CN-121797893-B - Intelligent material proportion self-adaptive regulation and control system based on sand mixer

CN121797893BCN 121797893 BCN121797893 BCN 121797893BCN-121797893-B

Abstract

The invention discloses an intelligent material proportion self-adaptive regulation and control system based on a sand mixer, which belongs to the technical field of sand mixing regulation and control and comprises a high-precision material characterization unit, a multi-mode process sensing unit, a global knowledge base and a core processor, wherein the high-precision material characterization unit is used for generating a dynamic material characteristic vector, the multi-mode process sensing unit is used for synchronously collecting a torque vibration signal, a visual image sequence and a chemical spectrum signal in a mixing process, the global knowledge base is used for storing historical production case data, an optimization strategy and model parameters, and the core processor comprises a multi-mode digital twin body, an autonomous decision engine, an abnormal processor and an instruction verification and smoother. According to the invention, the digital twin body is constructed by fusing multisource perception data, the visualization and the accurate prediction of the performance of the mixing process are realized, the mixing parameters and the material proportion are dynamically and cooperatively adjusted based on a rolling optimization and autonomous learning mechanism, and finally, the stable performance of molding sand, the energy consumption optimization and the self-adaptive toughness control on abnormal working conditions are realized.

Inventors

ZHANG ZHIFANG
GUO JINGWEN
LIU WENBAO
TAO YANG
WANG LIAN

Assignees

山东工将机械科技股份有限公司

Dates

Publication Date: 20260508
Application Date: 20260306

Claims (8)

