CN-121832249-B - Multi-component powder mixing uniformity evaluation method based on digital twin

CN121832249BCN 121832249 BCN121832249 BCN 121832249BCN-121832249-B

Abstract

The invention discloses a multi-component powder mixing uniformity evaluation method based on digital twinning, which relates to the technical field of powder mixing evaluation, and comprises the steps of establishing a virtual model for coupling computational fluid dynamics and discrete elements and synchronizing with monitoring data such as on-site flow, wind speed, particle count and speed, temperature and humidity, iteratively solving the forced movement of an airflow field and particles in a virtual space to obtain the residence time, circulation track and component concentration distribution of typical areas such as a spraying area, a fountain area and an annular space area, constructing an airflow non-uniformity index, particle residence time errors and alignment quality components, correcting parameters such as inlet speed, outlet pressure, airflow direction and screw speed by adopting feedback control, updating twinning parameters according to working conditions in stages, forming evaluation results based on the Raschig uniformity index and variance decomposition thereof, and outputting uniformity level, abnormal area positioning, key influence factors and traceable evaluation reports, and providing basis for subsequent process parameter setting and on-line adjustment.

Inventors

LI GUANGMING
QU LONG
ZHANG CHENGCHENG
LIU YONGLI
MA BEIYUE
BAN XIA
SONG NA
TIAN JIAN
GUO XINGKAI
WANG LIXIANG

Assignees

大石桥市美尔镁制品有限公司

Dates

Publication Date: 20260512
Application Date: 20260311

Claims (10)

