CN-122015462-A - Digital twin-driven drying system and energy flow-quality cooperative control method

CN122015462ACN 122015462 ACN122015462 ACN 122015462ACN-122015462-A

Abstract

The invention discloses a digital twin-driven drying system and an energy flow-quality cooperative control method, belonging to the technical fields of heat pump drying, industrial Internet and intelligent manufacturing; the system comprises a carbon dioxide heat pump energy supply subsystem, an air circulation and heat recovery subsystem, a drying device and an intelligent control and digital management subsystem, and integrates a plurality of functional modules, wherein the system constructs a data driving and mechanism constraint collaborative state prediction model, realizes partition/layering energy flow priority distribution, multi-actuator joint optimization and disturbance self-adaptive working condition identification and mode switching based on a collaborative strategy of reinforcement learning and model prediction control, and hashes key operation and quality data to support formulation production and whole process tracing.

Inventors

LI MING
Yao Muchi
LI GUOLIANG
WANG YUNFENG
ZHANG YING
YU QIONGFEN
DENG JIANHUAN

Assignees

云南师范大学

Dates

Publication Date: 20260512
Application Date: 20260323

Claims (10)

1. The digital twin-driven drying system is characterized by comprising a carbon dioxide heat pump energy supply subsystem, an air circulation and heat recovery subsystem, a drying device and an intelligent control and digital management subsystem; the carbon dioxide heat pump energy supply subsystem comprises a compressor, an air cooler, a throttling element and an evaporator, and is used for providing heat for the drying device and forming heat exchange coupling with the air circulation and heat recovery subsystem; the air circulation and heat recovery subsystem comprises a return air channel, a fresh air channel, a moisture removal channel, a waste heat recoverer, a blower, a return air machine, a moisture removal blower and a layered air supply executing mechanism, and is used for carrying out waste heat recovery, condensation dehumidification on the return air and layered air supply to the drying device; The intelligent control and digital management subsystem is electrically connected with the carbon dioxide heat pump energy supply subsystem and the air circulation and heat recovery subsystem and is used for generating cooperative control instructions.
2. A digitally twinned driven drying system in accordance with claim 1 wherein: The intelligent control and digital management subsystem comprises a multi-source heterogeneous process data perception fusion module, a machine vision and water content prediction module, a PINN-transducer coupling digital twin model module, an energy flow-quality cooperative intelligent regulation module, a disturbance self-adaptive working condition identification and mode switching module, a dynamic attention mechanism module, a cloud edge cooperative calculation module and an operation data trusted storage module; The outputs of the PINN-transducer coupled digital twin model module comprise air state parameters, spatial distribution of material moisture content, unit moisture removal energy consumption, drying rate and quality risk indexes, and are used for providing feedforward prediction for the energy flow-quality collaborative intelligent regulation module; The dynamic attention mechanism module is used for carrying out joint modeling on the multi-layer/multi-region temperature and humidity field, the air volume distribution and the water content distribution of the materials to form space-time characteristics for digital twin prediction and control optimization; the cloud edge cooperative computing module comprises an edge controller and a cloud server, wherein the edge controller is deployed on a drying site, and the edge controller is used for executing data fusion, digital twin real-time reasoning and control output; the operation data trusted memory module is used for generating hash values for operation data, quality data and control decisions, writing the hash values into a alliance chain or a private chain account book according to time sequence, and checking and tracing; The intelligent control and digital management subsystem is in butt joint with the SCADA system, the MES system or the industrial Internet of things platform through OPC UA, modbus or MQTT protocols and is used for process recipe distribution, batch management and equipment state monitoring.
3. The digital twin-driven drying system of claim 2, wherein the multi-source heterogeneous process data sensing fusion module comprises a temperature sensor, a relative humidity sensor, a wind speed/wind volume sensor, a pressure sensor, an online water content sensor, an energy consumption parameter acquisition unit and a material weight acquisition unit, and is used for acquiring side operation state and material state data of a carbon dioxide heat pump energy supply subsystem and an air circulation and heat recovery subsystem.
4. The drying system driven by digital twin according to claim 3, wherein the machine vision and water content prediction module comprises an industrial camera and an edge reasoning unit, and the object detection and texture feature extraction model is adopted to identify apparent features of the material and output an on-line prediction result of water content and drying uniformity of the material.
5. The digital twin-driven drying system of claim 4, wherein the energy flow-quality collaborative intelligent regulation module generates a layered energy flow priority allocation scheme based on a reinforcement learning strategy, and performs constraint optimization on compressor frequency, throttling element opening, fan rotation speed, layered air supply valve and delivery beat in combination with model predictive control.
