Fault diagnosis and automatic control system for packed bed continuous hydrogenation process unit

Abstract

The present invention belongs to the field of automatic control technology, and specifically relates to a fault diagnosis and automatic control system for a packed bed continuous hydrogenation process device. The system includes: a model building part, which is used to model the material balance of the reactor to obtain a material balance model; establish a reaction kinetics and heat balance coupling model; establish a pressure field distribution model; establish a catalyst deactivation model to obtain the activity distribution of the catalyst; a state observation part, which is used to perform tensor decomposition of the fault deviation vector according to the activity part of the catalyst, the pressure distribution, and the concentration distribution and temperature distribution of each component to obtain a fault feature tensor; a diagnosis and automatic control part, which is used to calculate the probability of abnormal state according to the fault feature tensor, generate a compensation controller according to the probability of abnormal state, and use the compensation controller for automatic control. The present invention realizes comprehensive monitoring and precise control of packed bed reactors.

Description

Technical Field

[0001] The invention belongs to the technical field of automatic control, and in particular relates to a fault diagnosis and automatic control system for a packed bed continuous hydrogenation process device. Background Art

[0002] In modern chemical production, the packed bed continuous hydrogenation process unit is a widely used type of high-efficiency reactor, especially suitable for hydrogenation reactions in the fields of petrochemicals, fine chemicals and pharmaceuticals. The basic principle of the packed bed reactor is to carry out continuous hydrogenation reactions on the catalyst bed, and to achieve the reaction process by using the contact between the reactants and the catalyst. Although this type of reactor has the advantages of high efficiency, low energy consumption and easy industrialization, in long-term operation, the packed bed reactor also faces a series of technical difficulties, including catalyst deactivation, temperature and pressure fluctuations, uneven fluid distribution, and identification and control of system failures.

[0003] In the prior art, the fault diagnosis and control strategies of packed bed reactors mainly rely on traditional single parameter monitoring methods, such as real-time observation of the temperature, pressure or fluid flow rate inside the reactor. This monitoring method based on a single parameter can provide basic status information of the reactor to a certain extent, but due to the complex multiphase flow and chemical reaction involved in the packed bed reactor, a single parameter is often difficult to fully reflect the operating status of the system. This technical limitation leads to the following major problems faced by existing packed bed reactors during operation: First, in terms of catalyst deactivation, the catalyst in the packed bed reactor is easily affected by various factors such as carbon deposition, metal pollution, and thermal deactivation due to long-term contact with reactants and products, resulting in a gradual decrease in activity. In the prior art, most monitoring methods mainly evaluate the activity changes of catalysts through periodic reaction performance tests. This method requires the operation of the reactor to be suspended for detection, which not only affects the production efficiency, but also makes it difficult to identify the failure process of the catalyst in a timely manner. In addition, due to the certain accumulation and irreversibility of the catalyst deactivation process, traditional monitoring methods often cannot warn of the occurrence of deactivation in advance, but can only take remedial measures after failure, resulting in large production losses. Secondly, the temperature distribution and pressure fluctuations in the packed bed reactor are also important factors affecting the stability of the reactor. Existing temperature monitoring technology usually obtains local temperature data by arranging temperature sensors inside the reactor. However, due to the significant spatial heterogeneity of the temperature distribution inside the reactor, single-point or multi-point temperature monitoring methods are often difficult to fully capture the overall temperature changes in the reactor. This limitation of temperature monitoring may lead to untimely response to temperature fluctuations, which in turn causes reaction rate fluctuations or local overheating failure of the catalyst. Similarly, monitoring of pressure fluctuations is also a difficulty in the prior art. The fluid flow in a packed bed reactor is greatly affected by the catalyst particle bed, and the flow resistance will change with changes in the bed structure. Existing pressure monitoring methods are mainly based on direct measurement of pressure sensors, but this method can often only provide axial or local pressure change information, and it is difficult to fully reflect the overall pressure field distribution in the reactor, resulting in insufficient accuracy when dealing with problems such as fluid blockage or particle breakage.

[0004] In addition, in terms of fault diagnosis and control, traditional packed bed reactor control strategies mostly use fixed control parameters, which are difficult to dynamically adjust according to changes in actual operating conditions. This fixed control strategy often shows insufficient adaptability when faced with complex and changeable reaction conditions. For example, when the concentration or flow rate of the reactants fluctuates, the fixed control parameters may not be able to effectively maintain the stable operation of the reactor, resulting in reduced product quality or excessive stress accumulation in the reactor. At the same time, the control methods in the prior art usually rely on a single PID control or linear controller, which is difficult to cope with multiple failure modes and nonlinear complex conditions. Summary of the invention

[0005] The main purpose of the present invention is to provide a fault diagnosis and automatic control system for a packed bed continuous hydrogenation process device, thereby achieving comprehensive monitoring and precise control of a packed bed reactor.

[0006] In order to solve the above technical problems, the present invention provides a fault diagnosis and automatic control system for a packed bed continuous hydrogenation process device, the system comprising: a model building part, used to perform material balance modeling on the reactor to obtain a material balance model; according to the material balance model, the concentration distribution of each component is obtained to establish a reaction kinetics and heat balance coupling model; according to the reaction kinetics and heat balance coupling model, the temperature distribution is obtained, combined with the concentration part of each component, and a pressure field distribution model is established; according to the pressure field distribution model, the pressure distribution is obtained, combined with the concentration distribution and temperature distribution of each component, a catalyst deactivation model is established to obtain the activity distribution of the catalyst; a state observation part, used to construct a fault feature vector according to the activity part of the catalyst, the pressure distribution, combined with the concentration distribution and temperature distribution of each component, and perform deviation operation on the fault feature vector to obtain a fault deviation vector; perform tensor decomposition on the fault deviation vector to obtain a fault feature tensor; a diagnosis and automatic control part, used to calculate the probability of an abnormal state according to the fault feature tensor, generate a compensation controller according to the probability of the abnormal state, and use the compensation controller for automatic control.

[0007] Furthermore, the formula of the material balance model obtained in the model building part is:

[0008] ;

[0009] in, For components The concentration distribution of For time; stands for Laplace operator; For components The effective diffusion coefficient of In the spatial diffusion behavior within the reactor, since the packed bed reactor has particulate catalysts, the diffusion behavior is limited by the internal pores and external particle size, so the effective diffusion coefficient is used to characterize the actual diffusion situation; is the total number of reactions, which indicates the number of all chemical reactions in the reactor; is the stoichiometric coefficient, indicating the component In reaction The amount of generation or consumption in . A negative value indicates that the component In reaction If it is a positive value, it means it is generated; For reaction The rate constant of Representative reaction The rate equation is the product of the concentrations of all relevant components, is the number of all components, It is a component In reaction The reaction order in ; Describe the components Convection process in the direction of fluid flow, is the linear velocity, which indicates the average velocity of the fluid in the reactor; It represents the axial coordinate of the reactor, that is, the coordinate inside the reactor along the direction of fluid flow.

[0010] Furthermore, the formula of the reaction kinetics and heat balance coupling model is:

[0011] ;

[0012] in, is the fluid density; is the specific heat capacity; is the temperature distribution; It represents the effective thermal conductivity, which reflects the thermal conductivity between the fluid and solid particles in the reactor; is the fluid flow rate; Indicates The heat of reaction of each reaction, a positive value indicates an endothermic reaction, a negative value indicates an exothermic reaction; Indicates the temperature of the jacket; Indicates the diameter of the reactor; heat transfer coefficient It reflects the heat transfer effect between the reactor and the jacket.

