Search

CN-122007198-A - Dynamic pressure compensation method for zero-working aluminum profile extrusion die

CN122007198ACN 122007198 ACN122007198 ACN 122007198ACN-122007198-A

Abstract

The invention relates to the technical field of metal plastic processing, in particular to a dynamic pressure compensation method of a zero-working extrusion die with aluminum profiles, which comprises the following steps of synchronously collecting main cylinder pressure in the extrusion process according to a set control period Actual speed of extrusion rod Temperature of die Process variable for characterizing extrusion processes Based on the sensor data in the current collection and past preset time window, constructing a multidimensional input characteristic vector containing time sequence characteristics and context characteristics And inputting the characteristic vector Inputting into a pre-trained neural network prediction model, calculating and outputting the extrusion rod speed control quantity for pressure compensation in real time . According to the invention, pressure and flow fluctuation caused by dynamic factors such as temperature change, material difference, die abrasion and the like are effectively compensated, and high stability control of the outlet speed is realized.

Inventors

  • ZHANG LIGANG
  • TANG XIANMING

Assignees

  • 江苏江顺精密科技集团股份有限公司

Dates

Publication Date
20260512
Application Date
20251230

Claims (10)

  1. 1. The dynamic pressure compensation method of the zero-working extrusion die with the aluminum profile is characterized by comprising the following steps of: Step 1, synchronously collecting the master cylinder pressure in the extrusion process according to a set control period Actual speed of extrusion rod Temperature of die Process variable for characterizing extrusion processes Sensor data; Step 2, based on the sensor data in the current collection and the past preset time window, constructing a multidimensional input characteristic vector containing time sequence characteristics and context characteristics ; Step 3, the input characteristic vector is processed Inputting into a pre-trained neural network prediction model, calculating and outputting the extrusion rod speed control quantity for pressure compensation in real time ; Step 4, controlling the quantity according to the output speed The servo actuating mechanism of the extruder is driven in real time to dynamically compensate pressure fluctuation caused by die temperature rise, material rheological stress change and die abrasion factors.
  2. 2. A dynamic pressure compensation method for a zero-working aluminum profile extrusion die as set forth in claim 1, wherein in step 1, process variables for characterizing the extrusion process are used For the extrusion time Or extruded length 。
  3. 3. The method for dynamic pressure compensation of a zero-working extrusion die for aluminum profiles according to claim 1, wherein the feature vector is as follows The method specifically comprises the following steps: Defining a current time and a previous time Sliding time windows of historical moments, and acquiring time sequence data matrix in window ; For the time sequence data matrix Extracting features to generate a combined feature set containing original values, statistical features and change rate features; splicing the combined feature set with a static context feature vector C to form the input feature vector Wherein the contextual feature vector C is comprised of an encoding of alloy brands and die numbers.
  4. 4. A dynamic pressure compensation method for a zero-working aluminum profile extrusion die as set forth in claim 3, wherein said statistical features include an average value of master cylinder pressures within said time window And standard deviation of The change rate characteristic comprises a change rate of the temperature of the mold And master cylinder pressure rate of change 。
  5. 5. The method for compensating dynamic pressure of a zero-working extrusion die with aluminum profiles according to claim 1, wherein the neural network prediction model is a time sequence prediction model based on a long-short-term memory network LSTM.
  6. 6. The method for dynamic pressure compensation of a zero-working extrusion die with aluminum profile as set forth in claim 5, wherein the neural network prediction model adopts an encoder-decoder structure, wherein the encoder is an LSTM layer for extracting input feature vectors The decoder is a full connection layer for mapping the high-level time sequence characteristics into the speed control quantity 。
  7. 7. The method for dynamic pressure compensation of a zero-working aluminum profile extrusion die according to claim 1, wherein in step S3, the method is specifically implemented as a first control mode: the neural network prediction model uses the time sequence data in the sliding time window And the context characteristic C is taken as input, and the target set speed value of the extrusion rod in the next control period is directly output, namely And (2) and Wherein And predicting a model for the neural network.
  8. 8. The method for dynamic pressure compensation of a zero-working aluminum profile extrusion die according to claim 1, wherein in step S3, the method is specifically implemented as a second control mode: the neural network prediction model uses the time sequence data in the sliding time window And the context characteristic C is used as input to output the predicted value of the master cylinder pressure variation after the future set time length, namely ; Subsequently, the predicted value is calculated With the current pressure Adding to obtain a pressure predicted value ; Finally, a proportional-integral-derivative PID controller is used for controlling the pressure according to the target pressure And the predicted value of the pressure Deviation of (2) Calculating the speed compensation amount And accordingly updating the target set speed of the extrusion rod 。
  9. 9. The method for dynamic pressure compensation of a zero-working aluminum profile extrusion die according to claim 1, wherein in step S4, the driving servo actuator is specifically: setting the target speed A motion controller that sends to the extruder as a command value; The motion controller adjusts the current of the servo proportional valve or the displacement of the variable pump through a closed-loop control algorithm to enable the actual speed of the extrusion rod to follow the instruction value And (3) a change.
  10. 10. The method for dynamic pressure compensation of a zero-working aluminum profile extrusion die as claimed in claim 1, further comprising a model online self-adaptation step: continuously monitoring stability indexes of the profile outlet speed during the operation of the system; When the stability index exceeds a preset threshold or the production batch is switched, triggering a model updating flow, and performing incremental learning or fine adjustment on the neural network prediction model by utilizing the latest acquired production data.

