Heat exchanger full working condition running state simulation system based on sensing monitoring

Abstract

This invention relates to the field of artificial intelligence technology and discloses a full-condition operation simulation system for heat exchangers based on sensor monitoring. The system includes an environmental simulation module, a sensor monitoring module, a simulation processing module, an evaluation and optimization module, and a central control module. The environmental simulation module generates a set of full-condition scenarios covering startup, steady state, variable load, shutdown, and fault prediction. It introduces a multi-timescale coupling method and dynamic disturbance sequence to ensure that the simulation scenarios fully cover the actual operating environment and reduce the deviation between boundary conditions and actual operating conditions. A full-channel multi-physics dataset is constructed through a virtual sensor network. Combined with heterogeneous data alignment and dynamic mode decomposition algorithms, real-time monitoring data is mapped to a digital twin model to achieve dynamic reconstruction of irreversible thermodynamic processes and fluid transport characteristics, ensuring the consistency between simulation results and physical equipment states, and improving simulation accuracy and state restoration capabilities.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, specifically to a simulation system for the full-condition operation of heat exchangers based on sensor monitoring. Background Technology

[0002] A heat exchanger is a device that transfers part of the heat from a hot fluid to a cold fluid; heat exchangers play an important role in chemical, petroleum, power, food and many other industrial productions. In chemical production, heat exchangers can be used as heaters, coolers, condensers, evaporators and reboilers, etc., and are widely used.

[0003] Currently, due to the complex and varied operating scenarios of heat exchangers in industrial production such as chemical, petroleum, power, and food, existing heat exchanger operating state simulations are mostly limited to rated operating conditions or single parameter boundaries, making it difficult to dynamically coordinate multi-dimensional operating boundary conditions. This may result in incomplete state perception, delayed identification of abnormal signs, and insufficient accuracy in energy efficiency analysis and safety assessment.

[0004] Therefore, a sensor-based simulation system for the full-condition operation of heat exchangers is proposed to solve the above problems. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a full-condition operation simulation system for heat exchangers based on sensor monitoring, which solves the problems mentioned in the background technology, such as incomplete state perception and delayed identification of abnormal signs, leading to insufficient accuracy in energy efficiency analysis and safety assessment.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a heat exchanger full-condition operation simulation system based on sensor monitoring, the system comprising an environmental simulation module, a sensor monitoring module, a simulation processing module, an evaluation and optimization module, and a central control module; The environmental simulation module is used to construct multi-dimensional operating boundary conditions throughout the entire life cycle of the heat exchanger, and generate a set of full-condition scenarios covering startup, steady state, variable load, shutdown and fault simulation. The boundary conditions include thermodynamic parameters, fluid dynamic parameters and external disturbance parameters. The sensing and monitoring module is used to collect multi-physics response data of the heat exchanger under all operating conditions, obtain the spatiotemporal distribution characteristics of temperature field, pressure field, flow velocity field and stress field through virtual sensing network, and identify the evolution law of abnormal operating state. The simulation processing module is used to establish a digital twin model based on a multi-physics coupling mechanism, map real-time monitoring data to a virtual simulation space, and support the dynamic reconstruction of irreversible thermodynamic processes and fluid transport characteristics inside the heat exchanger under all operating conditions. The evaluation and optimization module is used to perform energy efficiency analysis and reliability assessment on the simulation results, identify heat transfer performance degradation nodes and weak links in safety margin, and generate multi-objective optimization strategies. The central control module is used to coordinate the runtime sequence and data interaction of each module, and dynamically adjusts the simulation accuracy and boundary conditions through an adaptive learning algorithm to form a closed-loop control logic.

[0007] Preferably, the environment simulation module includes a working condition generation unit, a boundary reconstruction unit, and a disturbance injection unit; The operating condition generation unit constructs a continuous variable load curve and a discrete fault mode library through parametric modeling, covering the complete operating range from rated operating conditions to extreme operating conditions. The boundary reconstruction unit dynamically corrects the heat exchanger inlet fluid parameters, ambient temperature and heat dissipation conditions based on historical operating data and external environmental factors, thereby reproducing the unsteady boundary characteristics in actual operation. The disturbance injection unit is used to simulate the progressive degradation process of sensor drift, fluid impurity deposition, and pipe wall scaling, generating a dynamic disturbance sequence with time evolution characteristics.

[0008] Preferably, the operating condition generation unit adopts a multi-timescale coupling method to superimpose long-term aging effects with short-term transient fluctuations to construct a composite operating condition scenario that includes seasonal climate change, production load fluctuations, and equipment start-up and shutdown impacts. The boundary reconstruction unit introduces external meteorological data and process flow parameters to establish a dynamic energy balance relationship between the heat exchanger and the upstream and downstream systems, and collaboratively simulates boundary conditions. The disturbance injection unit generates multi-source uncertainty disturbances through a stochastic process model to simulate the coupling mechanism between the cumulative effect of minor faults and sudden anomalies.