1. Intelligent material proportion self-adaptive regulation and control system based on sand mixer, its characterized in that includes: the high-precision material characterization unit is used for detecting the particle size distribution, the true density, the water content and the surface activity index of each batch of raw materials on line and generating a dynamic material characteristic vector; the multi-mode process sensing unit comprises a multi-axis torque sensor arranged on a main shaft of the sand mixer, an internal high-definition camera array and an on-line near infrared spectrometer at a discharge port, and is used for synchronously acquiring a torque vibration signal, a visual image sequence and a chemical spectrum signal of the mixing process; The global knowledge base is used for storing historical production case data, optimization strategies and model parameters; the core processor is respectively connected with the high-precision material characterization unit, the multi-mode process sensing unit and the global knowledge base, and comprises: the multi-mode digital twin body is used for receiving real-time sensing data and material characteristic vectors, and simulating and predicting performance indexes of the molding sand at the mixing endpoint through a built-in physical property emergence prediction model; The autonomous decision engine is internally provided with a dynamic optimization objective function, and calculates a stirring parameter adjustment instruction and a material proportioning correction amount through a rolling time domain optimization algorithm based on the prediction output of the multi-modal digital twin body; The abnormal processor is used for identifying abnormal working conditions according to abrupt change of the sensing signals and triggering corresponding toughness control programs; the command checking and smoothing device is used for carrying out feasibility checking and smoothing filtering on the proportioning correction quantity output by the autonomous decision engine; the calculation process of the physical property emergence prediction model comprises the steps of firstly, calculating an average motion vector field of a particle group from an image sequence of a high-definition camera array through an optical flow method, and extracting the statistical characteristics of the rotation and the divergence of the average motion vector field as macroscopic mixing dynamics characteristic vectors Simultaneously, extracting power spectrum entropy characteristics from multi-axis torque signals Then inverting the uniformity of the distribution of the binder from the near infrared spectrum signal Finally, the macroscopic mixing dynamics characteristic vector Entropy characterization of power spectrum Uniformity of binder distribution And the current material characteristic vector Fusing, inputting a pre-trained deep neural network, and outputting the deep neural network as the compaction rate of molding sand for a mixed endpoint With wet compression strength The mapping relation of the joint predicted value of (2) is expressed as: ; Dynamically optimizing objective functions in the autonomous decision engine The construction process of the method comprises the following steps of taking a molding sand performance predicted value close to a target value and the minimum energy consumption in the prediction process as an optimization target, wherein the mathematical expression is as follows: , wherein, And The performance prediction vector and the target vector are respectively, Representing the square of the euclidean norm, And Is a weighting coefficient, and the integral term represents the current time To a predetermined mixing end time Is used for the pre-estimated energy consumption of the vehicle, For stirring rotation speed predicted according to torque model The function of the torque to be associated with, To be over time A varying rotational speed function.
2. The intelligent material proportion self-adaptive control system based on the sand mixer according to claim 1, wherein the rolling time domain optimization algorithm is implemented by searching a rotation speed control function in a future limited time domain at each optimization trigger time A matching correction vector To optimize the variables, the rotational speed control function and the proportioning correction vector jointly form a hybrid process cooperative control sequence to minimize the objective function of the multi-modal digital twin body prediction at the time domain endpoint under the action of the hybrid process cooperative control sequence The value is the target, and the ratio correction vector is fixed at first when solving Pseudo-spectral method is used to control the function for the rotation speed The continuous time optimization problem of (2) is converted into a nonlinear programming problem and solved, and then a gradient descent method is adopted to calculate a proportioning correction vector capable of further reducing the performance prediction deviation based on the obtained rotating speed control function And outputting an optimization instruction after performing a plurality of iterations.
3. The intelligent material proportion self-adaptive regulation and control system based on the sand mixer is characterized in that a deep neural network in a physical property emergence prediction model is provided with an online self-updating mechanism, and the updating process comprises the steps of inputting a process sensing data sequence into the model to obtain a final performance predicted value after each batch is completed, calculating a predicted error with a laboratory measured value, storing a training sample formed by the predicted error and corresponding process data into an experience pool, periodically sampling from the experience pool, and performing fine tuning training on the deep neural network to update network weights.
4. The intelligent material proportioning self-adaptive control system based on sand mixer as claimed in claim 3, wherein said global knowledge base further stores a strategy network mapping specific material characteristics and working conditions to optimal adjustment strategies, when starting batches with novel material characteristic combinations, said autonomous decision engine searches similar cases in the global knowledge base first, if no matching cases, starts strategy search based on evolution algorithm, uses strategy network parameters as gene codes to generate initial population, utilizes digital twin to evaluate objective function values corresponding to each strategy in the population And updating strategy network parameters by adopting a covariance matrix self-adaptive evolution strategy algorithm until strategies meeting the conditions are obtained and archived.
5. The intelligent material proportion self-adaptive regulation and control system based on a sand mixer according to claim 4, wherein the parameter updating process of the covariance matrix self-adaptive evolution strategy algorithm comprises the steps of maintaining a mean vector of current optimal strategy parameters And a covariance matrix At each generation, from multiple normal distribution The method comprises the steps of sampling to generate a preset number of child strategy parameter vectors, evaluating and selecting the best-performing child strategy parameter vector by utilizing a digital twin body According to this Recalculating the parameters of the offspring Sum covariance matrix And adjusting the step length The mean value updating formula is as follows: Wherein Is the first The parameter vector of the one of the optimal offspring, Is a weight coefficient.
6. The intelligent material proportion self-adaptive regulation and control system based on a sand mixer as claimed in claim 5, wherein the high-precision material characterization unit detects the surface activity index of the raw material on line through a micro fluidized bed device The detection process comprises the steps of performing controllable airflow fluidization on a raw material sample, measuring a scattered light intensity fluctuation signal of a particle swarm through laser scattering, performing fast Fourier transform on the light intensity fluctuation signal, and calculating energy integral within a predefined characteristic frequency bandwidth Total energy of the entire spectrum Index of surface Activity Obtained by calculation of the formula: 。
7. The intelligent material proportion self-adaptive regulation and control system based on the sand mixer is characterized in that the operation process of the abnormal processor is that mutation of signals of a multi-mode process sensing unit is monitored in real time, when the signals exceed a historical normal fluctuation range, abnormal working conditions are judged, then a pre-stored abnormal case library in the global knowledge base is queried for pattern matching, and after matching is successful, a corresponding standardized toughness intervention program is executed.
8. The intelligent material proportioning self-adaptive control system based on the sand mixer, as set forth in claim 7, wherein the operation process of the command checking and smoothing device comprises the steps of performing projection correction on proportioning correction amount according to the residual amount of the material bin, the precision range of the blanking mechanism and the process constraint rule to obtain a feasible solution, and then performing exponential weighting moving average processing on feasible proportioning correction amount sequences generated by a plurality of continuous batches, wherein a smoothing formula is as follows: , wherein, As a smoothing factor, the smoothing factor is used, For the indexing of the batch(s), Is the first The feasible proportioning correction quantity obtained by batch calculation, Is the first And finally outputting correction quantity after batch smoothing.

