CN-122018312-A - Multi-parameter feedback and LSTM-based aluminum profile heat treatment system

CN122018312ACN 122018312 ACN122018312 ACN 122018312ACN-122018312-A

Abstract

The invention relates to the technical field of aluminum profile heat treatment control, in particular to an aluminum profile heat treatment system based on multi-parameter feedback and LSTM, which comprises a server and an execution terminal, wherein the server comprises a multi-dimensional parameter acquisition module, a feature fusion pretreatment module, an LSTM prediction optimization module, a deviation grading control module and a dynamic iteration storage module. The system comprises a multidimensional parameter acquisition module, a characteristic fusion preprocessing module, an LSTM prediction optimization module, a deviation grading control module, a dynamic iteration storage module, a periodic incremental training model and a parameter mapping relation optimization module, wherein the multidimensional parameter acquisition module acquires real-time heat treatment temperature, section thickness, cooling medium flow rate and finished product tensile strength, the characteristic fusion preprocessing module outputs a state parameter vector after data purification optimization, the LSTM prediction optimization module predicts the tensile strength based on the vector and combines a deviation value and section thickness output parameter adjustment quantity, the deviation grading control module performs differentiation adjustment according to a deviation grade, and the dynamic iteration storage module stores data and regularly increments the training model. According to the invention, through multi-parameter cooperation and LSTM nonlinear adaptation and dynamic iteration, the stability and the adaptability of the heat treatment quality of the aluminum profile are improved.

Inventors

HU ZHENGLIANG
HU YONGDE
ZHANG XIAOSHU
ZHANG BANGQUAN

Assignees

贵州正合轻合金科技有限责任公司

Dates

Publication Date: 20260512
Application Date: 20251230

Claims (5)

1. The aluminum profile heat treatment system based on multi-parameter feedback and LSTM is characterized by comprising a server and an execution terminal, wherein the server comprises a multi-dimensional parameter acquisition module, a characteristic fusion preprocessing module, an LSTM prediction optimization module, a deviation grading control module and a dynamic iteration storage module; The multi-dimensional parameter acquisition module is used for controlling the multi-sensor array to acquire original state parameters of the aluminum profile heat treatment process and the tensile strength of the heat treated finished product, wherein the original state parameters comprise real-time heat treatment temperature, profile thickness and cooling medium flow rate; the feature fusion preprocessing module is used for carrying out noise removal, key feature extraction, feature cross fusion and standardization processing on the original state parameters and then outputting state parameter vectors; The LSTM prediction optimization module is internally provided with an LSTM deep learning model and is used for outputting a target tensile strength predicted value of the aluminum profile after heat treatment by inputting a state parameter vector, outputting an original state parameter adjustment quantity comprising a real-time heat treatment temperature adjustment quantity and a cooling medium flow speed adjustment quantity based on a deviation value of the target tensile strength predicted value and a preset target tensile strength and combining profile thickness suitability; The deviation grading control module is used for grading deviation grades according to the deviation values, correspondingly executing a differential dynamic adjustment strategy, and associating the adjustment strategy with the deviation grades, the original state parameter adjustment quantity and the profile thickness suitability; The dynamic iteration storage module is used for storing original state parameters, finished product tensile strength, state parameter vectors, target tensile strength predicted values, state parameter adjustment instructions and deviation grades; The dynamic iteration storage module is internally provided with an incremental training unit, when the stored data quantity reaches a preset storage threshold value, the incremental iterative training is carried out on the LSTM deep learning model by combining the historical core parameter data, and the mapping relation between the state parameter vector and the target tensile strength predicted value and the original state parameter adjustment quantity and the model parameters are updated; The execution terminal comprises a heating control unit and a cooling control unit, wherein the temperature is adjusted by the heating control unit to realize real-time heat treatment temperature adjustment, and the rotating speed is adjusted by the cooling control unit to realize cooling medium flow speed adjustment.
2. The aluminum profile heat treatment system based on multi-parameter feedback and LSTM as set forth in claim 1, wherein the initial state parameters further comprise an aluminum profile initial temperature, the multi-sensor array comprises a temperature sensor for collecting real-time heat treatment temperature, a thickness sensor for collecting profile thickness, a flow rate sensor for collecting cooling medium flow rate, an intensity detection sensor for collecting finished product tensile strength and an infrared thermometer for collecting the aluminum profile initial temperature, and the execution terminal further comprises an initial temperature adjusting unit for adjusting the aluminum profile initial temperature.
3. The aluminum profile heat treatment system based on multi-parameter feedback and LSTM according to claim 2, wherein the processing flow of the characteristic fusion pretreatment module comprises the steps of removing random noise in original state parameters through Kalman filtering, extracting key characteristics through principal component analysis, fusing characteristic vectors of different dimensional parameters into a high-dimensional characteristic matrix through a characteristic cross fusion unit, normalizing the characteristic matrix to a preset interval to generate a state parameter vector, wherein in the LSTM prediction optimization module, calculation of an original state parameter adjustment quantity is positively correlated with an offset value, and the subintervals are divided according to profile thickness, and different subintervals correspond to different adjustment quantity adaptation coefficients to adapt to heat conduction differences of aluminum profiles with different thicknesses.
4. The aluminum profile heat treatment system based on multi-parameter feedback and LSTM as set forth in claim 3, wherein the deviation classification control module is divided into three levels, wherein the first level deviation is not greater than a first threshold value, the second level deviation is not greater than the first threshold value and is not greater than a second threshold value, the third level deviation is greater than the second threshold value, the corresponding adjustment strategy is that the first level deviation keeps the current original state parameters and records the related data, the second level deviation preferentially executes the original state parameter adjustment amount and adapts the thickness adjustment coefficient, the third level deviation starts the emergency adjustment mode and executes the adjustment amount, the adapted thickness adjustment coefficient and the starting equipment self-test according to the upper limit.
5. The multi-parameter feedback and LSTM based aluminum profile heat treatment system as set forth in claim 4, wherein the dynamic iterative storage module further comprises a feature weight analysis unit for counting core original state parameter weight duty ratios and optimal adjustment coefficients of different aluminum profile series to form a material, thickness and adjustment coefficient mapping table, and providing initial parameter matching basis for new material and new specification aluminum profile heat treatment.