Description

Dynamic pressure compensation method for zero-working aluminum profile extrusion die Technical Field The invention relates to a dynamic pressure compensation method of a zero-working extrusion die with an aluminum profile, and belongs to the technical field of metal plastic processing. Background Extrusion production of aluminum profiles is a complex dynamic process, the product quality (such as dimensional accuracy, surface quality and tissue performance) and productivity of which are deeply affected by process stability, and during extrusion, various dynamic factors can cause fluctuation of master cylinder pressure and metal flow state, and the factors include, but are not limited to, temperature gradient of an aluminum ingot head, a middle tail, continuous temperature rise and heat balance of a die in operation, temperature change of an extrusion barrel, and die working belt abrasion which gradually occurs with increase of production rod times; The conventional control method generally adopts a constant-speed extrusion mode or a constant-pressure extrusion mode, wherein the constant-speed extrusion mode is adopted, the fixed extrusion rod speed is preset, the master cylinder pressure is maintained constant through simple PID feedback adjustment, however, the methods are all 'post-remediation' type control, dynamic interference of multivariable coupling cannot be foreseen and counteracted, so that the problems of profile outlet speed fluctuation, size out-of-tolerance, surface defects (such as corrugation and bright band) and the like are generated, and abnormal abrasion of a die is aggravated, so that improvement on a dynamic pressure compensation method of a zero-working aluminum profile extrusion die is needed to solve the problems. Disclosure of Invention The invention aims to provide a dynamic pressure compensation method for a zero-working extrusion die with aluminum profiles, which aims to solve the problems that the traditional control method usually adopts a constant-speed extrusion mode or a constant-pressure extrusion mode, the former is used for presetting a fixed extrusion rod speed, and the latter is used for maintaining the pressure of a main cylinder to be constant through simple PID feedback adjustment, however, the methods are all 'post-remedy' type control, the dynamic interference of multivariable coupling cannot be foreseen and counteracted, so that the speed of an outlet of the profile fluctuates, and further the problems of dimension out-of-tolerance, surface defects (such as ripples and bright bands) and the like are generated, and the abnormal abrasion of the die is aggravated. In order to achieve the above purpose, the present invention provides the following technical solutions: A dynamic pressure compensation method of a zero-working extrusion die with aluminum profiles comprises the following steps: Step 1, synchronously collecting various sensor data installed on an extruder and a die in a fixed control period (preferably 10-100 milliseconds) in the extrusion process; the acquired data at least comprises master cylinder pressure Actual displacement/speed of extrusion rodTemperature of dieAnd variables characterizing the extrusion process(Extrusion time)Or extruded length); Step 2, converting the original time sequence data into characteristic vectors capable of fully representing the dynamic state of the current systemThe method specifically comprises the following steps: Sliding window interception, defining a window containing current time t and front Sliding time windows of historical moments, wherein the window length w is set according to process dynamic characteristics (corresponding to 5-30 seconds of data), and a time sequence data matrix in the window is obtainedWhereinSensor data vector for time i; feature engineering calculation, namely, window data Performing online calculation, and extracting three types of features: original features sensor readings at the present instant ; Statistical characteristics, calculating statistics of key parameters in window, mean value of master cylinder pressureAnd standard deviationThe pressure sensor is used for representing the average level and fluctuation intensity of pressure; calculating the change trend of key parameters and the change rate of the temperature of the die And master cylinder pressure rate of changeCapturing a dynamic direction of a process; Generating a static context feature vector C, wherein the static context feature vector C comprises an alloy brand and a die number of a current production task; Feature splicing, namely splicing the original features, the statistical features and the change rate features obtained by calculation with the context feature vector C to form a final multidimensional input feature vector ; Step 3, constructing the feature vectorInput to a pre-trained neural network prediction modelIn the method, the model can learn the mapping relation from complex working condition characteristi