316L stainless steel and has a filter screen at its front end to block large-diameter solid catalyst particles. The in-situ sampling probe is sequentially and sealed to an online impurity removal component, a defoaming component, and a constant temperature and pressure stabilizing component via high-pressure resistant pipelines, forming an integrated online pretreatment pipeline assembly.

[0040] The online impurity removal component is an online precision filtration component with a built-in sintered metal filter element with a filtration precision of 5-10 μm, which can effectively trap nickel-based catalyst particles (approximately 5-50 μm in diameter) in the hydrogenation reaction materials. To verify the technical effect of the filtration precision endpoint, this embodiment tested filter elements with filtration precisions of 5 μm and 10 μm respectively: when using a 5 μm filter element, the retention rate of catalyst particles with a diameter ≥5 μm is ≥99.5%, but the pressure drop increases by approximately 15% compared to an 8 μm filter element; when using a 10 μm filter element, the pressure drop is smaller, but the retention rate of 5-8 μm particles decreases slightly. In practical applications, an appropriate precision can be selected according to the catalyst particle size distribution, with 8 μm being the preferred value, ensuring both retention effect and operational economy. The filtration component adopts a backflushing cleaning structure, which can automatically backflush periodically to remove catalyst particles adhering to the filter element surface, ensuring long-term filtration effect.

[0041] The defoaming component is an ultrasonic defoaming component, comprising an ultrasonic generator and a defoaming chamber. When material containing air bubbles flows through the defoaming chamber, the ultrasonic generator emits high-frequency sound waves of 20-40kHz, causing the air bubbles to resonate, coalesce, and burst under the influence of the sound field, thereby eliminating air bubble interference in the material. The defoamed material then enters a constant temperature and pressure stabilizing component.

[0042] The constant temperature and pressure stabilizing component includes a heat exchanger and a pressure regulator. The heat exchanger controls the material temperature within the range of 25-35℃, while the pressure regulator stabilizes the material pressure at 0.1-0.3MPa through a back pressure valve, ensuring that the material entering the spectral detection unit is in a constant temperature and pressure state, eliminating the influence of temperature and pressure fluctuations on the spectral signal.

[0043] The interference-free test material, after being processed by the sampling preprocessing unit, is transported by pipeline to the spectral detection unit for real-time spectral acquisition.

[0044] Example 2: Spectral Detection Unit

[0045] The spectral detection unit is an explosion-proof near-infrared spectral detection unit adapted to the highly hazardous flammable atmosphere of hydrogenation reactions. Its housing adopts an explosion-proof design, meeting the ExdⅡBT4 explosion-proof rating requirements. The output terminals of the spectral detection unit and the sampling preprocessing unit are sealed and connected, and it incorporates a quantitative prediction model for trans fatty acids and iodine values ​​in hydrogenated oils.

[0046] Spectral acquisition and preprocessing:

[0047] The spectral detection unit employs diffuse reflectance or transmission sampling methods, acquiring near-infrared spectra in the range of 1000-2500 nm. The spectral preprocessing module utilizes a multivariate scattering correction (MSC) algorithm combined with a wavelet transform denoising algorithm to process the acquired raw near-infrared spectra.

[0048] The specific implementation of the multivariate scattering correction algorithm is as follows: Let the original spectral matrix be... ,in For the sample size, This represents the number of wavelength points. First, calculate the average spectrum:

[0049] Then, linear regression was performed on each spectrum:

[0050] in, and For regression coefficients, This is the residual. The corrected spectrum is:

[0051] Wavelet transform denoising was performed using the db4 wavelet basis function, with a decomposition level of 5. High-frequency noise signals were removed using a soft thresholding method. The preprocessed standardized spectral data was then input into the quantitative prediction model.

[0052] Construction of quantitative prediction models:

[0053] The quantitative prediction model is established using the partial least squares regression (PLSR) algorithm. Let the spectral matrix be... Trans fatty acid content matrix The relationship between them is:

[0054] in, This is the regression coefficient matrix. Let be the residual matrix. The number of principal components is determined through cross-validation to minimize the sum of squared predicted residuals.

[0055] To address the adaptation issue of different raw material oils (soybean oil, rapeseed oil, and palm oil), this embodiment employs a dynamic generalization prediction model based on transfer learning. First, a source domain model is established using soybean oil hydrogenation samples as the base dataset. Then, a domain adaptation algorithm is used to transfer the model to rapeseed oil and palm oil scenarios.