[0009] Preferably, the sensing and monitoring module includes a virtual sensor network, a data fusion unit, and a status recognition unit; The virtual sensing network generates high-resolution virtual monitoring data in sparse areas of physical sensor deployment through spatial interpolation and time series prediction methods, and constructs a multi-physics dataset covering the entire flow channel. The data fusion unit employs heterogeneous data alignment technology to eliminate spatiotemporal deviations between monitoring data with different sampling frequencies, dimensions, and precisions, and extracts feature vectors reflecting the internal state of the heat exchanger. The state recognition unit is based on a dynamic mode decomposition algorithm to extract dominant modes from multiphysics data and identify precursor features of abnormal states such as flow separation, local boiling, and heat transfer deterioration.

[0010] Preferably, the virtual sensing network combines physical mechanism constraints with a data-driven model to invert internal state variables in areas where key parameters such as pipe wall temperature and inlet / outlet pressure difference cannot be directly measured, through the fluid-thermal-structure coupling relationship. The data fusion unit introduces an uncertainty quantification method to separate random errors and systematic errors in the monitoring data and construct a state estimation result with a confidence interval. The state recognition unit uses nonlinear dimensionality reduction technology to map high-dimensional monitoring data to a low-dimensional feature space, and determines the degree to which the heat exchanger deviates from the design conditions through trajectory evolution analysis.

[0011] Preferably, the simulation processing module includes a multiphysics coupling unit, a model order reduction unit, and a digital twin mapping unit; The multiphysics coupling unit establishes a set of nonlinear partial differential equations based on conservation laws, including convection heat transfer, heat conduction, radiation and phase change, to describe the multi-scale energy transfer process inside the heat exchanger under all operating conditions. The model reduction unit adopts a combination of intrinsic orthogonal decomposition and dynamic mode decomposition to reduce the computational dimension while ensuring physical consistency. The digital twin mapping unit uses online parameter identification technology to synchronize the real-time status of the physical heat exchanger to the virtual model and dynamically corrects the model parameters to match the actual degradation characteristics.

[0012] Preferably, the multiphysics coupling unit introduces a flow-induced vibration and thermal stress coupling mechanism to simulate the feedback effect of pipe wall deformation caused by temperature gradient on flow resistance and heat transfer coefficient. The model reduction unit establishes a mapping relationship between the high-precision detailed model and the low-precision simplified model, and ensures the prediction accuracy of the reduced model in the entire working condition range through residual correction. The digital twin mapping unit uses a Bayesian update method to dynamically adjust the probability distribution of model parameters based on newly acquired monitoring data.

[0013] Preferably, the evaluation and optimization module includes an energy efficiency analysis unit, a safety margin evaluation unit, and a multi-objective optimization unit; The energy efficiency analysis unit quantifies the energy loss distribution under different operating conditions based on the analysis method, and identifies the main energy efficiency influencing factors such as heat exchange surface fouling thermal resistance, flow dead zone and bypass leakage. The safety margin assessment unit predicts the fatigue life of key components under alternating thermal loads and assesses the remaining safety reserve under the current operating condition through stress and life curve and damage tolerance analysis. The multi-objective optimization unit adopts the Pareto optimality method to generate the optimal combination of operating parameters with the objectives of improving energy efficiency, extending lifespan, and reducing maintenance costs, while satisfying safety constraints.

[0014] Preferably, the energy efficiency analysis unit combines the second law of thermodynamics with economic evaluation methods to calculate the unit heat exchange cost and carbon emission intensity under different load rates, and constructs a full-condition energy efficiency evaluation system. The safety margin assessment unit introduces a fuzzy comprehensive evaluation method to transform the deterministic analysis results into risk level indicators with probabilistic significance, thereby quantifying the impact of uncertain factors on safety assessment. The multi-objective optimization unit uses a combination of evolutionary algorithms and surrogate models to quickly search for the global optimal solution in a large-scale solution space, and determines the adjustment priority of key influencing parameters through sensitivity analysis.

[0015] Preferably, the central control module includes a scheduling and coordination unit, an adaptive learning unit, and a knowledge management unit; The scheduling and coordination unit dynamically allocates the running priority and data interaction bandwidth of each module according to the urgency of the simulation task and the computing resource usage. The adaptive learning unit uses reinforcement learning to adjust the boundary condition settings and simulation step size selection strategy of the environmental simulation based on feedback from actual operation results, thereby optimizing the overall response speed of the system. The knowledge management unit is used to store historical simulation cases, optimization strategies and evaluation conclusions, construct a searchable domain knowledge graph, and support solution matching and experience transfer under new working conditions.

[0016] Compared with existing technologies, this invention provides a simulation system for the full-condition operation of heat exchangers based on sensor monitoring, which has the following advantages: 1. In this invention, when constructing a full-condition operation simulation of a heat exchanger, a set of full-condition scenarios covering startup, steady state, variable load, shutdown, and fault pre-simulation is generated through an environmental simulation module. A multi-timescale coupling method and dynamic disturbance sequence are introduced to reproduce the multi-dimensional operating boundary conditions and non-steady-state evolution characteristics throughout the entire life cycle, ensuring that the simulation scenario fully covers the actual operating environment and reducing the deviation between boundary conditions and actual operating conditions.

[0017] 2. In this invention, when performing heat exchanger state perception and simulation reconstruction, a full-channel multi-physics dataset is constructed through a virtual sensor network. Combined with heterogeneous data alignment and dynamic mode decomposition algorithms, the precursor characteristics of abnormal states such as flow separation, local boiling, and heat transfer deterioration are identified. Real-time monitoring data is mapped to a digital twin model to achieve dynamic reconstruction of irreversible thermodynamic processes and fluid transport characteristics, ensuring the consistency between simulation results and physical equipment state, and improving simulation accuracy and state restoration capability.