6. The digital twin-driven drying system of claim 5, wherein the disturbance adaptive condition identification and mode switching module constructs disturbance indicators based on environmental temperature and humidity fluctuations, material moisture content gradients and energy flow distribution offsets, adaptively switches between closed circulation, open dehumidification, deep condensation dehumidification and energy replenishment and frost/defrost scheduling modes, and imposes constraints on pressure change rates, exhaust temperature and heat supply fluctuations.
7. A method of cooperative control of energy flow and quality using a digitally twinned driven drying system according to any one of claims 1-6, comprising the steps of: S1, collecting working condition parameters of a carbon dioxide heat pump energy supply subsystem, temperature, humidity, air quantity parameters of multiple layers/regions in a drying device, energy consumption parameters and material state data, and cleaning and synchronizing the data; S2, performing target detection and texture feature extraction on the acquired image data to obtain apparent characteristics of the materials and outputting online prediction results of the water content and the drying uniformity of the materials; s3, carrying out space-time feature fusion on the multi-source data obtained in the S1 and the S2 by utilizing a dynamic attention mechanism to form a feature vector for twinning reasoning; S4, driving PINN-transducer coupled digital twin models based on heat and mass transfer and energy conservation constraint, and predicting air state parameters, spatial distribution of material moisture content, energy consumption for removing water per unit and quality risks; S5, constructing an energy consumption-humidity control-quality multi-objective optimization problem based on the predicted quantity obtained in the S4, screening key control variables through characteristic importance analysis, establishing a mapping relation among heat pump COP, material apparent characteristics and macroscopic drying rate, and dynamically adjusting multi-objective weights; s6, generating layered energy flow priority allocation and multi-executor cooperative control instructions based on cooperative strategies of reinforcement learning and model predictive control, and carrying out working condition identification and mode switching according to disturbance indexes; S7, performing trusted verification on the operation data, the quality data and the control decision and associating batch and formula information.
8. The energy flow-quality cooperative control method according to claim 7, wherein the specific content of the PINN-transducer coupled digital twin model based on heat transfer mass transfer and energy conservation constraint driving in S4 is that heat transfer mass transfer and air state constraint are embedded in a data driving network, and a material moisture content diffusion equation, a surface convection mass transfer boundary condition, a partition air energy conservation relationship and an air enthalpy-humidity relationship are used as physical constraint equations in a digital twin model training and updating process, so that high-precision coupling prediction of a temperature field, a humidity field, moisture content evolution and a system energy flow state in a drying process is realized; wherein, the material moisture content diffusion equation is as follows: ; In the formula, The water content of the material is; Is equivalent diffusion coefficient; The second derivative term of the space for the water content represents the diffusion and transmission of the water in the material; The surface convection mass transfer boundary condition has the expression: ; In the formula, The water content gradient is along the external normal direction of the surface of the material; the mass transfer coefficient is the surface convection; Is the moisture content of the material surface; To balance the water content; First, the The air energy conservation in the drying zone is expressed as follows: ; In the formula, Air mass flow for the j-th drying zone; The specific heat capacity is fixed for dry air; And The inlet air temperature and the outlet air temperature of the j-th drying zone respectively; heat exchange for the j-th drying zone; is the latent heat of vaporization of the moisture; Evaporating the moisture mass flow rate for the j-th drying zone; the air enthalpy-humidity relationship is expressed as follows: ; ; In the formula, Is the moisture content of the humid air; is the partial pressure of water vapor; Is the total pressure of the humid air; Is the specific enthalpy of wet air; The specific heat capacity is fixed for dry air; Is the air temperature; is the vaporization latent heat of water in a reference state at 0 ℃; The specific heat capacity is fixed for the water vapor; obtaining a prediction model through combining optimization data fitting loss and physical residual loss, and constructing a loss function, wherein the expression is as follows: ; In the formula, As a total loss function; Fitting a loss term to the data; Is a physical residual loss term; is a regularization loss term; loss of weight coefficients for physical residuals; Loss weight coefficients for regularization; Wherein, data fitting loss The expression is as follows: ; In the formula, Is the number of samples; a predicted value for the i-th output; an actual measurement value of the ith output