[0013] Furthermore, the formula of the pressure field distribution model is:

[0014] ∂ P ∂ t =- 150 m (1- ϵ ) 2 d p 2 ϵ 3 v s - 1.75 r (1- ϵ ) d p ϵ 3 v s 2 + ∂ ∂ z [ K p ∂ P ∂ z ]+ R g ∑ i =1 N c C i ∂ T ∂ t ;

[0015] in, is the pressure distribution; is the fluid viscosity; is the bed void ratio; is the catalyst particle size; is the pressure transmission coefficient; is the gas constant.

[0016] Furthermore, the catalyst deactivation model is formulated as:

[0017] ;

[0018] in, is the catalyst activity concentration distribution; is the inactivation rate constant; is the activation energy of deactivation; For components Impact index on inactivation; is the influence coefficient of pressure on inactivation; is the reference pressure.

[0019] Furthermore, the formula of the fault feature vector constructed by the state observation part is:

[0020] ;

[0021] in, For components The corresponding fault feature vector , a total of Fault feature vector.

[0022] Furthermore, the fault feature vector is subjected to deviation operation, and the formula for the fault deviation vector is obtained as follows:

[0023] ;

[0024] in, Components The corresponding fault deviation vector; For components The corresponding reference concentration distribution; is the reference temperature distribution; is the reference pressure distribution; is the reference catalyst activity concentration distribution; is the L2 norm; the fault deviation vectors corresponding to all components are regarded as one element, forming a vector as the system fault deviation vector;

[0025] Furthermore, the system fault deviation vector is decomposed into four different dimensions by the following formula to obtain the system fault feature tensor:

[0026] F = ∑ k =1 K l k ⋅[ C k (Φ)⊗ T k (Φ)⊗ P k (Φ)⊗ a k (F)] ⋅exp(- o k D t ) ;

[0027] in, is the number of eigenvalues ​​of the system fault deviation vector; is the tensor corresponding to the concentration dimension; is the tensor corresponding to the temperature dimension; is the tensor corresponding to the pressure dimension; is the tensor corresponding to the catalyst activity dimension; For the The time decay coefficient corresponding to the eigenvalue; For the feature values.

[0028] Furthermore, the diagnosis and automation control part calculates the probability of abnormal state through the following formula: :

[0029] ;

[0030] in, is the fault template tensor; is the total system failure rate set based on prior historical data; is the total system failure rate under the fault template tensor set according to the prior historical data; Modulo operation on vectors.

[0031] Furthermore, the compensation controller formula is:

[0032] u ( t )= K c ( t )⋅[ e ( t )+ 1 t I ∫ 0 t e ( t ) dt + t D de ( t ) dt ]+ ∑ i =1 N c P( pro ) ⋅ C i (|Φ i |) ;

[0033] in, Control inputs generated for compensation controllers; To control the gain in real time; is the error signal; is the integration time constant; is the differential time constant; For components The gamma function value corresponding to the modulus of the corresponding fault deviation vector; is the integration variable.

[0034] The fault diagnosis and automatic control system of the packed bed continuous hydrogenation process device of the present invention has the following beneficial effects: the present invention covers multiple key factors such as material balance, reaction kinetics and heat balance coupling, pressure field distribution and catalyst deactivation by establishing a multi-layer coupling model, so that the system can comprehensively and accurately capture various physical and chemical behaviors inside the reactor. By constructing the material balance model, the concentration changes of each component can be dynamically grasped, while the reaction kinetics and heat balance coupling model combines the reaction rate and thermal effect to accurately reflect the temperature changes in the reactor. The pressure field distribution model and the catalyst deactivation model further supplement the description of fluid flow and catalyst activity. This multi-level coupling modeling method enables the system to provide comprehensive state data and effectively improves the overall grasp of the reactor operation state. The present invention integrates multiple key parameters such as concentration, temperature, pressure and catalyst activity into a multidimensional vector by constructing a fault feature vector, which can more comprehensively reflect the operation state of the reactor. Compared with the traditional single parameter monitoring method, the multi-dimensional feature extraction method can not only capture the changes of each key parameter, but also identify complex fault modes through multi-parameter fusion. In particular, by introducing tensor decomposition technology, the fault deviation vector is decomposed in multiple dimensions, and the main fault features can be extracted from high-dimensional data. The use of tensor decomposition technology not only improves the accuracy of fault feature extraction, but also effectively solves the dimensionality disaster problem in high-dimensional data processing. Through this method, the system can identify early signs of multiple fault modes earlier and more accurately, providing strong data support for fault diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0036] Figure 1 A schematic diagram of the system structure of a fault diagnosis and automation control system for a packed bed continuous hydrogenation process device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0037] The method of the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments of the present invention.

[0038] Example 1, reference Figure 1 : A fault diagnosis and automatic control system for a packed bed continuous hydrogenation process device, the system comprising: a model building part, used to perform material balance modeling on the reactor to obtain a material balance model; based on the material balance model, obtain the concentration distribution of each component and establish a reaction kinetics and heat balance coupling model; based on the reaction kinetics and heat balance coupling model, obtain the temperature distribution and establish a pressure field distribution model in combination with the concentration part of each component; based on the pressure field distribution model, obtain the pressure distribution and establish a catalyst deactivation model in combination with the concentration distribution and temperature distribution of each component to obtain the activity distribution of the catalyst; a state observation part, used to construct a fault feature vector based on the activity part of the catalyst, the pressure distribution, and the concentration distribution and temperature distribution of each component, perform deviation calculation on the fault feature vector to obtain a fault deviation vector; perform tensor decomposition on the fault deviation vector to obtain a fault feature tensor; a diagnosis and automatic control part, used to calculate the probability of an abnormal state based on the fault feature tensor, generate a compensation controller based on the probability of an abnormal state, and use the compensation controller for automatic control.

[0039] Specifically, the material balance model determines the trend of each component in the reaction process through the stoichiometric relationship, so that the concentration of each chemical substance in the reaction process can be monitored and predicted in real time. This is very important because the change in concentration will not only affect the reaction rate, but also have a significant impact on the reaction heat effect, which will lead to temperature changes. Therefore, material balance modeling is not limited to the quantitative balance control of each component, but also provides a concentration basis for the subsequent fine analysis of reaction kinetics and heat distribution. After obtaining the concentration distribution, the system further combines the reaction kinetics and heat balance models for coupled modeling. The construction of the reaction kinetics model is based on the mechanism of chemical reactions. The main purpose is to capture the chemical reaction rate and reaction order in the reaction process, which is closely related to factors such as temperature, pressure, and concentration. Through kinetic modeling, the reaction rate and the path of the reaction can be determined, and this rate change will further affect the heat change in the reaction process. In order to enable the model to accurately reflect the real situation inside the reactor, the system combines reaction kinetics with heat balance to describe the temperature change and energy flow in the reaction process in the form of a coupled model.