[0056] Specific implementation of the domain adaptive algorithm (MMD alignment): Assume the source domain spectral data distribution is as follows: The distribution of the target domain spectral data is as follows Maximum Mean Discrepancy (MMD) is used to measure the distance between two distributions. MMD is defined as:

[0057] in, For kernel function mapping (using Gaussian radial basis kernel function, bandwidth parameter) (Selected using the median heuristic) Let Hilbert space be the regenerating kernel. Distribution alignment is achieved by minimizing the MMD distance between the source and target domains while maintaining the discriminative performance of the prediction model. Specifically, gradient descent is used for iterative optimization, with a maximum number of iterations of 500, a learning rate of 0.01, and a convergence threshold of [value missing]. The aligned source domain model can be directly applied to target domain samples, significantly reducing the calibration cost of the model for new raw material oils.

[0058] Spectral signal quality self-diagnosis:

[0059] The spectral signal quality self-diagnostic module monitors the signal-to-noise ratio (SNR) and baseline drift of the near-infrared spectrum in real time. When the SNR is lower than a preset threshold (e.g., SNR < 30 dB) or the baseline drift exceeds a preset range (e.g., ± 0.05 Abs), a signal is sent to the control unit. The control unit then instructs the isothermal and pressure-regulating component in the sampling preprocessing unit to perform a parameter calibration process. After calibration, the spectral data is reacquired until the overall spectral quality index meets the preset admission criteria.

[0060] Example 3: Control Unit and Multi-Parameter Coupled Coordinated Control Model

[0061] The control unit has a built-in trans fatty acid targeted closed-loop regulation module, which receives real-time trans fatty acid content data and iodine value data output by the spectral detection unit.

[0062] Construction of a multi-parameter coupled synergistic regulation model:

[0063] The multi-parameter coupled synergistic regulation model pre-defines a weight matrix for the coupling effects of hydrogenation process parameters on trans fatty acid formation. Hydrogenation process parameters include reaction temperature (…). (unit: °C), hydrogen pressure ( (unit MPa), hydrogen flow rate ( (unit: L / min) and stirring rate ( (unit: r / min).

[0064] weight matrix The model was constructed using a combination of response surface methodology and historical data regression. The specific method is as follows: Experiments were conducted using a central composite design (CCD) to examine the effects of various process parameters on trans fatty acid formation, and a quadratic response surface model was established.

[0065] in, These are the normalized process parameters. These are the regression coefficients. The weighting coefficients for the coupling effects of each parameter are obtained by normalizing the regression coefficients:

[0066] The model is based on the deviation between the real-time trans fatty acid content and the target threshold for trans fatty acids. And the deviation between real-time iodine value data and the target iodine value threshold. The multi-parameter linkage adjustment combination that converges the deviation is calculated and output through the weight matrix:

[0067] in, For process parameter adjustment vectors, Here is the gain matrix. This is the deviation vector.

[0068] In this embodiment, the weight matrix This serves as the baseline matrix. In actual closed-loop control, its elements will be adjusted in real time according to a dynamic weight adjustment mechanism (i.e., ...). Become ), and receive catalyst activity correction factor Global scaling (i.e.) Become This enables adaptive control.

[0069] Bi-objective collaborative optimization objective function and SQP solution algorithm:

[0070] The targeted closed-loop control module achieves the first and second priority control objectives by constructing a constrained dual-objective collaborative optimization objective function. The objective function is defined as follows:

[0071] The constraints are:

[0072] in, This is a function for predicting trans fatty acid content. This is a prediction function for iodine value. and Priority weight coefficient ( ).

[0073] The objective function is solved using the Sequential Quadratic Programming (SQP) algorithm, specifically implemented as follows: At each iteration point... The search direction is determined by solving a quadratic programming subproblem. :

[0074] in, This is an approximation of the Hessian matrix for the Lagrange function (updated using the BFGS formula). Let be the constraint function. After solving for the search direction, determine the step size through linear search and update the iteration points. Set the convergence condition to the gradient norm being less than 1. The algorithm can be iterated more than 100 times. It can quickly converge to the optimal solution within the target iodine value range while satisfying the trans fatty acid constraint.