[0018] 3. In this invention, when evaluating and optimizing the operation of the heat exchanger, the energy loss distribution is quantified by the energy efficiency analysis unit and fatigue life is predicted by the safety margin assessment unit. The Pareto optimal method is used to generate a multi-objective optimization strategy that takes into account energy efficiency improvement, life extension and maintenance cost reduction. At the same time, closed-loop control and experience transfer are realized through the adaptive learning and knowledge management of the central control module to ensure the safety and economy of the heat exchanger under all operating conditions and improve the overall response speed and decision reliability of the system. Attached Figure Description

[0019] Fig. 1 This is a schematic diagram of the simulation system for full-condition operation of heat exchangers based on sensor monitoring, as described in this invention. Fig. 2 This is a unit architecture diagram of the environment simulation module in this invention; Fig. 3 This is a flowchart of the operation steps of the heat exchanger full-condition operation simulation system based on sensor monitoring according to the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] For specific implementation examples, please refer to: Figs. 1-3 A simulation system for the full-condition operation of heat exchangers based on sensor monitoring. The system includes an environmental simulation module, a sensor monitoring module, a simulation processing module, an evaluation and optimization module, and a central control module. The environmental simulation module is used to construct multi-dimensional operating boundary conditions throughout the entire life cycle of the heat exchanger, generating a set of full-condition scenarios covering startup, steady state, variable load, shutdown, and fault simulation. The boundary conditions include thermodynamic parameters, fluid dynamic parameters, and external disturbance parameters. The sensing and monitoring module is used to collect multi-physics response data of the heat exchanger under all operating conditions. It obtains the spatiotemporal distribution characteristics of temperature field, pressure field, flow velocity field and stress field through virtual sensing network, and identifies the evolution law of abnormal operating state. The simulation processing module is used to establish a digital twin model based on the multi-physics coupling mechanism, map real-time monitoring data to the virtual simulation space, and support the dynamic reconstruction of the irreversible thermodynamic process and fluid transport characteristics inside the heat exchanger under all operating conditions. The evaluation and optimization module is used to perform energy efficiency analysis and reliability assessment on the simulation results, identify heat transfer performance degradation nodes and weak links in the safety margin, and generate multi-objective optimization strategies. The central control module coordinates the runtime sequence and data interaction of each module, and dynamically adjusts the simulation accuracy and boundary conditions through an adaptive learning algorithm to form a closed-loop control logic.

[0022] The environment simulation module includes a working condition generation unit, a boundary reconstruction unit, and a disturbance injection unit; The operating condition generation unit constructs a continuous variable load curve and a discrete fault mode library through parametric modeling, covering the complete operating range from rated operating conditions to extreme operating conditions. The continuously variable load curve describes the load rate of the heat exchanger as it changes over time, and its mathematical expression is: ; in The load factor varies with time t. For the first The amplitude coefficient of each harmonic component, For the first The angular frequency of each harmonic component For the first Phase angle of each harmonic component As a load factor constant bias, This represents the total number of harmonic components. The discrete fault mode library is established by enumeration and includes pipe wall scaling factor, fluid impurity blockage coefficient and sensor drift bias. Based on historical operating data and external environmental factors, the boundary reconstruction unit dynamically corrects the heat exchanger inlet fluid parameters, ambient temperature and heat dissipation conditions to reproduce the unsteady boundary characteristics in actual operation. The corrected formula is defined as follows: ; in For the first The boundary parameters after time correction, For the first Boundary parameters predicted by the time-matter model. For the first The Kalman gain matrix at each time step. For the first Real-time monitoring and observation values The observation matrix; The perturbation injection unit is used to simulate the progressive degradation process of sensor drift, fluid impurity deposition, and pipe wall scaling, generating a dynamic perturbation sequence with time evolution characteristics; The disturbance injection unit employs the Markov chain Monte Carlo method to simulate the progressive degradation process of sensor drift, fluid impurity deposition, and pipe wall scaling; for the scaling process, a time evolution characteristic function is established: ; in This indicates the rate of change of scaling thermal resistance over time. The scaling rate constant is defined as 1.0 × 10⁻⁶. -8 -5.0×10 -7 m 2 ·K / J, This represents the scaling reaction order, with a value ranging from 1.0 to 1.5. For heat exchange surface area, The temperature difference between the inside and outside of the pipe wall is represented by the differential equation. Solving this differential equation generates a dynamic perturbation sequence with time evolution characteristics.