quantity; Wherein, physical residual error is lost The expression is as follows: ; In the formula, The number of points is configured for physical configuration; a residual operator is physically constrained; predicted physical field variables for the j-th configuration point; the spatial coordinates of the j-th configuration point; time coordinates of the j-th configuration point; Wherein the smoothing regularization term The expression is as follows: ; In the formula, Is the kth trainable parameter in the model; Is the total number of trainable parameters; setting a discrete control period delta t, and carrying out online assimilation by a digital twin model in a state space mode, wherein the expression is as follows: ; ; In the formula, The system state vector is the t moment; The control quantity is t time; the disturbance quantity at the moment t; the vector is output for the system at the moment t; is a state transfer function; the mapping function is observed; the PINN-transducer coupled digital twin model adopts a physical residual error self-adaptive weighting strategy to adjust the weights of different physical process constraints on line, and maintains the prediction stability under the states of dewing, strong dehumidification or abrupt load change and the like.
9. The energy flow-quality cooperative control method according to claim 8, wherein the specific content of the dynamic adjustment of the multi-objective weight in S5 is as follows: The energy flow-quality cooperative intelligent regulation module fuses thermodynamic parameters of the heat pump, a temperature and humidity field of the drying oven and a material dryness gradient, and constructs an energy flow regulation model driven by deep learning; the expression of the energy flow priority of the constructed partition is as follows: ; In the formula, The energy flow priority index of the j-th drying area; The water content deviation or dryness gradient of the j-th drying area; the quality risk index of the j-th drying area; a heat supply requirement index for the j-th drying area; the water content deviation weight is given; Is the quality risk weight; J is the number of the drying partition; The objective function expression for model predictive control optimization is constructed as follows: ; In the formula, Predicting and controlling an objective function for a model, wherein N is the length of a prediction time domain, and k is a prediction step; Predicting and outputting for the t+k step; Q is an output error weighting matrix; Controlling increment for the t+k step, R is controlling increment weighting matrix, Punishment coefficients for the water cut variance; Predicting a variance term of the water content state quantity for the t+k step; Predicting a water content state quantity for the t+k step; the constraint conditions satisfied by the model predictive control optimization are as follows: and satisfies the safety interlock constraint; In the formula, And The lower limit and the upper limit of the control quantity are respectively; And The lower limit and the upper limit of the control increment are respectively; the expression for constructing the reinforcement learning reward function is as follows: In the formula, The function value is rewarded for the time t; The energy consumption of the system or the energy consumption for removing water per unit at the time t; Punishment weights for energy consumption; The average water content at the moment t; The water content is the target water content; the water content variance term of each drying area; a quality penalty term at the moment t; tracking weights for average water cut; Punishment weights for the water cut variance; The weights are penalized for quality.
10. The energy flow-quality cooperative control method according to claim 9, wherein the specific contents of the condition identification and the mode switching according to the disturbance index in S6 are as follows: When the disturbance index exceeds a preset threshold value, the humidity control weight and the quality weight are adaptively improved, and energy flows are preferentially distributed to the areas with high water content so as to improve the drying uniformity and the quality consistency; the mode switching comprises a closed circulation mode, an open dehumidification mode, a deep condensation dehumidification energy supplementing mode and a frost suppression/defrosting scheduling mode, and the pressure change rate, the exhaust temperature and the heat supply fluctuation are restrained in the mode switching process; The expression for constructing the disturbance index and the frosting risk index is as follows: ; In the formula, The disturbance index is the disturbance index at the moment t; the frosting risk index is the time t; is the inlet air temperature; Is inlet air relative humidity; the air quantity or the wind speed representation quantity is used for supplying air; is the surface temperature of the evaporator or the heat exchanger; Corresponding dew point temperature; Is an indication function; , , , the weight coefficient of each component of the disturbance index; accumulating weight coefficients for the frosting risks; 、 And Normalized reference amounts of temperature, relative humidity and air quantity are respectively; is the sampling interval; Accumulating a time window for frost formation; Is a disturbance index threshold; is a frosting risk threshold; is a duration threshold; When meeting the requirements > And duration exceeds When the system is switched to a high disturbance regulation mode, and when the system meets the requirement > And when the system is switched to a defrosting or defrosting scheduling mode.