[0075] Dynamic weight adjustment mechanism:

[0076] The dynamic weight adjustment mechanism automatically adjusts the weight coefficients of each process parameter in the coupling influence weight matrix based on the deviation between the real-time trans fatty acid content and the target threshold for trans fatty acids. The specific rules are as follows:

[0077] When the deviation is less than the preset approach threshold (e.g., deviation < 5%), the weighting coefficients of reaction temperature and hydrogen pressure are increased first to stabilize the iodine value. When the deviation is greater than the preset deviation threshold (e.g., deviation > 15%), the weighting coefficients of hydrogen flow rate and stirring rate are increased first to quickly suppress the formation of trans fatty acids. When the deviation is between the two, multi-parameter coordinated adjustment is adopted, and linkage control is performed according to the preset weight matrix.

[0078] The approach threshold and deviation threshold are determined by a preset self-tuning algorithm. The specific implementation of the self-tuning algorithm is as follows: the system records the historical data sequence of trans fatty acid deviation in real time. ( (Take a range of 100-200), calculate the standard deviation of the deviation under steady-state conditions. Take the threshold as... Deviation from the threshold is To prevent individual outliers from causing drastic fluctuations in the threshold, a sliding window averaging method is used. Smoothing is performed, with a window size of 20. The system automatically updates the threshold after every 10 production batches or after accumulating 200 data points. The above is a specific implementation example of a preset self-tuning algorithm. Those skilled in the art will understand that other methods based on historical data statistics (such as moving percentiles, control chart principles) or model simulation can also be used to achieve automatic threshold tuning.

[0079] Example 4: Trend Prediction Submodule

[0080] The targeted closed-loop control module incorporates a sub-module for predicting the trans fatty acid formation trend based on hydrogenation reaction kinetics. This sub-module extrapolates and predicts the trans fatty acid content change trend within the next 10–30 minutes based on real-time detection data and current process parameters, and outputs pre-control commands in advance.

[0081] Dynamic model construction:

[0082] The formation of trans fatty acids can be simplified to a first-order reaction kinetic model: in:

[0083] Trans fatty acid concentration, in %

[0084] Pre-exponential factor, unit

[0085] For activation energy, unit

[0086] Let be the gas constant, and take .

[0087] The reaction temperature is expressed in Kelvin (K).

[0088] This is the partial pressure of hydrogen, in MPa.

[0089] The reaction order is the partial pressure of hydrogen.

[0090] This represents the concentration of unsaturated fatty acids, in % (%).

[0091] Conversion relationship with iodine value: The concentration of unsaturated fatty acids is positively correlated with the iodine value (IV). According to fatty acid composition analysis, the two approximately satisfy a linear relationship. ,in, The conversion factor is set at 0.035-0.045 for soybean oil and 0.040-0.050 for rapeseed oil. In practical applications, the system calculates the concentration of unsaturated fatty acids using real-time detected iodine value data, thus achieving closed-loop correction of the kinetic model.

[0092] Model parameter calibration:

[0093] The kinetic parameters were calibrated using experimental data. Hydrogenation reactions were carried out under different temperatures (150-250℃) and hydrogen pressures (0.3-1.5MPa), and the trans fatty acid content was measured at regular intervals. The results were obtained using nonlinear least squares fitting.

[0094] activation energy

[0095] Pre-exponential factors

[0096] Hydrogen partial pressure reaction order

[0097] Specific experimental conditions for parameter calibration: Using soybean oil as raw material, with a nickel-based catalyst dosage of 0.5 wt%, a baseline experiment was conducted at a temperature of 180℃ and a hydrogen pressure of 0.6 MPa. Reaction rate data under different operating conditions were obtained by changing a single parameter. A total of 20 experimental points were designed for parameter fitting, and the coefficient of determination was determined. Root mean square error of model prediction .