[0023] The operating condition generation unit adopts a multi-timescale coupling method to superimpose long-term aging effects with short-term transient fluctuations, thereby constructing a composite operating condition scenario that includes seasonal climate change, production load fluctuations, and equipment start-up and shutdown impacts. Composite functions for complex operating conditions: ; in This is a function for complex working conditions. This is a function of long-term aging effects. It is a short-term transient fluctuation function; The boundary reconstruction unit introduces external meteorological data and process flow parameters to establish a dynamic energy balance relationship between the heat exchanger and the upstream and downstream systems, and collaboratively simulates boundary conditions. The energy balance equation is expressed as follows: ; in The rate of change of the total internal energy of the heat exchanger system. The heat power flowing into the system, The heat power flowing out of the system, This represents the heat power lost by the system to the environment. By solving this first-order ordinary differential equation, a collaborative simulation of cross-system boundary conditions is achieved, ensuring strict conservation of mass and energy during the simulation process. The disturbance injection unit generates multi-source uncertain disturbances through a stochastic process model to simulate the coupling mechanism between the cumulative effect of minor faults and sudden anomalies. The perturbation injection unit generates multi-source uncertainty perturbations through a Gaussian random walk stochastic process model; the probability density function of the perturbation term is defined as: ; in For random variables The probability density function, For multi-source uncertainties and disturbance variables, The mean of the disturbance. Let the standard deviation of the disturbance be . It is a natural constant. The value is pi. This distribution function is used to simulate the coupling mechanism between the cumulative effect of minor faults and sudden anomalies. The generated random disturbance signal is injected into the key performance parameters of the heat exchanger to improve the robustness of the simulation.

[0024] The sensing and monitoring module includes a virtual sensor network, a data fusion unit, and a status recognition unit; Virtual sensor networks generate high-resolution virtual monitoring data in sparse areas of physical sensor deployment using spatial interpolation and time series prediction methods, and construct a multi-physics dataset covering the entire flow channel. The spatial interpolation formula is: ; in Unknown location Virtual monitoring prediction values, For the first Weight coefficients for known positions, For known location The measured values ​​of the physical sensors Given the total number of known positions, The coordinates of the sensor position to be predicted. For the first The location coordinates of a known physical sensor; The data fusion unit uses heterogeneous data alignment technology to eliminate the spatiotemporal deviation between monitoring data with different sampling frequencies, dimensions and precisions, and extract feature vectors that reflect the internal state of the heat exchanger. The data fusion unit employs heterogeneous data alignment technology. For monitoring data with different sampling frequencies, dimensions, and precisions, a dynamic time warping algorithm is introduced to eliminate spatiotemporal biases. The aligned data undergoes principal component analysis to extract feature vectors reflecting the internal state of the heat exchanger. The covariance matrix calculated by principal component analysis is as follows: ; in Let covariance matrix be the variance matrix. For expectation operator, It is the mean vector. This is the original multiphysics monitoring data matrix. This is the transpose operator; by calculating the eigenvalues ​​and eigenvectors of the covariance matrix, the eigenvector corresponding to the largest eigenvalue is selected as the dominant feature, thereby achieving dimensionality reduction and feature extraction of high-dimensional data; The state recognition unit is based on the dynamic mode decomposition algorithm to extract the dominant mode from multi-physics data and identify the precursor features of abnormal states such as flow separation, local boiling and heat transfer deterioration. The core of the dynamic pattern decomposition algorithm lies in solving for the eigenvalues ​​and eigenvectors of the linear operator, satisfying... ,in For the first The system state vector at time t. For linear operators, For the first The system state vector at time t; the spectrum calculated using the dynamic mode decomposition algorithm. Extract the modal amplitude corresponding to a specific frequency, where It is a natural constant. The imaginary unit, For the first The system detects the angular frequency of each mode; when the amplitude of the mode of vortex shedding frequency in flow separation suddenly increases beyond the threshold, the system automatically identifies it as a precursor to abnormal states such as flow separation, local boiling, and heat transfer deterioration, and triggers the early warning mechanism; the modal amplitude threshold is the average modal amplitude under normal operating conditions plus 3 times the standard deviation as the threshold.

[0025] Virtual sensor networks combine physical mechanism constraints with data-driven models to invert internal state variables in regions where key parameters such as pipe wall temperature and inlet / outlet pressure difference cannot be directly measured, through fluid-thermal-structure coupling relationships. The data fusion unit introduces an uncertainty quantification method to separate random errors and systematic errors in the monitoring data and construct state estimation results with confidence intervals; The data fusion unit introduces an uncertainty quantification method, namely Bayesian linear regression. Random errors in monitoring data With systematic error Separate; Measured values The probability model is defined as follows: ,in The true values ​​of physical quantities are used; using Markov chain Monte Carlo sampling technique, state estimation results with confidence intervals are constructed. ,in This is the state estimate. The sample mean. The quantiles of the standard normal distribution The standard deviation is the sample standard deviation. The status recognition unit uses nonlinear dimensionality reduction technology to map high-dimensional monitoring data to a low-dimensional feature space, and judges the degree of deviation of the heat exchanger from the design conditions through trajectory evolution analysis. The objective function of nonlinear dimensionality reduction techniques is defined as: ; in The objective function value, High-dimensional distribution With low-dimensional distribution Kullback-Leibler divergence, Midpoint of low-dimensional space With point The joint probability, For the first The true conditional probability distribution of each sample; by calculating the difference in probability distribution of data points in high-dimensional and low-dimensional spaces, complex operating states are mapped to trajectories in low-dimensional space; Trajectory evolution analysis, which calculates the distance between the trajectory and the cluster center under normal operating conditions, determines the degree to which the heat exchanger deviates from the design operating conditions. The greater the distance, the more severe the deviation. ; in This represents the deviation between the current state and normal operating conditions. This is the current running state vector. This represents the cluster center vector under normal operating conditions.