Description

Intelligent material proportion self-adaptive regulation and control system based on sand mixer Technical Field The invention relates to the technical field of sand mixing regulation, in particular to an intelligent material proportion self-adaptive regulation system based on a sand mixer. Background In casting production, the sand mixing procedure is used for uniformly mixing raw sand, binder, water and other additives according to a certain proportion so as to obtain molding sand with qualified performance. The sand mixing quality directly affects the surface quality, dimensional accuracy and internal quality of the castings. Currently, most sand mixing processes still rely on manual experience to set proportioning and process parameters, or employ simple automation systems based on fixed proportioning and time control. The method is difficult to adapt to the fluctuation of raw material batches, the environmental change and the nonlinear time-varying characteristic in the mixing process, and the problems of uneven mixing, unstable performance, high energy consumption and the like are easily caused. In recent years, some advanced sand mixing devices begin to introduce sensors for process monitoring, such as torque monitoring, near infrared spectrum analysis, and the like. However, the existing system is limited to single signal feedback and local parameter adjustment, lacks multi-dimensional fusion perception of macroscopic and microscopic states of the mixing process, and further fails to realize prospective optimization and cross-batch continuous learning based on mixing endpoint performance prediction. Therefore, a person skilled in the art provides an intelligent material proportioning self-adaptive control system based on a sand mixer so as to solve the problems in the background art. Disclosure of Invention The invention aims to solve the technical problem of overcoming the defects of the prior art and providing a material proportion regulating and controlling system of a sand mixer based on multi-mode sensing and autonomous optimization. The system constructs a digital twin body by fusing multisource perception data, realizes visualization and accurate prediction of performance in a mixing process, dynamically and cooperatively adjusts stirring parameters and material proportion based on a rolling optimization and autonomous learning mechanism, and finally realizes stable molding sand performance, energy consumption optimization and self-adaptive toughness control on abnormal working conditions. In order to achieve the above purpose, the present invention provides the following technical solutions: Intelligent material proportion self-adaptive regulation and control system based on sand mixer includes: the high-precision material characterization unit is used for detecting the particle size distribution, the true density, the water content and the surface activity index of each batch of raw materials on line and generating a dynamic material characteristic vector; the multi-mode process sensing unit comprises a multi-axis torque sensor arranged on a main shaft of the sand mixer, an internal high-definition camera array and an on-line near infrared spectrometer at a discharge port, and is used for synchronously acquiring a torque vibration signal, a visual image sequence and a chemical spectrum signal of the mixing process; The global knowledge base is used for storing historical production case data, optimization strategies and model parameters; the core processor is respectively connected with the high-precision material characterization unit, the multi-mode process sensing unit and the global knowledge base, and comprises: the multi-mode digital twin body is used for receiving real-time sensing data and material characteristic vectors, and simulating and predicting performance indexes of the molding sand at the mixing endpoint through a built-in physical property emergence prediction model; The autonomous decision engine is internally provided with a dynamic optimization objective function, and calculates a stirring parameter adjustment instruction and a material proportioning correction amount through a rolling time domain optimization algorithm based on the prediction output of the multi-modal digital twin body; The abnormal processor is used for identifying abnormal working conditions according to abrupt change of the sensing signals and triggering corresponding toughness control programs; the command checking and smoothing device is used for carrying out feasibility checking and smoothing filtering on the proportioning correction quantity output by the autonomous decision engine; The control flow of the system comprises the steps of starting mixing based on initial proportioning, predicting performance by utilizing a multi-mode digital twin body, generating a collaborative optimization instruction by an autonomous decision engine, wherein the collaborative optimization instruction comprises st