Description

Multi-parameter feedback and LSTM-based aluminum profile heat treatment system Technical Field The invention relates to the technical field of aluminum profile heat treatment control, in particular to an aluminum profile heat treatment system based on multi-parameter feedback and LSTM. Background The heat treatment quality of the aluminum profile as a common material in industrial production directly influences the mechanical property and the service life of the product. The existing aluminum profile heat treatment technology mainly depends on traditional fixed parameter control or single parameter feedback adjustment, and has the defects that temperature is used as a core control parameter, key influencing factors such as profile thickness, cooling speed and the like are ignored, so that the tensile strength deviation of the aluminum profiles in the same batch is large, the product quality stability is poor, the traditional PID control algorithm is difficult to adapt to nonlinear characteristics in the heat treatment process, the response to process parameter changes is lagged, the adjustment precision is limited, the adaptability is lacking, the process parameters of the aluminum profiles in different batches and different specifications are required to be manually debugged again, the operation is complicated, the production efficiency is low, no historical production data optimization control strategy cannot be adopted, and the control precision is easy to drop after long-term use. Therefore, an intelligent heat treatment system combining multi-parameter feedback and intelligent algorithm is needed, the problems of single parameter control, poor nonlinear adaptation and insufficient self-adaptability in the prior art are solved, and the accuracy and the intelligent level of heat treatment of aluminum profiles are improved. Disclosure of Invention Aiming at the defects of the prior art, the invention aims to provide an aluminum profile heat treatment system based on multi-parameter feedback and LSTM, solves the problems of weak adaptive capacity, lack of self-adaptability and low control precision of the traditional aluminum profile heat treatment system, can realize dynamic optimization of process parameters and improves the precision of aluminum profile heat treatment. The invention provides a basic scheme that an aluminum profile heat treatment system based on multi-parameter feedback and LSTM comprises a server and an execution terminal, wherein the server comprises a multi-dimensional parameter acquisition module, a feature fusion preprocessing module, an LSTM prediction optimization module, a deviation grading control module and a dynamic iteration storage module; The multi-dimensional parameter acquisition module is used for controlling the multi-sensor array to acquire original state parameters of the aluminum profile heat treatment process and the tensile strength of the heat treated finished product, wherein the original state parameters comprise real-time heat treatment temperature, profile thickness and cooling medium flow rate; the feature fusion preprocessing module is used for carrying out noise removal, key feature extraction, feature cross fusion and standardization processing on the original state parameters and then outputting state parameter vectors; The LSTM prediction optimization module is internally provided with an LSTM deep learning model and is used for outputting a target tensile strength predicted value of the aluminum profile after heat treatment by inputting a state parameter vector, outputting an original state parameter adjustment quantity comprising a real-time heat treatment temperature adjustment quantity and a cooling medium flow speed adjustment quantity based on a deviation value of the target tensile strength predicted value and a preset target tensile strength and combining profile thickness suitability; The deviation grading control module is used for grading deviation grades according to the deviation values, correspondingly executing a differential dynamic adjustment strategy, and associating the adjustment strategy with the deviation grades, the original state parameter adjustment quantity and the profile thickness suitability; The dynamic iteration storage module is used for storing original state parameters, finished product tensile strength, state parameter vectors, target tensile strength predicted values, state parameter adjustment instructions and deviation grades; The dynamic iteration storage module is internally provided with an incremental training unit, when the stored data quantity reaches a preset storage threshold value, the incremental iterative training is carried out on the LSTM deep learning model by combining the historical core parameter data, and the mapping relation between the state parameter vector and the target tensile strength predicted value and the original state parameter adjustment quantity and the model parameters are updated; Th