[0026] The simulation processing module includes a multiphysics coupling unit, a model order reduction unit, and a digital twin mapping unit; The multiphysics coupling unit establishes a set of nonlinear partial differential equations based on conservation laws, which includes convection heat transfer, heat conduction, radiation and phase change, to describe the multi-scale energy transfer process inside the heat exchanger under all operating conditions. Energy conservation governing equations: ; in For fluid density, The specific heat capacity at constant pressure of the fluid. For temperature Regarding time The partial derivatives, For flow velocity vectors, For temperature gradient, For thermal conductivity, For internal heat source, For radiative heat transfer, The phase transformation heat term is used; the above equations are spatially discretized and time-integrated using the finite element method to describe the multi-scale energy transfer process inside the heat exchanger under all operating conditions. The model reduction unit adopts a combination of intrinsic orthogonal decomposition and dynamic mode decomposition to reduce the computational dimension while ensuring physical consistency. Intrinsic orthogonal decomposition obtains basis functions by performing singular value decomposition on high-fidelity simulation data, and then projects the original high-dimensional state variables onto a low-dimensional subspace to obtain a reduced-order model. ; in Low-dimensional state vector Time derivative, It is a low-dimensional state vector. For the system input vector, and This is the reduced-order system matrix; this reduced-order model lowers the computational dimensionality, achieving a balance between real-time simulation and long-term prediction. The digital twin mapping unit uses online parameter identification technology to synchronize the real-time status of the physical heat exchanger to the virtual model and dynamically corrects the model parameters to match the actual degradation characteristics. The objective function for parameter correction is: ; in For the model parameter vector Minimize operation, This is the vector of model parameters to be corrected. For the length of the data window, For the first The time-matter model in the model parameter vector The output below, For the first The actual measured value at that moment.

[0027] The multiphysics coupling unit introduces the coupling mechanism of flow-induced vibration and thermal stress to simulate the feedback effect of pipe wall deformation caused by temperature gradient on flow resistance and heat transfer coefficient. The relationship between flow-induced vibration displacement and flow velocity is described by random vibration theory, satisfying the following conditions: ;in For flow-induced vibration displacement The second derivative, For flow-induced vibration displacement The first derivative, For the damping ratio, For the natural frequency, Fluid excitation force, For the mass of the vibration system; The thermal stress in the pipe wall is calculated using the thermoelastic equation: ; in For the thermal stress of the pipe wall, For elastic modulus, The coefficient of thermal expansion is The temperature difference between the inside and outside of the pipe wall; when the thermal stress exceeds the yield strength of the material, the pipe wall undergoes plastic deformation, which in turn changes the geometry of the flow channel; The model reduction unit establishes a mapping relationship between the high-precision detailed model and the low-precision simplified model, and ensures the prediction accuracy of the reduced model in the entire working condition range through residual correction; Define the residual correction term for the reduced-order model: ; in This is the residual correction term. For high-fidelity simulation output, This is the output of the reduced-order model; a residual neural network is introduced. Modeling the residuals, the output of the corrected reduced-order model is: ; in This is the output of the corrected reduced-order model. The residual is the estimated value. Through the above residual correction mechanism, the prediction accuracy of the reduced-order model is guaranteed to be higher than the preset threshold in the entire working condition range. The digital twin mapping unit uses a Bayesian update method to dynamically adjust the probability distribution of model parameters based on newly acquired monitoring data; Prior distribution of parameters The likelihood function is determined using historical data. Based on current monitoring data Construction; according to Bayes' theorem, the posterior distribution Calculated as ; The Markov chain Monte Carlo sampling technique is used to generate samples from the posterior distribution and dynamically obtain the optimal estimates of the model parameters and their uncertainty range.

[0028] The evaluation and optimization module includes an energy efficiency analysis unit, a safety margin assessment unit, and a multi-objective optimization unit; The energy efficiency analysis unit quantifies the energy loss distribution under different operating conditions based on the analysis method, and identifies the main energy efficiency influencing factors such as heat exchange surface fouling thermal resistance, flow dead zone and bypass leakage. Based on the tamper analysis method, an evaluation model of the second law of thermodynamics under all operating conditions is established; by calculating the tamper distribution under different operating conditions, the degree of energy level matching of hot and cold fluids is quantified, and the weak link with the greatest irreversible loss is identified. ; in The loss per unit time. For ambient temperature, For entropy production rate, The mass flow rate of the fluid. Specific enthalpy of the fluid inlet. Enthalpy of fluid outlet The specific entropy at the fluid inlet. The specific entropy at the fluid outlet; The safety margin assessment unit predicts the fatigue life of key components under alternating thermal loads and assesses the remaining safety reserves under the current operating conditions through stress and life curves and damage tolerance analysis. Based on the full-condition thermal stress time-domain data output by the simulation processing module, the complex variable amplitude stress history is simplified into an equivalent constant amplitude stress cycle by using the rainflow counting method; combined with the stress and life curves of the heat exchanger tubes and Miner's linear cumulative damage theory, the cumulative fatigue damage factor of key components under alternating heat load is calculated. ; in To accumulate fatigue damage factors, This represents the actual number of loops. For stress amplitude, Stress ratio, The theoretical number of iterations, The stress level is the grade number. This represents the actual number of loops. This is the critical damage threshold, typically set to 1.0; The multi-objective optimization unit adopts the Pareto optimal method to generate the optimal combination of operating parameters with the objectives of improving energy efficiency, extending lifespan, and reducing maintenance costs, while satisfying safety constraints. The energy efficiency output of the energy efficiency analysis unit is used as the benefit objective, and the cumulative damage rate and maintenance cost are used as the cost objectives. A multi-objective optimization problem is constructed to find the Pareto optimal frontier under the premise of ensuring safety constraints. A set of equilibrium solutions that cannot improve one objective without harming another is identified for decision-makers to choose from.