[0098] Forecasting and Pre-regulation:

[0099] The fourth-order Runge-Kutta method was used to numerically solve the kinetic equations. Using the current detection data as initial conditions, the prediction was extrapolated forward by 10-30 minutes to obtain the trend of trans fatty acid changes. To verify the applicability of different prediction durations, this embodiment tested prediction durations of 10 minutes, 20 minutes, and 30 minutes respectively: the prediction error was smallest when the prediction duration was 10 minutes. This is suitable for fast-response scenarios; however, the prediction error increases slightly when the prediction duration is 30 minutes. While offering more time for preparation and control, this method provides a more comprehensive control measure. In practical applications, the prediction duration can be selected based on the process response speed, with 20 minutes being the preferred value, balancing prediction accuracy and control lead time. When the predicted value exceeds the warning threshold (e.g., 80% of the target threshold) within the prediction duration, a pre-control instruction is output in advance to intervene in the process parameters. The core purpose of the trend prediction submodule is to qualitatively or semi-quantitatively determine the changing trend of trans fatty acids and provide timely warnings, offering a basis for pre-control decisions. Effective implementation examples show that this prediction capability can effectively prevent exceedances (pass rate increased to 98.5%), meeting the needs of industrial applications.

[0100] Example 5: Catalyst Management Module

[0101] The control unit also has a built-in catalyst management module, which includes a catalyst activity self-learning submodule, a catalyst type adaptive submodule, and a historical data traceability submodule.

[0102] Catalyst activity self-learning submodule:

[0103] The catalyst activity self-learning submodule records the rate of trans fatty acid formation during each hydrogenation reaction, compares and analyzes it with historical reaction data, and dynamically updates the catalyst activity correction factor in the multi-parameter coupled synergistic regulation model. .

[0104] The activity correction factor is updated using the Recursive Least Squares (RLS) algorithm. Let the predicted value of the kinetic model be... The actual measured value is The prediction error is then... Correction factor The update formula is:

[0105] in, The gain matrix is ​​updated in real time using a recursive formula:

[0106] in, For the regression vector, The forgetting factor (ranged from 0.95 to 0.99) controls the rate at which historical data is forgotten. This mechanism helps to offset the generation deviation caused by batch-to-batch differences and activity decay of the catalyst in real time.

[0107] Catalyst type adaptive submodule:

[0108] The catalyst type adaptive submodule automatically switches the set of control parameters in the multi-parameter coupled synergistic regulation model according to the type of catalyst used in the current hydrogenation reaction (nickel-based catalyst, noble metal catalyst, or different supported catalyst).

[0109] Specifically, the catalyst type parameter library includes:

[0110] Nickel-based catalyst control parameter set: applicable to Pricat 9908, 9910, 9920 and other nickel-based catalysts, reaction temperature range 150-200℃, hydrogen pressure range 0.3-0.8MPa, weight matrix biased towards temperature regulation (temperature weight coefficient 0.4, pressure 0.3, hydrogen flow rate 0.2, stirring 0.1).

[0111] Noble metal catalyst control parameter set: applicable to palladium, platinum and other noble metal catalysts, reaction temperature range 80-120℃, hydrogen pressure range 0.5-1.0MPa, weight matrix biased towards pressure regulation (pressure weight coefficient 0.4, temperature 0.3, hydrogen flow rate 0.2, stirring 0.1).

[0112] Supported catalyst control parameter set: applicable to supported catalysts such as Ni-Ag / SBA-15, reaction temperature range 100-150℃, hydrogen pressure range 0.4-0.6MPa, weight matrix biased towards hydrogen flow rate adjustment (hydrogen flow rate weight coefficient 0.35, temperature 0.3, pressure 0.25, stirring 0.1).

[0113] Historical data tracing submodule:

[0114] The historical data traceability submodule records the reaction temperature, hydrogen pressure, hydrogen flow rate, stirring rate, real-time trans fatty acid content, and iodine value of each hydrogenation reaction, forming a complete product quality traceability record and supporting batch query and trend analysis.

[0115] Example 6: Multi-mode control switching and safety interlocking

[0116] Multi-mode control switching logic:

[0117] The control unit has built-in multi-mode control switching logic, which automatically selects the control mode based on the deviation between the real-time trans fatty acid content and the target threshold for trans fatty acids.

[0118] When the deviation exceeds a preset first threshold (e.g., deviation > 15%), a fast response control mode is adopted, prioritizing the adjustment of hydrogen flow rate and reaction temperature to quickly suppress the further formation of trans fatty acids.

[0119] When the deviation is less than the preset second threshold (e.g., deviation < 5%), a fine adjustment control mode is adopted, prioritizing the adjustment of the stirring rate to achieve fine-tuning control.

[0120] When the deviation is between the first and second thresholds, a multi-parameter collaborative control mode is adopted to synchronously adjust all process parameters.