1. The method for evaluating the mixing uniformity of the multi-component powder based on digital twinning is characterized by comprising the following steps of; Step S1, a digital twin computational fluid dynamics method and discrete element method coupling model is established, equipment geometric parameters and operation condition parameters are collected, boundary conditions and initial fields are set, a continuity equation and a momentum equation are solved, normal contact force and tangential contact force are calculated in a particle model, a particle translational equation and a rotation equation are combined, drag force is calculated according to a Gidaspow model, a simulation sequence of airflow speed and pressure and a simulation sequence of particle track and residence time are formed, and a speed sampling point sequence and a residence time partition sequence required by calculation are determined; Step S2, calculating an airflow distribution non-uniformity index UI and particle residence time statistics based on a simulation sequence, identifying a working area to be regulated according to UI and residence time distribution, setting an inlet airflow speed, an outlet pressure and an airflow direction according to the working area, synchronously setting a screw rotating speed and a particle supply amount, calculating a Lacey index for the updated simulation sequence, and recording; step S3, arranging a flow sensor, a wind speed sensor, a particle counter, a particle speed sensor and a temperature and humidity sensor, collecting an airflow speed sequence and a particle residence time sequence in real time, and fusing data; calculating an air flow uniformity error and a residence time uniformity error by a real-time sequence, inputting the errors into a proportional-integral-derivative controller to generate a control quantity, updating an inlet air flow speed, an outlet pressure, an air flow direction, a screw speed and a particle supply quantity according to the control quantity, calculating a Lacey index according to the period, and recording; And S4, converging long-term operation data, rolling and updating a digital twin model and an error sequence, constructing an iterative optimization flow based on feedback, adopting optimization calculation combining a reinforcement learning method and a model predictive control method, generating a control strategy according to a reward function and an objective function, and outputting each round of control parameters, an index sequence and a versioning record.
2. The method for evaluating the mixing uniformity of multi-component powder based on digital twinning according to claim 1, wherein a refined three-dimensional computing grid of the device is constructed based on the geometric parameters; converting the operation condition parameters into simulated initial conditions and material properties; Setting a turbulence model and a solving algorithm for multiphase flow in a CFD (computational fluid dynamics) solver, and setting particle size distribution, a shape factor and contact parameters in a DEM solver; the speed, volume fraction and interaction force data of the gas-solid phase are exchanged in each time step through the coupling interface.
3. The method for evaluating the mixing uniformity of the multi-component powder based on digital twinning according to claim 2, wherein generating the simulation data sequence of the airflow velocity field, the pressure field, the particle motion trail, the spatial distribution and the partition residence time specifically comprises: the complete mixing period is simulated through transient simulation, and the airflow velocity vector and the pressure scalar of each grid unit in the equipment, the position coordinate, the velocity vector and the belonging statistical partition number of each trace particle are recorded in a time sequence form; And generating a curve of the change of the particle quantity of each preset statistical zone along with time by utilizing particle track data post-processing, and then integrating to calculate the average residence time and the distribution thereof.
4. The method for evaluating the mixing uniformity of multi-component powder based on digital twinning according to claim 1, wherein the process of intelligently identifying the working area to be optimized comprises the following steps: Setting a reasonable interval of a threshold value of the UI and particle residence time; Clustering units with UI exceeding a threshold value in the simulation domain grid into an air flow non-uniform region, and marking a statistical partition with average particle residence time exceeding a reasonable interval as a particle movement abnormal region; And (3) combining the spatial position relation of the two types of areas, and determining one or more key working areas which finally need to be subjected to targeted adjustment by combining the structural characteristics of equipment.
5. The method for evaluating the mixing uniformity of a multi-component powder based on digital twinning according to claim 4, wherein the generated optimization strategy scheme further comprises: aiming at the flow field characteristics and particle motion modes of different key working areas, differential adjusting logic is designed; the strategy of enhancing directional airflow is adopted for the airflow dead zone, the strategy of adjusting the motion parameters of mechanical components or introducing auxiliary airflow is adopted for the particle detention zone, and all local regulation schemes are integrated into a unified control instruction set which is globally coordinated.
6. The method for evaluating the mixing uniformity of the multi-component powder based on the digital twin system according to claim 1, wherein the constructed real-time data acquisition network adopts a distributed architecture, and sensor data is transmitted to an edge computing gateway or a central server through an industrial internet of things protocol; The space-time alignment and data fusion processing comprises the steps of interpolating or resampling sensor data from different sampling rates to achieve time synchronization, and mapping and matching sensor space coordinates with virtual monitoring point coordinates in a digital twin model to achieve space alignment of physical data and virtual data.
7. The method for evaluating the mixing uniformity of the multi-component powder based on digital twinning according to claim 6, wherein the proportion, integral and differential coefficients of the PID controller with the self-adaptive gain adjustment function can be self-set on line according to the size and the change trend of error indexes and the current equipment operation stage; The adopted multivariable predictive controller takes a digital twin model as an internal predictive model, and based on the current measured value and the future set value, the optimal control sequence in a future period of time is optimally solved and rolling optimization is implemented.
8. The method for evaluating the mixing uniformity of multi-component powder based on digital twin according to claim 6, wherein when the Lacey index of the actual mixing process is periodically calculated on line, the sampling method comprises obtaining powder samples on line or intermittently from an outlet of a mixing device or a designated sampling port, and rapidly analyzing concentration distribution of each component by near infrared spectrum, image analysis or conductivity measurement technology, thereby calculating the Lacey index of the current batch.
9. The method for evaluating the mixing uniformity of a multi-component powder based on digital twinning according to claim 1, wherein the rolling updating of the digital twinning model comprises: Performing Bayesian updating or least square fitting calibration on key experience parameters in the model by using long-term operation data; meanwhile, according to equipment abrasion or transformation conditions, the geometric configuration in the model is synchronously updated, and the prediction accuracy of the model is kept.
10. The method for evaluating the mixing uniformity of the multi-component powder based on the digital twin system according to claim 1, wherein the state space of the reinforcement learning intelligent agent adopted by the constructed iterative optimization framework comprises real-time sensor data, current control parameters of equipment and historical KPI trend, the action space is an adjustable control parameter combination, and the reward function is the reinforcement learning reward function constructed according to the non-uniformity error of air flow, the stay time error of particles and the adjustment weight coefficient; the intelligent agent continuously optimizes the control strategy by interactive learning with the simulation environment constructed by the digital twin model and updated by historical data rolling.