[0029] The energy efficiency analysis unit combines the second law of thermodynamics with economic evaluation methods to calculate the unit heat exchange cost and carbon emission intensity under different load rates, and constructs a full-condition energy efficiency evaluation system. Unit heat exchange cost: ; in Cost per unit of heat exchange For initial investment costs, This is the capital recovery coefficient. Annual maintenance costs Total annual heat exchange; Carbon emission intensity: ; in Carbon emission intensity per unit of heat exchange, For the first Energy consumption For the first Carbon emission factors of various energy sources; The safety margin assessment unit introduces the fuzzy comprehensive evaluation method to transform the deterministic analysis results into risk level indicators with probabilistic significance, thereby quantifying the impact of uncertain factors on safety assessment. Since traditional deterministic assessments cannot fully reflect the impact of uncertainties such as material discreteness and manufacturing errors, fuzzy mathematics theory needs to be introduced. High-risk, medium-risk, and low-risk rating sets are defined, and membership functions, i.e. Gaussian membership functions, are constructed for deterministic assessment results such as stress, temperature, and corrosion rate. Through fuzzy synthesis operations, deterministic physical quantities are transformed into risk level indicators with probabilistic significance, thereby quantifying the impact of uncertainties on safety assessments. The multi-objective optimization unit uses a combination of evolutionary algorithms and surrogate models to quickly search for the global optimum in a large-scale solution space and determines the adjustment priority of key influencing parameters through sensitivity analysis. First, initial sample points are generated in the design space using Latin hypercube sampling. The objective function values ​​of these sample points are then evaluated using a high-precision simulation model, and a surrogate model is trained. The approximate expression of the surrogate model is: ; in This is an approximation of the objective function. As weight, As basis functions, For the input variable vector, For the first The input vector of each training sample, The total number of training samples; Evolutionary algorithms are used to optimize the surrogate model, and the variance contribution rate of each input parameter is calculated through sensitivity analysis to determine the adjustment priority of key influencing parameters, guiding parameter fine-tuning in actual operation. ; in For the first The variance contribution rate of each input parameter and , The variance contribution of the i-th parameter. This represents the total variance.

[0030] The central control module includes a scheduling and coordination unit, an adaptive learning unit, and a knowledge management unit; The scheduling and coordination unit dynamically allocates the running priority and data interaction bandwidth of each module according to the urgency of the simulation task and the computing resource usage. The priority allocation strategy uses a weighted scoring method: ; in For the first Priority rating for each task. , Let be the weighting coefficients, and , Assigning weights based on the urgency of the task. For the first Resource utilization rate of each task; The adaptive learning unit uses reinforcement learning to adjust the boundary condition settings and simulation step size selection strategy of the environmental simulation based on feedback from actual operation results, thereby optimizing the overall response speed of the system. The simulation system is treated as an intelligent agent, and the actual operating deviation of the heat exchanger is taken as the environmental state. A reward function is defined, which comprehensively considers simulation accuracy, computation time and resource consumption. The intelligent agent is trained through trial and error using a deep reinforcement learning algorithm to learn the optimal strategy: that is, in different operating conditions, it automatically adjusts the fineness of the boundary conditions and the simulation step size of the environmental simulation. Reinforcement learning reward function: ; in In order to perform the action Post-state Instant rewards received This represents the improvement in simulation accuracy relative to the threshold. This is the reduction amount used in simulation calculations. To calculate resource consumption , , To balance the weighting coefficients of various indicators and ; The knowledge management unit is used to store historical simulation cases, optimization strategies and evaluation conclusions, build a searchable domain knowledge graph, and support solution matching and experience transfer under new working conditions.

[0031] The operation steps of the sensor-based full-condition operation simulation system for heat exchangers are as follows: Step 1: Construction of Full-Condition Scenarios and Generation of Boundary Conditions First, the environmental simulation module constructs multi-dimensional operating boundary conditions throughout the entire life cycle of the heat exchanger. The operating condition generation unit uses a multi-timescale coupling method to superimpose long-term aging effects and short-term transient fluctuations to generate a set of full operating condition scenarios covering startup, steady state, variable load, shutdown, and fault simulation. At the same time, the boundary reconstruction unit introduces external meteorological data and process flow-related parameters to dynamically correct inlet fluid parameters and environmental heat dissipation conditions. The disturbance injection unit simulates sensor drift and scaling degradation processes through a stochastic process model, jointly reproducing the unsteady boundary characteristics and dynamic disturbance sequences in actual operation.