Description

Digital twin-driven drying system and energy flow-quality cooperative control method Technical Field The invention belongs to the technical fields of heat pump drying, industrial Internet and intelligent manufacturing, and particularly relates to a digital twin-driven drying system and an energy flow-quality cooperative control method. Background Carbon dioxide heat pump drying is concerned by environmental protection working medium and low temperature adaptability, but under severe cold low temperature, large temperature rise heat supply working condition and temperature-humidity alternation or fluctuation environment, the state of a drying medium and a material water content field are in strong nonlinear change, and problems of energy flow distribution mismatch, insufficient cooperation of dehumidification and heat supply, lag regulation and control, fluctuation of product quality and the like are easy to occur. The control strategy of the existing drying equipment is mostly dependent on a fixed threshold regulation or single-loop feedback control mode, and lacks of real-time online modeling and accurate sensing capability for a multi-region temperature and humidity field, a material dryness gradient and a quality dynamic evolution process. Even if part of technologies try to optimize control logic and adopt PLC to realize quality feedback control of CO 2 heat pump drying, the technology is still limited to single closed cycle and single quality target, has no cooperative optimization design of energy flow and quality, and part of technologies only aims at discretization sectional control of rice grain design while considering energy efficiency and quality, has no accurate energy flow distribution model and cannot adapt to the characteristic requirements of heat sensitive materials. The method leads to the difficulty in realizing dynamic matching of energy supply and accurate humidity control according to the need when the prior art faces complex conditions such as environmental disturbance, load fluctuation, material anisotropy and the like, and further aggravates the control contradiction between energy efficiency improvement and quality assurance. Meanwhile, along with the deep penetration of the intelligent manufacturing technology, the requirements of a drying production line on standardized management of a formulation process, whole-flow batch tracing and remote operation and maintenance management and control are increasingly urgent, but the prior art still has obvious short plates. In the digital twin application level, the method is limited to unidirectional optimization in the research and development stage, has no real-time control capability of the production process, and lacks deep modeling and decision issuing functions of energy flow and quality parameters. In the heat pump adaptation and cooperative control layer, the temperature and humidity control of the focused air source heat pump is not related to the CO 2 transcritical circulation working condition, and the quality parameter-free closed loop design cannot meet the multi-target management and control requirement under the complex working condition. In summary, the traditional CO 2 heat pump drying system generally lacks a multisource data fusion sensing, digital twin reasoning decision and trusted data storage and verification mechanism, cannot open a full-link closed loop of 'energy flow efficiency-product quality-process tracing', and is difficult to adapt to intelligent management and control requirements of modern production. Based on the multiple bottlenecks in the prior art, it is highly desirable to provide an intelligent drying system and a control method for a digital twin-drive carbon dioxide heat pump, which take data drive and multi-physical-field coupling modeling as cores and can adapt to alternating disturbance of temperature and humidity to realize cooperation of energy flow optimal allocation and quality accurate regulation. Disclosure of Invention The invention aims to provide a digital twin-driven drying system and an energy flow-quality cooperative control method, which solve the problems that a traditional CO 2 heat pump drying system in the prior art generally lacks a multisource data fusion sensing, digital twin reasoning decision and trusted data storage and verification mechanism, a full-link closed loop of 'energy flow efficiency-product quality-process tracing' cannot be opened, intelligent management and control requirements of modern production are difficult to adapt, and the like. In order to achieve the aim, the invention provides a digital twin-driven drying system, which comprises a carbon dioxide heat pump energy supply subsystem, an air circulation and heat recovery subsystem, a drying device and an intelligent control and digital management subsystem; the carbon dioxide heat pump energy supply subsystem comprises a compressor, an air cooler, a throttling element and an evaporator, and is used for p