[0040] The significance of coupled modeling is that it can simultaneously consider the consumption and generation of substances in the reaction process, as well as the thermal effects associated with the reaction. In this way, the temperature distribution inside the reactor can be accurately predicted, which is crucial for the control of packed bed reactors. Changes in temperature not only affect the reaction rate, but also cause changes in the pressure distribution inside the reactor, which in turn affects the operating state and operational stability of the reactor. Therefore, when establishing a coupled model of reaction kinetics and heat balance, it is necessary to further combine the rate equation and thermodynamic equation of the chemical reaction based on the concentration information obtained from the material balance to ensure accurate simulation of temperature changes. Next, after obtaining the temperature distribution through the reaction kinetics and heat balance model, the system further established a pressure field distribution model. Pressure distribution is one of the important parameters of a packed bed reactor, because the fluctuation of pressure during the reaction directly affects the reaction rate, mass transfer effect and the effective service life of the catalyst. The construction of the pressure field distribution model is based on the principle of fluid mechanics, combined with the concentration and temperature distribution inside the reactor, to describe the pressure changes in various regions of the reactor. In order to improve the accuracy of the model, the system not only considers the fluid dynamic characteristics inside the reactor, but also combines the dynamic changes of temperature and concentration to ensure the real-time and reliability of the pressure field distribution.

[0041] By establishing a material balance model, a reaction kinetics and heat balance coupling model, and a pressure field distribution model, the system further combines the output information of these models to construct a catalyst deactivation model. The catalyst is easily affected by carbon deposition, thermal deactivation, metal contamination, etc. during the reaction process, which leads to a gradual decrease in its activity. Therefore, the purpose of establishing a catalyst deactivation model is to monitor the activity changes of the catalyst in real time in order to warn and control the failure of the catalyst. The establishment of this model is based on the analysis of the deactivation rate. By combining key parameters such as concentration, temperature and pressure, the system can accurately predict the activity change trend of the catalyst. It does not just stop at a single material balance and reaction kinetics analysis, but through the coupling modeling of reaction kinetics and heat balance, it further combines the principles of fluid mechanics and the catalyst deactivation mechanism to construct a comprehensive and accurate reactor state monitoring and diagnosis model. This multi-level, multi-physical field coupling modeling method enables the system to grasp the changes in various key parameters inside the reactor in real time, identify abnormal conditions in a timely manner and respond. In addition, the modeling part of this system achieves efficient data integration and real-time information update through the interaction of multi-level models. Traditional packed bed reactor control systems often ignore the mutual coupling between different physical fields, which limits the accuracy and reliability of the model. However, the present invention organically combines multi-level models such as material balance, reaction kinetics, heat balance and pressure distribution, which can not only accurately predict the state changes inside the reactor, but also provide comprehensive data support for subsequent fault diagnosis and automatic control. This all-round, multi-level modeling method significantly improves the automation level and safety of the packed bed continuous hydrogenation process unit.

[0042] The state observation part first obtains the activity distribution of the catalyst in real time according to the established catalyst deactivation model. The activity of the catalyst is an important indicator reflecting the efficiency of the chemical reaction in the reactor, and its change directly affects the reaction rate and the conversion rate of the final product. Therefore, by observing the activity of the catalyst, the system can dynamically evaluate the catalytic reaction state of the reactor and identify potential deactivation trends. Combined with the activity distribution obtained by the catalyst deactivation model, this part also uses the pressure, concentration and temperature distribution data output by the material balance, reaction kinetics and heat balance coupling model to comprehensively analyze the operating conditions inside the reactor. The system provides rich basic information for the extraction of fault features by real-time collection and fusion of these multi-dimensional data. After obtaining these key parameters, the state observation part constructs a fault feature vector containing multiple state variables. The fault feature vector not only contains the concentration, temperature and pressure of each component in the reactor, but also combines the activity of the catalyst, which enables the feature vector to fully reflect the operating state of the reactor. Compared with the traditional single state monitoring method, this system can effectively capture the feature differences under different fault modes through multi-parameter and multi-dimensional data fusion. This multi-dimensional fusion of feature vectors enables the system to more accurately identify and classify fault characteristics when dealing with complex working conditions and multiple fault modes.

[0043] In order to further enhance the recognition accuracy of fault features, the state observation part introduces the calculation of fault deviation vector. The fault deviation vector quantitatively describes the abnormality of the reactor operation state by calculating the deviation between the current observed value of the fault feature vector and the historical normal value. The introduction of the deviation vector can effectively capture the change trend of the system when a fault occurs, and provide a clearer judgment basis for fault diagnosis by calculating the deviation degree of each state variable. In the calculation process of the fault deviation vector, the system will set the baseline feature value under the historical normal state and use it as a comparison benchmark to calculate the deviation of the current feature vector in real time. In this way, the state observation part can not only detect obvious fault features, but also capture potential faults caused by the accumulation of small changes in time. After obtaining the fault deviation vector, the state observation part further processes the fault deviation vector through tensor decomposition technology. The introduction of tensor decomposition is to better handle high-dimensional and complex fault feature data. By expressing the fault deviation vector as a multidimensional tensor, the system can more clearly identify the main components and change trends of the fault features in the multidimensional data space. Specifically, tensor decomposition can reduce the dimensionality of high-dimensional data while retaining the main features of the data, thereby simplifying the complexity of the data and improving the accuracy of fault diagnosis. In this process, the system uses advanced tensor decomposition algorithms, such as high-order singular value decomposition or high-order matrix decomposition, to extract fault feature tensors from fault deviation vectors. The generation of fault feature tensors provides a high-dimensional data expression for further diagnosis and control of the system, greatly enhancing the identifiability of fault features.

[0044] Compared with the traditional fault diagnosis method, this part not only improves the accuracy of feature extraction through multi-dimensional data fusion, but also solves the common dimensional disaster problem in high-dimensional data processing by introducing tensor decomposition technology, so that the system can still maintain a high diagnostic accuracy under complex working conditions. This innovative design enables the fault diagnosis system of the packed bed continuous hydrogenation process device to still have strong adaptability and robustness when facing a variety of fault modes and operating conditions. In addition, the state observation part can provide accurate fault information for subsequent automatic control through real-time observation and deviation analysis of fault characteristics. Compared with the traditional single control strategy, this system can dynamically adjust the control strategy according to the observed multi-dimensional fault characteristics, further improving the safety and operation efficiency of the reactor. Under the subject of the present invention, the state observation part forms an accurate and efficient state monitoring system by integrating the output data of multi-level models such as material balance, kinetic and thermal coupling model, and pressure distribution model. This multi-dimensional coupled data processing method significantly improves the fault detection and diagnosis capabilities of the system under complex working conditions, so that the packed bed continuous hydrogenation process device can achieve a higher level of automatic control and operation optimization.