[0121] The first and second thresholds are preset judgment thresholds based on the allowable deviation range of trans fatty acids, and can be set through the human-machine interface according to the production process requirements.

[0122] Early warning and safety interlocking unit:

[0123] The system also includes an early warning and safety interlock unit, which communicates with the control unit and has preset hierarchical safety trigger rules:

[0124] When the real-time content of trans fatty acids exceeds the preset warning threshold (such as 90% of the target threshold), an audible and visual alarm will be automatically triggered to alert the operator.

[0125] When the real-time content of trans fatty acids exceeds the preset shutdown safety threshold (e.g., 110% of the target threshold), or when the continuous calibration and repair of the spectral signal quality fails to return to the qualified range after a preset time (e.g., 30 minutes), the emergency shutdown interlock of the hydrogenation reactor is automatically triggered to cut off the hydrogen supply and ensure production safety.

[0126] Effect Example

[0127] The online monitoring and closed-loop control system for trans fatty acid content in hydrogenated oils, as described in this embodiment, was tested for industrial application in an oil processing enterprise. The verification period was 30 days of continuous operation. The raw material was soybean oil (initial iodine value 132 g I2 / 100g), the catalyst was a nickel-based catalyst, the target threshold for trans fatty acids was set to ≤5%, and the target range for iodine value was set to 80-90 g I2 / 100g.

[0128] To verify the technical effectiveness of this invention, a control group experiment was set up: the control group used a traditional offline detection method (sampling every 2 hours, sending samples to the laboratory for gas chromatography analysis, and manually adjusting process parameters); the experimental group used the system of this invention (real-time online detection, automatic closed-loop control). Both groups of experiments were run in parallel under the same conditions of raw materials, the same catalyst batch, and the same target threshold.

[0129] The application results are statistically analyzed as follows:

[0130] The pass rate for trans fatty acid control was 98.5% in the experimental group and 85.2% in the control group, representing an increase of 15.6%.

[0131] Iodine value stability control rate (stable within the range of 80-90 g I2 / 100g): 96.3% in the experimental group and 78.5% in the control group, representing an improvement of 17.8%.

[0132] The detection cycle was 2-4 hours per test for the control group, while it was shortened to 5 minutes per test for the experimental group, achieving true real-time monitoring.

[0133] Response time: From detection to completion of process parameter adjustment, the control group required an average of 45 minutes (including manual judgment and operation time), while the experimental group shortened it to less than 2 minutes, improving the control response speed by more than 95%.

[0134] Trans fatty acid exceedance early warning: Through the trend prediction submodule, the system can issue an early warning 15-20 minutes before the exceedance occurs, providing sufficient time for process adjustments. The control group has no early warning capability.

[0135] Catalyst activity compensation: After the catalyst management module was activated, the consistency of control performance between different batches of catalyst (measured by the standard deviation of trans fatty acid pass rate) decreased from 4.2% before activation to 2.9%, an improvement of approximately 30%.

[0136] Product quality batch consistency: The standard deviation of iodine value for the experimental group (30 batches) was 2.1 g I2 / 100g, while that for the control group was 4.8 g I2 / 100g, indicating a significant improvement in batch consistency.

[0137] The experimental data above show that the online monitoring and closed-loop control system for trans fatty acid content in hydrogenated oils of the present invention can effectively solve the problems existing in the prior art, realize the real-time monitoring and precise control of trans fatty acids, and has significant industrial application value.

[0138] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Various changes made within the scope of knowledge possessed by those skilled in the art without departing from the concept of the present invention still fall within the scope of protection of the present invention.