Description

Multi-component powder mixing uniformity evaluation method based on digital twin Technical Field The invention relates to the technical field of powder mixing evaluation, in particular to a multi-component powder mixing uniformity evaluation method based on digital twin. Background In the existing powder mixing uniformity evaluation, an off-line sampling detection and statistical index calculation combined mode is generally adopted, or a single mechanism model is based on simulation analysis of a mixing process, in equipment such as a spouted bed, a double-screw mixer and the like, a method for analyzing gas-solid interaction, particle distribution and mixing states by combining gas flow field simulation and particle motion simulation is also adopted, and in addition, an on-site sensor is used for collecting gas flow, particles and environmental information for process monitoring. The prior art has the following defects: The offline sampling has hysteresis, dynamic change and regional difference in the mixing process are difficult to reflect, the independent simulation is often insufficient in coupling with field data, the model parameters are difficult to correct in time when the actual working conditions deviate, the key causes such as uneven air flow distribution, particle retention and residence time difference are lack of uniform quantization associated indexes and closed loop correction paths, the self-adaptive updating mechanism under the long-term running of different working conditions is lacking, and the stability and the reusability of an evaluation result are limited when the working conditions drift. The present invention proposes a solution to the above-mentioned problems. Disclosure of Invention In order to overcome the above-mentioned drawbacks of the prior art, embodiments of the present invention provide a method for evaluating mixing uniformity of multi-component powder based on digital twinning, so as to solve the problems set forth in the background art. In order to achieve the above purpose, the present invention provides the following technical solutions: the method for evaluating the mixing uniformity of the multi-component powder based on digital twin comprises the following steps of; Step S1, a digital twin computational fluid dynamics method and discrete element method coupling model is established, equipment geometric parameters and operation condition parameters are collected, boundary conditions and initial fields are set, a continuity equation and a momentum equation are solved, normal contact force and tangential contact force are calculated in a particle model, a particle translational equation and a rotation equation are combined, drag force is calculated according to a Gidaspow model, a simulation sequence of airflow speed and pressure and a simulation sequence of particle track and residence time are formed, and a speed sampling point sequence and a residence time partition sequence required by calculation are determined; Step S2, calculating an airflow distribution non-uniformity index UI and particle residence time statistics based on a simulation sequence, identifying a working area to be regulated according to UI and residence time distribution, setting an inlet airflow speed, an outlet pressure and an airflow direction according to the working area, synchronously setting a screw rotating speed and a particle supply amount, calculating a Lacey index for the updated simulation sequence, and recording; step S3, arranging a flow sensor, a wind speed sensor, a particle counter, a particle speed sensor and a temperature and humidity sensor, collecting an airflow speed sequence and a particle residence time sequence in real time, and fusing data; calculating an air flow uniformity error and a residence time uniformity error by a real-time sequence, inputting the errors into a proportional-integral-derivative controller to generate a control quantity, updating an inlet air flow speed, an outlet pressure, an air flow direction, a screw speed and a particle supply quantity according to the control quantity, calculating a Lacey index according to the period, and recording; And S4, converging long-term operation data, rolling and updating a digital twin model and an error sequence, constructing an iterative optimization flow based on feedback, adopting optimization calculation combining a reinforcement learning method and a model predictive control method, generating a control strategy according to a reward function and an objective function, and outputting each round of control parameters, an index sequence and a versioning record. In a preferred embodiment, step S1 comprises the following: constructing a refined three-dimensional computational grid of the device based on the geometric parameters; converting the operation condition parameters into simulated initial conditions and material properties; Setting a turbulence model and a solving algorithm for multiphase flow in a CF