[0045] Based on the fault feature tensor, the system first evaluates the abnormal state of the reactor through a probability calculation model. The composition of the fault feature tensor contains multi-dimensional information, such as the concentration distribution of each component, temperature distribution, pressure distribution, and catalyst activity, all of which reflect the operating state of the reactor under the current working conditions. Therefore, when evaluating the abnormal state, the system uses a model based on Bayesian theory or machine learning algorithm to process these high-dimensional data. By inferring the joint probability distribution of these data, the probability of occurrence of a specific fault mode can be calculated. This probability-based evaluation method can not only quantitatively describe the risk of fault occurrence, but also effectively distinguish between multiple fault modes, thereby avoiding the misjudgment and missed judgment problems caused by simple threshold judgment in traditional fault diagnosis. Once the system calculates the abnormal state probability of the reactor based on the fault feature tensor, the diagnosis and automation control part will generate a compensation controller based on this probability. The generation process of the compensation controller is dynamically adjusted according to different fault modes and states. This process draws on the theory of adaptive control and fuzzy control, and realizes automatic compensation for faults by adjusting the operating parameters of the reactor in real time, such as temperature, pressure, flow, etc. The core of the compensation controller is that it can automatically adjust the control strategy according to the operating state and fault mode of the reactor to ensure that the reactor can still maintain a stable operating state when a fault occurs. Unlike the traditional fixed control strategy, the compensation controller realizes the ability to dynamically adjust the control strategy according to the fault situation through an adaptive algorithm, which greatly improves the intelligence and robustness of the system. In addition, another key principle of the diagnosis and automatic control part is to use the feedback loop to optimize the control effect. After generating the compensation controller, the system will adjust the operating parameters of the reactor according to the output of the compensation controller. At the same time, the system will continue to observe the state changes of the reactor and feed back the new observation data to the control system to update the parameters of the compensation controller in real time. This closed-loop control method keeps the operating state of the reactor within the preset safety range through continuous observation and adjustment, avoiding problems such as overcompensation or insufficient control caused by a single adjustment. Through this real-time feedback and dynamic adjustment strategy, the system can always maintain precise control of the reactor state under various complex working conditions.

[0046] Example 2: The formula of the material balance model obtained in the model building part is:

[0047] ;

[0048] in, For components The concentration distribution of For time; stands for Laplace operator; For components The effective diffusion coefficient of In the spatial diffusion behavior within the reactor, since the packed bed reactor has particulate catalysts, the diffusion behavior is limited by the internal pores and external particle size, so the effective diffusion coefficient is used to characterize the actual diffusion situation; is the total number of reactions, which indicates the number of all chemical reactions in the reactor; is the stoichiometric coefficient, indicating the component In reaction The amount of generation or consumption in . A negative value indicates that the component In reaction If it is a positive value, it means it is generated; For reaction The rate constant of Representative reaction The rate equation is the product of the concentrations of all relevant components, is the number of all components, It is a component In reaction The reaction order in ; Describe the components Convection process in the direction of fluid flow, is the linear velocity, which indicates the average velocity of the fluid in the reactor; It represents the axial coordinate of the reactor, that is, the coordinate inside the reactor along the direction of fluid flow.

[0049] Specifically, part of the formula involves the components Time Variation , which reflects the components in the reactor The concentration of the dynamic change over time. For the packed bed continuous hydrogenation process unit, the time change rate is a key indicator, which can indicate the rate of reaction and mass transfer in the system. Therefore, by observing this term, the dynamic change trend of each component can be grasped in real time, and whether there is abnormal concentration fluctuation inside the reactor can be identified, thereby providing early warning for fault diagnosis. Secondly, the diffusion term in the formula Describes the diffusion behavior of components in space. Diffusion refers to the natural migration of substances from high concentration areas to low concentration areas. In packed bed reactors, due to the presence of internal catalyst particles, the diffusion behavior is not only affected by the catalyst pores, but also related to the size and shape of the particles. Therefore, the effective diffusion coefficient , the model can more realistically reflect the actual diffusion of each component in the reactor. In the process of fault diagnosis, the change of this diffusion term can reflect the abnormal mass transfer phenomenon in the system, such as blockage, carbon deposition and other problems, which will directly affect the reaction rate and product quality.

[0050] In addition, the chemical reaction term in the formula describes the generation and consumption of components during the reaction. The reaction term is composed of multiple reaction rate equations, which are expressed by stoichiometric coefficients. and rate constant For each reaction, the form of the rate equation depends on the kinetic characteristics of the reaction, that is, the concentration of each component involved in the reaction and its reaction order. This part can accurately describe the dynamic process of multiple chemical reactions taking place simultaneously. Since there are usually multiple parallel or series reactions in the packed bed hydrogenation process, the reaction rate is crucial to the overall behavior of the system. If the rate of a certain reaction is abnormal, it may mean the deactivation of the catalyst or changes in certain reaction conditions. Therefore, through real-time monitoring of this term, possible faults in the reactor can be detected. In addition, the convection term in the formula Describes the convective mass transfer process of fluid in the reactor. The convection effect is usually accompanied by the overall movement of the fluid in the reactor, while in a packed bed reactor, the convection process is mainly affected by the fluid linear velocity. The existence of convection enables the material to be transferred along the axial direction of the reactor, thus forming a stable concentration gradient and temperature gradient. For hydrogenation reaction systems, convective mass transfer is crucial because it affects the material supply in each reaction zone and the discharge of reaction products. Changes in convection rate may reflect problems such as improper flow rate control or pressure fluctuations in the system, which may cause serious safety hazards under high-pressure operating conditions. Therefore, by monitoring the convection item, it is possible to determine whether the operating status of the reactor is normal and detect potential faults in a timely manner.

[0051] In general, the formula of the material balance model takes into account factors such as diffusion, reaction and convection, and can fully reflect the material migration and reaction characteristics of the packed bed reactor. Compared with the traditional chemical reaction model, this formula is more in line with the operating characteristics of the actual process unit. and convection terms, which can more accurately reflect the interaction between the catalyst and the fluid. This multi-factor coupled model design enables the system to rely on accurate data analysis to form effective decision support when performing fault diagnosis and automatic control. During the fault diagnosis process, the model can provide the system with comprehensive fault characteristic data by real-time monitoring of the time change of concentration, diffusion behavior, reaction rate and convection effect. When the concentration distribution or change rate of a component is abnormal, combined with the changes in each term in the formula, the potential source of the fault can be quickly locked, and through subsequent tensor decomposition and automatic control modules, effective fault detection and control strategy adjustment can be achieved. Therefore, the importance of the model in reactor fault diagnosis and automatic control is self-evident. It not only provides a basis for fault detection, but also lays a solid foundation for the generation of compensation controllers and the optimization of control strategies.

[0052] Example 3: The formula of the reaction kinetics and heat balance coupling model is:

[0053] ;

[0054] in, is the fluid density; is the specific heat capacity; is the temperature distribution; It represents the effective thermal conductivity, which reflects the thermal conductivity between the fluid and solid particles in the reactor; is the fluid flow rate; Indicates The heat of reaction of a reaction, a positive value indicates an endothermic reaction, a negative value indicates an exothermic reaction; Indicates the temperature of the jacket; Indicates the diameter of the reactor; heat transfer coefficient It reflects the heat transfer effect between the reactor and the jacket.

[0055] Specifically, the left side of the formula represents the rate of change of the temperature in the reactor over time. represents the density of the fluid, and Then represents the specific heat capacity. It reflects the ability of the fluid inside the reactor to absorb or release heat under specific density and specific heat capacity conditions. , which reveals how the temperature in the reactor changes over time. This term is critical for packed bed reactors because the dynamic change of temperature directly affects the reaction rate, equilibrium constant, and catalyst activity. On the right side of the formula, the first term describes the heat conduction process in the reactor. The effective thermal conductivity here is It is used to comprehensively characterize the heat conduction capacity between fluid and catalyst particles. In a packed bed reactor, due to the presence of catalyst particles, the heat conduction behavior is not only affected by the fluid, but also closely related to factors such as the material, particle size, and porosity of the catalyst. To characterize this complex heat conduction effect, the heat flow characteristics in the reactor can be more accurately reflected. Through this item, the system can accurately describe the temperature difference at different locations, which is very important for the control system to adjust the temperature gradient in the reactor.