Claims

1. An on-line monitoring and closed-loop control system for the trans-fatty acid content of hydrogenated oils, comprising a sampling and pretreatment unit, a spectral detection unit, a control unit, and a hydrogenation execution unit, which are配套设置于氢化油脂加氢反应装置,所述加氢反应装置包括加氢反应釜、氢气供给管路、搅拌组件,其特征在于: The sampling end of the sampling and pretreatment unit extends into the reaction material cavity of the hydrogenation reaction kettle, and is used for on-line impurity removal, defoaming, constant temperature, constant pressure and pressure stabilization treatment of the high-temperature and high-pressure solid-containing material during the hydrogenation reaction process, and outputs interference-free待测物料;The sampling and pretreatment unit includes a high-temperature and high-pressure resistant in-situ sampling probe extending into the hydrogenation reaction kettle, and an on-line impurity removal component, a defoaming component and a constant temperature, constant pressure and pressure stabilization component that are sequentially sealed and connected to form an integrated on-line pretreatment pipeline assembly; The spectral detection unit is an explosion-proof near-infrared spectral detection unit adapted to the high-risk flammable atmosphere of the hydrogenation reaction, and is hermetically connected to the output end of the sampling and pretreatment unit. It is内置有氢化油脂反式脂肪酸与碘值的定量预测模型,用于对所述无干扰待测物料进行实时光谱采集与分析,同步输出反式脂肪酸实时含量数据和碘值实时数据; The control unit is内置有反式脂肪酸靶向闭环调控模块,The input end of the target closed-loop control module is communicatively connected to the output end of the spectral detection unit, and is used for receiving the real-time trans-fatty acid content data and the real-time iodine value data, taking the trans-fatty acid target threshold pre-stored in the system or set through the human-machine interface as the first-priority control target and the iodine value target threshold as the second-priority control target, constructing a multi-parameter coupling and collaborative control model, and outputting a hydrogenation process control instruction; The hydrogenation execution unit includes an electronically controlled proportional regulating valve arranged on the hydrogen supply pipeline, a variable-frequency speed-regulating motor传动连接于所述搅拌组件、以及与所述加氢反应釜配套的温度调控组件,The controlled end of the hydrogenation execution unit is communicatively connected to the output end of the control unit, and is used for receiving the hydrogenation process control instruction and adjusting the process parameters of the hydrogenation reaction in real time, so as to form a closed-loop feedback control link of multi-parameter coordination of flow-pressure-temperature-stirring for the whole process of the hydrogenation reaction of hydrogenated oils, and实现工艺参数的动态稳定与产品质量指标的收敛控制。 2. The online monitoring and closed-loop control system for trans fatty acid content in hydrogenated oils according to claim 1, characterized in that, The online impurity removal component is an online precision filtration component with a filtration accuracy of 5-10μm, suitable for the precision purification of high-temperature, high-viscosity materials containing solid catalyst particles; the defoaming component is an ultrasonic defoaming component; the multi-parameter coupled synergistic control model is preset with a weight matrix of the coupling influence of hydrogenation process parameters on trans fatty acid formation, the hydrogenation process parameters including reaction temperature, hydrogen pressure, hydrogen flow rate and stirring rate; the model is based on the deviation between the real-time trans fatty acid content and the target threshold of trans fatty acids, and the deviation between the real-time iodine value data and the target threshold of iodine value, and calculates and outputs a multi-parameter linkage adjustment combination that converges the deviation through the weight matrix, and outputs the corresponding hydrogenation process control command; the targeted closed-loop control module also has a built-in trans fatty acid formation trend prediction submodule based on hydrogenation reaction kinetics, which predicts the change trend of trans fatty acid content in the next 10-30 minutes based on real-time detection data and current process parameters, and outputs pre-control commands in advance.

3. The online monitoring and closed-loop control system for trans fatty acid content in hydrogenated oils according to claim 1, characterized in that, The targeted closed-loop control module achieves the first priority and second priority control objectives by constructing a dual-objective collaborative optimization objective function with constraints. The constraint is that the real-time trans fatty acid content does not exceed the target threshold for trans fatty acids, and the optimization objective is to stably control the iodine value within the target threshold range, thus achieving synergistic optimization of the hydrogenation functionality of the oil and the safety management of trans fatty acids. The targeted closed-loop control module also incorporates a dynamic weight adjustment mechanism. This mechanism automatically adjusts the weight coefficients of each process parameter in the coupled influence weight matrix based on the deviation between the real-time trans fatty acid content and the target threshold: when the deviation is less than a preset approach threshold, the weight coefficients of reaction temperature and / or hydrogen pressure are preferentially increased; when the deviation is greater than a preset deviation threshold, the weight coefficients of hydrogen flow rate and / or stirring rate are preferentially increased. The approach threshold and deviation threshold are deviation judgment critical values ​​determined by a preset self-tuning algorithm or a manual preset method.