[0032] Step 2: Multiphysics Data Sensing and State Recognition The system collects multi-physics response data of the heat exchanger under all operating conditions using a sensor monitoring module. It then uses a virtual sensor network to generate high-resolution virtual monitoring data in sparse areas of physical sensors through spatial interpolation and time series prediction methods, constructing a dataset of temperature, pressure, velocity, and stress fields covering the entire flow channel. Subsequently, the data fusion unit uses heterogeneous data alignment technology and uncertainty quantification methods to eliminate spatiotemporal biases. The state recognition unit extracts the dominant modes and identifies the precursor features of abnormal states such as flow separation, local boiling, and heat transfer deterioration based on dynamic mode decomposition algorithm and nonlinear dimensionality reduction technology.

[0033] Step 3: Digital Twin Model Mapping and Dynamic Reconstruction A digital twin model is established based on a multiphysics coupling mechanism through a simulation processing module. The multiphysics coupling unit establishes a set of nonlinear partial differential equations including convection heat transfer, heat conduction, radiation, and phase change based on conservation laws. The coupling mechanism of flow-induced vibration and thermal stress is introduced to describe the multi-scale energy transfer process. The model reduction unit adopts a combination of intrinsic orthogonal decomposition and dynamic mode decomposition to reduce the computational dimension while ensuring physical consistency. The digital twin mapping unit synchronizes real-time monitoring data to the virtual model and dynamically corrects parameters to match the actual degradation characteristics through online parameter identification and Bayesian update methods.

[0034] Step 4: Energy efficiency and safety assessment and multi-objective optimization The evaluation and optimization module is used to conduct energy efficiency analysis and reliability assessment on the simulation results. The energy efficiency analysis unit combines the second law of thermodynamics and economic evaluation methods to quantify the energy loss distribution under different operating conditions and identify energy efficiency influencing factors such as fouling thermal resistance and flow dead zone. The safety margin assessment unit predicts the fatigue life of key components and quantifies the risk level through stress life curves and fuzzy comprehensive evaluation methods. Finally, the multi-objective optimization unit uses the Pareto optimal method and evolutionary algorithm to generate the optimal combination of operating parameters that takes into account energy efficiency improvement, life extension and maintenance cost reduction under the premise of meeting safety constraints.

[0035] Step 5: Closed-loop control and knowledge collaborative management The central control module coordinates the runtime sequence and data interaction of each module. The scheduling and coordination unit dynamically allocates running priorities and data interaction bandwidth based on the urgency of the task and the usage of computing resources. The adaptive learning unit adjusts the boundary condition settings and simulation step size selection strategy of the environment simulation based on the feedback of the actual running effect through reinforcement learning methods. The knowledge management unit stores historical simulation cases and optimization strategies to build a domain knowledge graph, which supports scheme matching and experience transfer under new working conditions, forming an adaptive closed-loop control logic.

[0036] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0037] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A simulation system for the full-condition operation of a heat exchanger based on sensor monitoring, characterized in that: The system includes an environmental simulation module, a sensing and monitoring module, a simulation processing module, an evaluation and optimization module, and a central control module. The environmental simulation module is used to construct multi-dimensional operating boundary conditions throughout the entire life cycle of the heat exchanger, and generate a set of full-condition scenarios covering startup, steady state, variable load, shutdown and fault simulation. The boundary conditions include thermodynamic parameters, fluid dynamic parameters and external disturbance parameters. The sensing and monitoring module is used to collect multi-physics response data of the heat exchanger under all operating conditions, obtain the spatiotemporal distribution characteristics of temperature field, pressure field, flow velocity field and stress field through virtual sensing network, and identify the evolution law of abnormal operating state. The simulation processing module is used to establish a digital twin model based on a multi-physics coupling mechanism, map real-time monitoring data to a virtual simulation space, and support the dynamic reconstruction of irreversible thermodynamic processes and fluid transport characteristics inside the heat exchanger under all operating conditions. The evaluation and optimization module is used to perform energy efficiency analysis and reliability assessment on the simulation results, identify heat transfer performance degradation nodes and weak links in safety margin, and generate multi-objective optimization strategies. The central control module is used to coordinate the runtime sequence and data interaction of each module, and dynamically adjusts the simulation accuracy and boundary conditions through an adaptive learning algorithm to form a closed-loop control logic.

2. The heat exchanger full-condition operation simulation system based on sensor monitoring according to claim 1, characterized in that: The environment simulation module includes a working condition generation unit, a boundary reconstruction unit, and a disturbance injection unit; The operating condition generation unit constructs a continuous variable load curve and a discrete fault mode library through parametric modeling, covering the complete operating range from rated operating conditions to extreme operating conditions. The boundary reconstruction unit dynamically corrects the heat exchanger inlet fluid parameters, ambient temperature and heat dissipation conditions based on historical operating data and external environmental factors, thereby reproducing the unsteady boundary characteristics in actual operation. The disturbance injection unit is used to simulate the progressive degradation process of sensor drift, fluid impurity deposition, and pipe wall scaling, generating a dynamic disturbance sequence with time evolution characteristics.

3. The heat exchanger full-condition operation simulation system based on sensor monitoring according to claim 2, characterized in that: The operating condition generation unit adopts a multi-timescale coupling method to superimpose long-term aging effects with short-term transient fluctuations to construct a composite operating condition scenario that includes seasonal climate change, production load fluctuations, and equipment start-up and shutdown impacts. The boundary reconstruction unit introduces external meteorological data and process flow parameters to establish a dynamic energy balance relationship between the heat exchanger and the upstream and downstream systems, and collaboratively simulates boundary conditions. The disturbance injection unit generates multi-source uncertainty disturbances through a stochastic process model to simulate the coupling mechanism between the cumulative effect of minor faults and sudden anomalies.