[0056] Next, the second It describes the axial convection heat transfer process of the fluid in the reactor. represents the velocity of the fluid, and the negative sign indicates that this term describes the heat transfer along the direction of fluid flow. As the fluid in the reactor is constantly flowing, the convection effect causes the heat to be transferred axially, thereby forming a stable temperature gradient inside the fluid. For a packed bed reactor, this convection term reflects the process of heat migration driven by the fluid, thereby affecting the temperature distribution of the entire reactor. In the context of fault diagnosis, the observation of this term can help identify changes in fluid flow rate and abnormal heat transfer phenomena. The third term It directly reflects the contribution of the chemical reaction in the reactor to the temperature change. The heat of reaction for each reaction, exothermic reaction is a negative value, and the endothermic reaction is a positive value. The rate equation in this term Indicates that the reaction rate changes with the change of component concentration. This design can more accurately simulate the impact of reaction heat effect on temperature in packed bed reactors. Exothermic reactions release a lot of heat, causing local temperature rise, while endothermic reactions reduce the temperature. Therefore, through the accurate description of this item, the system can capture the temperature fluctuations caused by chemical reactions in the reactor in real time, so as to promptly identify abnormal thermal effects caused by changes in reaction conditions or catalyst deactivation. The last item describes the heat exchange process between the reactor and the jacket. represents the heat transfer coefficient, which is used to characterize the heat exchange capacity between the reactor wall and the jacket, while This term reflects the temperature difference between the inner and outer walls of the reactor. In a packed bed reactor, the jacket is usually used to maintain the temperature of the reactor. Therefore, the heat exchange between the reactor and the jacket is essential to maintain the thermal balance of the system. and heat transfer coefficient , the temperature distribution of the reactor can be controlled, thereby affecting the stability of the entire reaction process.

[0057] Embodiment 4: The formula of the pressure field distribution model is:

[0058] ∂ P ∂ t =- 150 m (1- ϵ ) 2 d p 2 ϵ 3 v s - 1.75 r (1- ϵ ) d p ϵ 3 v s 2 + ∂ ∂ z [ K p ∂ P ∂ z ]+ R g ∑ i =1 N c C i ∂ T ∂ t ;

[0059] in, is the pressure distribution; is the fluid viscosity; is the bed void ratio; is the catalyst particle size; is the pressure transmission coefficient; is the gas constant.

[0060] Specifically, in the fault diagnosis and automatic control system of the packed bed continuous hydrogenation process device, the pressure field distribution model proposed in Example 4 is used to describe the dynamic changes of the pressure inside the reactor over time and space. This formula comprehensively reflects the influence of various key factors on the pressure distribution in the packed bed reactor by combining the fluid dynamics characteristics, the particle bed characteristics and the heat transfer effect. This model plays a vital role in the entire system because the pressure change directly affects the reaction rate, mass transfer efficiency and stability of the reactor.

[0061] First, the left side of the formula It represents the rate of change of pressure over time. Pressure is an important parameter of the reactor operating status, which can reflect the resistance of the fluid to flow in the bed and the operating stability in the reactor. In a packed bed reactor, since the flow of the fluid in the catalyst bed is affected by many factors such as particles, porosity and viscosity, dynamic monitoring of pressure changes is crucial for fault diagnosis. Through real-time observation of this time change rate, abnormal increases in flow resistance or changes in flow rate can be identified, thereby helping the system to detect possible faults in a timely manner. The first term of the formula and the second Describes the linear pressure drop and nonlinear pressure drop caused by fluid flow, respectively. These two terms are derived from Darcy's law and Forchheimer equation, which are used to describe the pressure loss of fluid under low and high flow conditions, respectively. The fluid viscosity in the first term , bed void ratio and catalyst particle size It is an important parameter that determines the flow resistance. The porosity reflects the pore structure of the catalyst bed, while the particle size This affects the friction between the fluid and the particles. For lower flow rates, the viscosity of the fluid dominates the pressure loss, so this term shows a linear change. At higher flow rates, the inertial effect of the fluid becomes significant, and the nonlinear pressure drop term reflects this change through the square term. Packed bed reactors usually operate over a wide range of flow rates, so the combined description of these two terms can better simulate the pressure changes under different flow conditions. The third term Describes the process of pressure transmission in the axial direction. It stands for pressure conductivity, which characterizes the transfer capacity of the fluid in the bed. For packed bed reactors, the axial change of pressure is mainly affected by the pore structure of the bed and the physical properties of the fluid. The introduction of the pressure conductivity enables the model to more accurately describe the change of pressure along the axial direction of the reactor, which is crucial for monitoring the overall flow state of the bed. Through dynamic monitoring of this item, the system can identify abnormal pressure fluctuations caused by uneven distribution of the bed or blockage of the flow channel. The last item reflects the influence of thermal effect on pressure change. represents the gas constant, and the concentration and temperature change rate This shows the contribution of the concentration and temperature changes of different components to the pressure. In a packed bed reactor, chemical reactions are often accompanied by exothermic or endothermic reactions, and temperature changes can cause changes in the volume of the gas inside the reactor, which in turn leads to changes in pressure. Especially for reactions with a large amount of gas products, the effect of temperature changes on pressure is particularly significant. Through this item, the model can take into account the pressure fluctuations caused by the heat of the reaction, thereby more comprehensively describing the pressure changes in the reactor.

[0062] Overall, the pressure field distribution model can comprehensively and dynamically describe the pressure changes of the packed bed reactor by combining multiple factors such as linear and nonlinear flow resistance, axial pressure conduction and thermal effect. Unlike the traditional single flow model, this formula can more accurately capture the pressure change characteristics of the reactor under different working conditions and operating conditions by introducing multiple pressure change factors. In the packed bed continuous hydrogenation process unit, the change of pressure is an important factor affecting the stability and operation safety of the reactor. By real-time monitoring of the changes in the pressure field distribution, the system can detect faults caused by bed blockage, catalyst failure, abnormal fluid flow rate, etc. in a timely manner, thereby providing more comprehensive fault feature information. This multi-factor coupled pressure field distribution model not only enhances the system's pressure monitoring capability under complex flow conditions, but also improves the recognition accuracy of changes in the physical and chemical states inside the reactor by combining thermal effects. For the packed bed continuous hydrogenation process unit, the introduction of this model significantly improves the system's fault diagnosis accuracy and automation control level, enabling the unit to better cope with complex reaction process changes and operational challenges. Through this multi-dimensional pressure monitoring and analysis, the system can more accurately locate potential fault sources and adjust control strategies according to the severity of the fault, thereby achieving more efficient and safe process operations.

[0063] Example 5: The formula for the catalyst deactivation model is:

[0064] ;

[0065] in, is the catalyst activity concentration distribution; is the inactivation rate constant; is the activation energy of deactivation; For components Impact index on inactivation; is the influence coefficient of pressure on inactivation; is the reference pressure.