4. The online monitoring and closed-loop control system for trans fatty acid content in hydrogenated oils according to claim 1, characterized in that, The quantitative prediction model for trans fatty acids and iodine value of hydrogenated oils is a dynamic generalization prediction model built based on transfer learning. It uses hydrogenated oil samples from different raw material oils, different catalytic systems, and different hydrogenation conditions as the basic dataset. The model eliminates the distribution shift caused by the difference in operating conditions through a domain adaptive algorithm, so as to overcome the influence of spectral drift and sample distribution differences at the hydrogenation site and adapt to the hydrogenation reaction scenarios of various raw material oils such as soybean oil, rapeseed oil, and palm oil.

5. The online monitoring and closed-loop control system for trans fatty acid content in hydrogenated oils according to claim 1, characterized in that, The spectral detection unit has a built-in spectral preprocessing module and a spectral signal quality self-diagnosis module. The spectral preprocessing module is equipped with a multivariate scattering correction algorithm combined with a wavelet transform denoising algorithm to perform baseline correction, scattering interference elimination, and noise filtering on the acquired raw near-infrared spectrum, outputting standardized spectral data and inputting it into the quantitative prediction model. The spectral signal quality self-diagnosis module monitors the signal-to-noise ratio and baseline drift of the near-infrared spectrum in real time. When the signal-to-noise ratio is lower than a preset threshold or the baseline drift exceeds a preset range, it sends a signal to the control unit. The control unit then instructs the isothermal and pressure stabilizing component in the sampling preprocessing unit to perform a parameter calibration process. After calibration, the spectral data is reacquired until the comprehensive spectral quality index meets the preset admission criteria.

6. The online monitoring and closed-loop control system for trans fatty acid content in hydrogenated oils according to claim 1, characterized in that, The control unit also has a built-in catalyst management module, which includes a catalyst activity self-learning submodule, a catalyst type adaptive submodule, and a historical data traceability submodule; wherein: The catalyst activity self-learning submodule records the change curve of trans fatty acid formation rate during each hydrogenation reaction, compares and analyzes it with historical reaction data, and dynamically updates the catalyst activity correction factor in the multi-parameter coupled synergistic regulation model to offset the formation deviation caused by catalyst batch differences and activity decay in real time. The catalyst type adaptive submodule automatically switches the set of control parameters in the multi-parameter coupled synergistic regulation model according to the type of catalyst used in the current hydrogenation reaction, so as to adapt to the differences in hydrogenation characteristics of nickel-based catalysts, noble metal catalysts or catalysts with different supports. The historical data traceability submodule records the reaction temperature, hydrogen pressure, hydrogen flow rate, stirring rate, real-time trans fatty acid content, and iodine value of each hydrogenation reaction, which is used for full-chain traceability of product quality.

7. The online monitoring and closed-loop control system for trans fatty acid content in hydrogenated oils according to claim 1, characterized in that, The control unit also has a built-in multi-mode control switching logic, which automatically selects the control mode based on the deviation between the real-time trans fatty acid content and the target threshold for trans fatty acids: when the deviation is greater than a preset first threshold, a fast response control mode is adopted, prioritizing the adjustment of hydrogen flow rate and reaction temperature; when the deviation is less than a preset second threshold, a fine adjustment control mode is adopted, prioritizing the adjustment of stirring rate; when the deviation is between the first and second thresholds, a multi-parameter collaborative control mode is adopted, synchronously adjusting all process parameters; wherein, the first threshold is greater than the second threshold, and the first and second thresholds are preset judgment thresholds based on the allowable deviation range of trans fatty acids.

8. The online monitoring and closed-loop control system for trans fatty acid content in hydrogenated oils according to claim 1, characterized in that, It also includes an early warning and safety interlock unit, which is communicatively connected to the control unit and has preset graded safety triggering rules: when the real-time content of trans fatty acids exceeds the preset early warning threshold, an audible and visual alarm is automatically triggered; when the real-time content of trans fatty acids exceeds the preset shutdown safety threshold, or when the spectral signal quality fails to return to the qualified range after a preset time of continuous calibration and repair attempts, an emergency shutdown interlock of the hydrogenation reaction unit is automatically triggered.

9. The online monitoring and closed-loop control system for trans fatty acid content in hydrogenated oils according to claim 6, characterized in that, The catalyst type adaptive submodule has a built-in catalyst type parameter library, which includes at least a nickel-based catalyst control parameter set, a noble metal catalyst control parameter set, and a supported catalyst control parameter set. It is used to automatically call the corresponding control parameter set according to the catalyst type used in the current hydrogenation reaction.

Images