4. The heat exchanger full-condition operation simulation system based on sensor monitoring according to claim 1, characterized in that: The sensing and monitoring module includes a virtual sensor network, a data fusion unit, and a status recognition unit. The virtual sensing network generates high-resolution virtual monitoring data in sparse areas of physical sensor deployment through spatial interpolation and time series prediction methods, and constructs a multi-physics dataset covering the entire flow channel. The data fusion unit employs heterogeneous data alignment technology to eliminate spatiotemporal deviations between monitoring data with different sampling frequencies, dimensions, and precisions, and extracts feature vectors reflecting the internal state of the heat exchanger. The state recognition unit is based on a dynamic mode decomposition algorithm to extract dominant modes from multiphysics data and identify precursor features of abnormal states such as flow separation, local boiling, and heat transfer deterioration.

5. The heat exchanger full-condition operation simulation system based on sensor monitoring according to claim 4, characterized in that: The virtual sensing network combines physical mechanism constraints with a data-driven model to invert internal state variables in regions where key parameters such as pipe wall temperature and inlet / outlet pressure difference cannot be directly measured, through the fluid-thermal-structure coupling relationship. The data fusion unit introduces an uncertainty quantification method to separate random errors and systematic errors in the monitoring data and construct a state estimation result with a confidence interval. The state recognition unit uses nonlinear dimensionality reduction technology to map high-dimensional monitoring data to a low-dimensional feature space, and determines the degree to which the heat exchanger deviates from the design conditions through trajectory evolution analysis.

6. The heat exchanger full-condition operation simulation system based on sensor monitoring according to claim 1, characterized in that: The simulation processing module includes a multiphysics coupling unit, a model order reduction unit, and a digital twin mapping unit; The multiphysics coupling unit establishes a set of nonlinear partial differential equations based on conservation laws, including convection heat transfer, heat conduction, radiation and phase change, to describe the multi-scale energy transfer process inside the heat exchanger under all operating conditions. The model reduction unit adopts a combination of intrinsic orthogonal decomposition and dynamic mode decomposition to reduce the computational dimension while ensuring physical consistency. The digital twin mapping unit uses online parameter identification technology to synchronize the real-time status of the physical heat exchanger to the virtual model and dynamically corrects the model parameters to match the actual degradation characteristics.

7. The heat exchanger full-condition operation simulation system based on sensor monitoring according to claim 6, characterized in that: The multiphysics coupling unit introduces a flow-induced vibration and thermal stress coupling mechanism to simulate the feedback effect of pipe wall deformation caused by temperature gradient on flow resistance and heat transfer coefficient. The model reduction unit establishes a mapping relationship between the high-precision detailed model and the low-precision simplified model, and ensures the prediction accuracy of the reduced model in the entire working condition range through residual correction. The digital twin mapping unit uses a Bayesian update method to dynamically adjust the probability distribution of model parameters based on newly acquired monitoring data.

8. The heat exchanger full-condition operation simulation system based on sensor monitoring according to claim 1, characterized in that: The evaluation and optimization module includes an energy efficiency analysis unit, a safety margin evaluation unit, and a multi-objective optimization unit; The energy efficiency analysis unit quantifies the energy loss distribution under different operating conditions based on the analysis method, and identifies the main energy efficiency influencing factors such as heat exchange surface fouling thermal resistance, flow dead zone and bypass leakage. The safety margin assessment unit predicts the fatigue life of key components under alternating thermal loads and assesses the remaining safety reserve under the current operating condition through stress and life curve and damage tolerance analysis. The multi-objective optimization unit adopts the Pareto optimality method to generate the optimal combination of operating parameters with the objectives of improving energy efficiency, extending lifespan, and reducing maintenance costs, while satisfying safety constraints.

9. The heat exchanger full-condition operation simulation system based on sensor monitoring according to claim 8, characterized in that: The energy efficiency analysis unit combines the second law of thermodynamics with economic evaluation methods to calculate the unit heat exchange cost and carbon emission intensity under different load rates, and constructs a full-condition energy efficiency evaluation system. The safety margin assessment unit introduces a fuzzy comprehensive evaluation method to transform the deterministic analysis results into risk level indicators with probabilistic significance, thereby quantifying the impact of uncertain factors on safety assessment. The multi-objective optimization unit uses a combination of evolutionary algorithms and surrogate models to quickly search for the global optimal solution in a large-scale solution space, and determines the adjustment priority of key influencing parameters through sensitivity analysis.

10. The heat exchanger full-condition operation simulation system based on sensor monitoring according to claim 1, characterized in that: The central control module includes a scheduling and coordination unit, an adaptive learning unit, and a knowledge management unit. The scheduling and coordination unit dynamically allocates the running priority and data interaction bandwidth of each module according to the urgency of the simulation task and the computing resource usage. The adaptive learning unit uses reinforcement learning to adjust the boundary condition settings and simulation step size selection strategy of the environmental simulation based on feedback from actual operation results, thereby optimizing the overall response speed of the system. The knowledge management unit is used to store historical simulation cases, optimization strategies and evaluation conclusions, construct a searchable domain knowledge graph, and support solution matching and experience transfer under new working conditions.

Images