[0066] Specifically, the left side of the formula Indicates the rate of change of catalyst activity concentration over time. Catalyst activity concentration It is a key indicator to measure the effective reaction ability of the catalyst in the reactor. By calculating the time derivative of the active concentration, the deactivation process of the catalyst can be monitored in real time, thus providing basic data for fault diagnosis. The first main part on the right side of the formula represents the inactivation rate constant The relationship between the exponential factors related to temperature. Here, reflects the deactivation rate of the catalyst under standard conditions, while the exponential factor The exponential factor is derived from the Arrhenius equation, which describes the relationship between the rate of a chemical reaction and temperature. represents the energy barrier required for catalyst deactivation, and the gas constant and temperature The deactivation rate of the catalyst is determined by the temperature increase. The deactivation process is usually accelerated by the temperature increase, which is particularly significant in the long-term operation of the reactor. By monitoring and controlling this part, the catalyst deactivation problem caused by abnormal temperature increase can be identified. Next, the formula includes a concentration-related product term , which indicates the effect of the concentration of each component on the catalyst deactivation. Indicates components The influence index on catalyst deactivation determines the contribution of component concentration to the deactivation rate. In a packed bed reactor, catalyst deactivation is often closely related to the concentration of the reactants. For example, the accumulation of some side reaction products may cause contamination or poisoning of the catalyst surface, thereby reducing its activity. Therefore, the introduction of concentration-related terms can help the system more comprehensively monitor the impact of each component on the deactivation process. In addition, a pressure-related exponential factor is also introduced in the formula Here reflects the effect of pressure on catalyst deactivation, while represents the current reactor pressure. The introduction of this factor shows that changes in pressure will also have a significant impact on catalyst deactivation. In a packed bed reactor, high-pressure operating conditions usually help to increase the reaction rate, but may also lead to mechanical wear or physical failure of the catalyst. In addition, the accumulation of by-products under high-pressure conditions may also aggravate the deactivation of the catalyst. Therefore, by monitoring this index factor, deactivation problems caused by pressure fluctuations can be effectively identified and corresponding control measures can be taken.

[0067] Embodiment 6: The formula of the fault feature vector constructed in the state observation part is:

[0068] ;

[0069] in, For components The corresponding fault feature vector , a total of Fault feature vector.

[0070] Specifically, the fault feature vector Contains components Four key parameters are related: concentration ,temperature ,pressure and catalyst activity The selection of these parameters is not random, but based on an in-depth analysis of the operating characteristics of the packed bed reactor. Concentration, temperature and pressure are the core state parameters of the reactor operation, while the activity of the catalyst directly affects the efficiency of the reaction. By organically combining these key parameters, the fault feature vector can effectively reflect the component The full range of conditions in the reactor provides an accurate data basis for fault diagnosis. First, the concentration in the fault feature vector Indicates a specific component Concentration distribution in the reactor. Changes in concentration not only reflect the consumption and generation of reactants, but are also closely related to the mass transfer and reaction rate of the reactor. In a packed bed reactor, the concentration change of each component is an important indicator of process operation. When the concentration of a component deviates significantly from the expected value, it may mean that the reaction conditions or operating parameters are abnormal. Therefore, incorporating concentration into the fault feature vector can help the system quickly identify component-related faults, such as feed anomalies, reaction rate fluctuations, or the occurrence of side reactions. Secondly, the temperature in the fault feature vector is another key state variable. Changes in temperature directly affect the reaction rate, equilibrium constant, and catalyst activity. For packed bed reactors, the reaction process is often accompanied by significant exothermic or endothermic phenomena, so temperature fluctuations are an important fault signal. Real-time temperature observation can help the system identify faults caused by fluctuations in reaction conditions, failure of the cooling system, or changes in reactants. Temperature fluctuations may also be related to thermal effects. If these abnormal temperature changes are not detected and corrected in time, they may cause reactor operation instability and even pose safety hazards. Pressure in the fault feature vector It is the third important indicator of the reactor operating status. The change in pressure is closely related to the flow characteristics of the fluid, the mass transfer efficiency of the reactor, and the operating conditions. Especially in packed bed reactors, the interaction between the fluid and the catalyst bed can cause pressure fluctuations, which may reflect bed blockage, catalyst particle rupture or other mechanical problems. Therefore, real-time monitoring of pressure can help the system identify potential faults caused by changes in flow resistance or decreased mass transfer performance, thereby providing more comprehensive fault feature information. Finally, the catalyst activity in the fault feature vector Directly reflects the effective reaction ability of the catalyst in the reactor. Catalyst deactivation is one of the main factors affecting reaction efficiency and product output. In a packed bed reactor, changes in catalyst activity are usually accompanied by carbon deposition, contamination or thermal failure, which will have a profound impact on the reaction process. By incorporating catalyst activity into the fault feature vector, the system can detect the decline in catalyst activity in a timely manner and take corresponding control measures to maintain the efficient progress of the reaction.

[0071] Embodiment 7: Perform deviation operation on the fault feature vector to obtain the formula of the fault deviation vector:

[0072] ;

[0073] in, Components The corresponding fault deviation vector; For components The corresponding reference concentration distribution; is the reference temperature distribution; is the reference pressure distribution; is the reference catalyst activity concentration distribution; is the L2 norm; the fault deviation vectors corresponding to all components are regarded as one element to form a vector as the system fault deviation vector.

[0074] Specifically, in the formula, the fault deviation vector Used to describe and The degree of deviation of the four key parameters of concentration, temperature, pressure and catalyst activity is calculated. The deviation of each parameter is expressed by calculating the ratio of the L2 norm (i.e., the Euclidean distance) between the current state and the reference state. Specifically, the reference state (e.g. , , and ) usually represents the stable parameter value of the reactor under normal operating conditions, while the current state reflects the actual operation of the system. By calculating the deviation, the system can detect significant changes in a parameter relative to normal conditions, thereby quickly identifying potential faults. In the first term of the formula, Describes the current concentration With reference concentration The relative deviation between them. The L2 norm is used here to measure the difference between the two, that is, by taking the sum of the squares of the difference between the current concentration and the reference concentration, and then dividing it by the L2 norm of the reference concentration. This method of calculating the relative deviation can not only reflect the absolute difference between the current state and the ideal state, but also can be normalized by the reference value, making the measurement of the deviation more intuitive and unified. In a packed bed reactor, changes in the concentration of each component may reflect abnormalities in the reaction conditions, reaction rate or feed flow rate. Therefore, by monitoring the concentration deviation, these potential faults can be identified in a timely manner. Similarly, the second term of the formula It describes the current temperature With reference temperature The relative deviation between the temperature and the reaction temperature. Temperature is a key factor affecting the chemical reaction rate and reaction equilibrium, so temperature changes are often accompanied by significant fluctuations in the reactor state. In a packed bed continuous hydrogenation process unit, temperature changes may be caused by the exothermic or endothermic effects of the chemical reaction, or by the failure of the cooling system. By monitoring temperature deviations, the system can identify faults caused by these reasons, thereby preventing temperature fluctuations from adversely affecting the stable operation of the reactor. describes the current pressure With reference pressure The relative deviation between pressure and flow. Pressure change is an important indicator of the operating status of the reactor because it directly reflects the flow resistance and flow characteristics of the fluid in the catalyst bed. Significant deviations in pressure may mean problems such as abnormal changes in flow rate, blockage of the catalyst bed, or particle rupture. Therefore, real-time observation of pressure deviations can help the system identify these fluid flow-related faults in a timely manner. Finally, the fourth term in the formula The current catalyst activity Activity of reference catalyst The relative deviation between the catalyst activity and the product conversion rate. Catalyst activity is a key indicator that affects the reaction efficiency and product conversion rate, so catalyst deactivation directly affects the production efficiency of the reactor. Changes in catalyst activity may be caused by factors such as carbon deposition, pollution, and thermal deactivation, which will gradually reduce the effective reaction capacity of the catalyst. By monitoring the deviation of catalyst activity, the system can identify early signs of catalyst failure and take necessary remedial measures to restore the activity of the catalyst.

[0075] Embodiment 8: By using the following formula, the system fault deviation vector is decomposed into four different dimensions to obtain the system fault feature tensor:

[0076] F = ∑ k =1 K l k ⋅[ C k (Φ)⊗ T k (Φ)⊗ P k (Φ)⊗ a k (F)] ⋅exp(- o k D t ) ;

[0077] in, is the number of eigenvalues ​​of the system fault deviation vector; is the tensor corresponding to the concentration dimension; is the tensor corresponding to the temperature dimension; is the tensor corresponding to the pressure dimension; is the tensor corresponding to the catalyst activity dimension; For the The time decay coefficient corresponding to the eigenvalue; For the feature values.

[0078] Specifically, the left side of the formula Represents the fault feature tensor of the system. A tensor is a multidimensional data structure that can represent and process data in multiple dimensions. Unlike traditional two-dimensional matrices, tensors can more effectively express complex multidimensional relationships, so they are particularly suitable for fault diagnosis. By constructing a fault feature tensor, the system can identify the characteristic patterns of each key parameter in a multidimensional space, thereby more accurately locating and identifying faults. On the right side of the formula, describes the tensor decomposition of the system in four different dimensions (concentration, temperature, pressure, and catalyst activity). Here, Representative eigenvalues, which are used to measure the importance of each feature. Each eigenvalue corresponds to a set of multidimensional tensors that reflect the fault characteristics in different dimensions. In the tensor decomposition process, the outer product operation is used in the formula ( ), which means combining the tensors of each dimension to form an overall system fault feature tensor. Specifically, A tensor representing the concentration dimension, A tensor representing the temperature dimension, represents the tensor of the pressure dimension, and Then represents the tensor of the catalyst activity dimension. By combining these tensors by outer product, the system can describe the complex relationship between key parameters in multidimensional space. This multidimensional tensor decomposition method enables the system to comprehensively consider the characteristic information in multiple dimensions, thereby enhancing the comprehensiveness and robustness of fault diagnosis. In addition, the exponential decay term introduced in the formula It is used to describe the changes in the time dimension. Indicates The time decay coefficient of the eigenvalue is represents the change over time. By introducing this decay term, the formula can capture the dynamic change of fault features over time. This is very important for packed bed continuous hydrogenation process units, because the fault features in the reactor often accumulate and appear gradually over time. Through the exponential decay method, those fault modes that gradually decay or accumulate over time can be better described, thereby improving the timeliness and accuracy of fault diagnosis. The purpose of tensor decomposition is to reduce the dimensionality of originally complex high-dimensional data to extract the most important feature information. In this way, the system can deconstruct the features of multiple dimensions of the fault deviation vector into a set of independent tensors, and then perform feature identification in a simpler way. Compared with traditional matrix analysis methods, tensor decomposition can retain more multi-dimensional information and distinguish the significance of different fault features by the weight of the eigenvalues, so as to more accurately identify the main fault mode. In the fault diagnosis and automatic control system of packed bed continuous hydrogenation process units, the construction of fault feature tensors has significant advantages. It can not only comprehensively cover the four key dimensions of concentration, temperature, pressure and catalyst activity, but also capture the dynamic changes of fault features through time decay terms. This multi-dimensional feature extraction method enables the system to maintain efficient identification of various faults under complex operating conditions, thereby significantly improving the system's intelligence level and operational safety.

[0079] Example 9: Diagnosis and automation control part, calculate the abnormal state probability through the following formula :

[0080] ;

[0081] in, is the fault template tensor; is the total system failure rate set based on prior historical data; is the total system failure rate under the fault template tensor set according to the prior historical data; Modulo operation for vectors.

[0082] Specifically, in this formula represents the current system fault feature tensor, and Represents the preset fault template tensor. Fault Template Tensor It is a reference standard built based on the common failure modes in prior historical data. It represents a typical failure feature. This template is set based on the analysis and refinement of a large amount of historical failure data, and usually covers various known failure modes. , the system can compare the current fault characteristics with this template to evaluate the abnormality of the current state. middle, Calculate the current fault feature tensor With fault template The degree of deviation between them. The modulus operation here represents the magnitude or absolute value of the vector, specifically, it measures the degree of difference between the current state and the template. and The difference between the current state and the known failure mode is large, indicating that there is a significant deviation between the current state and the known failure mode. At this time, the magnitude of this deviation can quantify the degree of abnormality of the system. This means that in the known fault template This factor is used to weigh the reference value of the template. Assuming that the template The failure probability of the system The higher the value, the lower the non-fault probability. Therefore, this factor is introduced to comprehensively consider the credibility of the fault template. The denominator of the formula contains three products. The first It indicates the total failure rate of the system set according to prior historical data. This total failure rate reflects the overall probability of failure of the system under normal operation, and is usually calculated based on a large amount of historical data. This provides a basis for normalizing the formula so that the calculation results can better reflect the actual failure risk of the system. The relative deviation between the fault feature tensor and the fault template tensor is calculated. The relative deviation is calculated by dividing the fault template To achieve normalization, so that the deviations between different fault features can be compared on the same scale. The introduction of this term helps to improve the robustness of the formula, allowing the system to capture subtle fault changes more accurately. The third term in the denominator Indicates that in the fault template The probability of system failure under the condition of . This conditional probability reflects the possibility of system failure under the premise of a known template. By combining this conditional probability, the formula can dynamically adjust the probability calculation result of the abnormal state according to the matching degree between the current feature and the template.

[0083] Embodiment 10: The formula of the compensation controller is:

[0084] u ( t )= K c ( t )⋅[ e ( t )+ 1 t I ∫ 0 t e ( t ) dt + t D de ( t ) dt ]+ ∑ i =1 N c P( pro ) ⋅ C i (|Φ i |) ;

[0085] in, Control inputs generated for compensation controllers; To control the gain in real time; is the error signal; is the integration time constant; is the differential time constant; For components The gamma function value corresponding to the modulus of the corresponding fault deviation vector; is the integration variable.

[0086] Specifically, the first part of the formula The basic form of a proportional-integral-derivative (PID) controller is used. Specifically, Represents the error signal of the system, that is, the deviation between the current state and the target set value. The PID controller calculates the control input through the combined action of three parts: proportional, integral and differential. Proportional term It directly reflects the current error. Its function is to generate a control quantity proportional to the error to quickly correct the deviation. It is used to accumulate historical information of errors and eliminate the steady-state deviation of the system. Determines the strength of the integral effect. It reflects the rate of change of the error signal. By predicting the trend of the error, predictive control of the system can be achieved. Determines the sensitivity of the differential term. Unlike traditional PID controllers, this formula introduces a real-time control gain , to adapt to the dynamic needs of the system under different working conditions. The design of is to dynamically adjust the controller gain according to the current state or process conditions, thereby improving the system's adaptability. Through this real-time control gain adjustment, the compensation controller can better cope with complex process changes and enhance the system's robustness. The second part of the formula The compensation of fault deviation is introduced. This part uses the information of fault deviation vector to convert each component The deviation is calculated by the gamma function Specifically, Indicates components The modulus of the corresponding fault deviation vector, that is, the deviation between the current state and the normal state. Gamma function It is used to map the deviation into a compensation signal. The specific mapping relationship can be set according to different process requirements. For example, the gamma function can be linear, nonlinear or piecewise to adapt to different types of fault characteristics. In addition, the probability of abnormal state is also introduced in the compensation amount. By combining the abnormal state probability calculated previously , the system can dynamically adjust the compensation amount according to the risk level of the current fault. If the abnormal probability of the current state is high, the compensation controller will increase the compensation input to correct the fault more actively. This design enhances the intelligence level of the controller, allowing the system to flexibly respond to different fault conditions based on the state changes monitored in real time. In summary, the formula of the compensation controller combines the control gain adjusted in real time with the classic framework of the PID controller. and compensation input based on the fault deviation vector form a multi-level, dynamically adjusted control strategy. The proportional, integral, and differential parts provide the basic adjustment of the system error, while the fault deviation compensation enhances the system's adaptability in the face of complex working conditions and multiple fault modes. By comprehensively utilizing multiple information sources, the controller can not only accurately track the setpoint, but also adjust the control strategy in a timely manner according to changes in fault risks, thereby improving the stability and safety of the system.

[0087] Although the specific embodiments of the present invention are described above, it should be understood by those skilled in the art that these specific embodiments are only illustrative, and those skilled in the art may omit, replace, and change the details of the above methods and systems in various ways without departing from the principles and essence of the present invention. For example, merging the above method steps so as to perform substantially the same functions in substantially the same manner to achieve substantially the same results is within the scope of the present invention. Therefore, the scope of the present invention is limited only by the appended claims.

Claims

1. A fault diagnosis and automation control system for a packed bed continuous hydrogenation process device, characterized in that: The system comprises: a model building part, which is used to perform material balance modeling on the reactor to obtain a material balance model; according to the material balance model, the concentration distribution of each component is obtained to establish a reaction kinetics and heat balance coupling model; according to the reaction kinetics and heat balance coupling model, the temperature distribution is obtained, and the pressure field distribution model is established in combination with the concentration part of each component; according to the pressure field distribution model, the pressure distribution is obtained, and the concentration distribution and temperature distribution of each component are combined to establish a catalyst deactivation model to obtain the activity distribution of the catalyst; a state observation part, which is used to construct a fault feature vector according to the activity part of the catalyst, the pressure distribution, and the concentration distribution and temperature distribution of each component, and to perform deviation operation on the fault feature vector to obtain a fault deviation vector; and to perform tensor decomposition on the fault deviation vector to obtain a fault feature tensor; a diagnosis and automatic control part, which is used to calculate the probability of an abnormal state according to the fault feature tensor, generate a compensation controller according to the probability of the abnormal state, and use the compensation controller for automatic control; The system fault deviation vector is decomposed into four different dimensions by the following formula to obtain the system fault feature tensor: Where K is the number of eigenvalues ​​of the system fault deviation vector; C k (Φ) is the tensor corresponding to the concentration dimension; T k (Φ) is the tensor corresponding to the temperature dimension; P k (Φ) is the tensor corresponding to the pressure dimension; a k (Φ is the tensor corresponding to the catalyst activity dimension; o k is the time decay coefficient corresponding to the kth eigenvalue; k is the kth eigenvalue; In the diagnosis and automation control part, the abnormal state probability P(pro) is calculated by the following formula: Among them, F * is the fault template tensor; P(ov) is the total system failure rate set according to prior historical data; P(pro|F * ) is the total system failure rate under the fault template tensor set according to the prior historical data; |·| is the modular operation of the vector; The formula for the compensation controller is: Where u(t) is the control input generated by the compensation controller; K c (t) is the real-time control gain; e(t) is the error signal; τ I is the integration time constant; τ D is the differential time constant; Γ i (|Φ(i)|) is the gamma function value corresponding to the modulus of the fault deviation vector corresponding to component i; τ is the integral variable.

2. The fault diagnosis and automation control system of the packed bed continuous hydrogenation process device according to claim 1, characterized in that: The formula of the material balance model obtained in the model building part is: Among them, C i is the concentration distribution of component i; t is time; stands for Laplace operator; D eff,i is the effective diffusion coefficient of component i, which represents the spatial diffusion behavior of component i in the reactor. Since the packed bed reactor has particulate catalysts, the diffusion behavior is limited by the internal pores and external particle size, so the effective diffusion coefficient is used to characterize the actual diffusion situation; N r is the total number of reactions, indicating the number of all chemical reactions in the reactor; v ij is the stoichiometric coefficient, which indicates the amount of component i generated or consumed in reaction j; if v ij If it is a negative value, it means that component i is consumed in reaction j; if it is a positive value, it means that it is generated; k j is the rate constant of reaction j; represents the rate equation for reaction j, which is the product of the concentrations of all relevant components, N c is the number of all components, α mj is the reaction order of component m in reaction j; Describes the convection process of component i in the direction of fluid flow, u is the linear velocity, which represents the average velocity of the fluid in the reactor; z represents the axial coordinate of the reactor, that is, the coordinate inside the reactor along the direction of fluid flow.

3. The fault diagnosis and automatic control system of the packed bed continuous hydrogenation process device according to claim 2, characterized in that: The formula for the coupled model of reaction kinetics and heat balance is: Where ρ is the fluid density; C p is the specific heat capacity; T is the temperature distribution; λ eff represents the effective thermal conductivity, which reflects the thermal conductivity between the fluid and solid particles in the reactor; v s is the fluid flow rate; ΔH j represents the reaction heat of the jth reaction, a positive value indicates an endothermic reaction, and a negative value indicates an exothermic reaction; T w Indicates the temperature of the jacket; d t Represents the diameter of the reactor; the heat transfer coefficient U reflects the heat transfer effect between the reactor and the jacket.

4. The fault diagnosis and automatic control system of the packed bed continuous hydrogenation process device according to claim 3, characterized in that: The formula of the pressure field distribution model is: Where P is the pressure distribution; μ is the fluid viscosity; ∈ is the bed void fraction; d p is the catalyst particle size; K p is the pressure transmission coefficient; R g is the gas constant.

5. The fault diagnosis and automatic control system of the packed bed continuous hydrogenation process device according to claim 4, characterized in that: The formula for the catalyst deactivation model is: Where a is the catalyst activity concentration distribution; k d is the inactivation rate constant; E d is the deactivation activation energy; α i is the influence index of component i on inactivation; β P is the influence coefficient of pressure on inactivation; P ref is the reference pressure.

6. The fault diagnosis and automatic control system of the packed bed continuous hydrogenation process device according to claim 5, characterized in that: The formula of the fault feature vector constructed in the state observation part is: in, is the fault feature vector i corresponding to component i, with a total of N c Fault feature vector.

7. The fault diagnosis and automatic control system of the packed bed continuous hydrogenation process device according to claim 6, characterized in that: Perform deviation operation on the fault feature vector and obtain the formula of the fault deviation vector: Where Φ(i) is the fault deviation vector corresponding to component i; C i,ref is the reference concentration distribution corresponding to component i; T ref is the reference temperature distribution; P ref is the reference pressure distribution; a ref is the reference catalyst activity concentration distribution; ∥·∥2 is the L2 norm; the fault deviation vectors corresponding to all components are regarded as one element to form a vector as the system fault deviation vector.

Images