CN-122018592-A - Production temperature control method of wire drawing machine

CN122018592ACN 122018592 ACN122018592 ACN 122018592ACN-122018592-A

Abstract

The invention belongs to the technical field of intelligent control, and particularly relates to a production temperature control method of a wire drawing machine. The method comprises the steps of firstly obtaining temperature data of a plurality of temperature areas of a wire drawing machine, determining a temperature control setting range, processing temperature area coupling interference through an improved neural network decoupling algorithm to obtain independent controlled temperature values of all the temperature areas, constructing an orthogonal mode PID controller, establishing unique mapping relation between PID three parameters and orthogonal modes, extracting single-mode core feature quantity through self-adaptive orthogonal mode decomposition, completing three-parameter uncoupled self-adaptive setting, and adopting a sparrow searching algorithm for improving elite retention variation mechanism to globally optimizing control parameters to obtain an optimal parameter set so as to realize real-time temperature control of the temperature areas. The invention effectively solves the problems of coupling interference of multiple temperature areas and difficult PID parameter setting of the wire drawing machine, obviously improves the temperature control precision, the production stability and the product forming quality, and meets the high-precision temperature control requirement of high-end wire drawing production.

Inventors

FU CHONGJIAN
CHEN GONGSHENG
LI RUILIANG

Assignees

山东鑫大地控股集团有限公司

Dates

Publication Date: 20260512
Application Date: 20260326

Claims (7)

1. The production temperature control method of the wire drawing machine is characterized by comprising the following steps of: S1, acquiring temperature data of multiple temperature areas of a target wire drawing machine, and determining a set range of temperature control of each temperature area; s2, based on temperature data of multiple temperature areas, decoupling the coupling interference of each temperature area by improving a neural network decoupling algorithm to obtain independent controlled temperature values of each temperature area; S3, constructing an orthogonal mode PID controller with mode parameters bound one by one according to the independent controlled temperature values of the decoupled temperature areas, and constructing a unique mapping relation between PID three parameters and the orthogonal mode to realize decoupling of control parameters; S4, carrying out self-adaptive orthogonal mode decomposition on the real-time temperature response data of each temperature zone, and extracting single-mode core feature quantity which is bound with each PID parameter one by one based on a temperature set value and real-time output data; s5, based on the extracted core characteristic quantity of the corresponding mode, the coupling-free self-adaptive setting of the three parameters of proportion, integral and differential is completed, and an improved temperature controller is obtained; S6, adopting a sparrow search algorithm, improving an elite retention variation mechanism to globally optimize control parameters of an improved temperature controller, obtaining an optimal control parameter set, and regulating and controlling a corresponding temperature region of the wire drawing machine in real time.
2. The method for controlling the production temperature of the wire drawing machine according to claim 1, wherein the step S2 is characterized in that based on the temperature data of the multiple temperature areas, decoupling processing is performed on the coupling interference of each temperature area by improving a neural network decoupling algorithm, and the specific implementation of obtaining the independent controlled temperature value of each temperature area is as follows: S21, setting a temperature control system of the wire drawing machine to comprise n independent heating temperature areas, taking heating control quantity of each temperature area as system input and taking a real-time temperature sampling value of each temperature area as system output, and constructing a multi-input multi-output system model reflecting temperature coupling characteristics among temperature areas in a discrete domain; S22, constructing a decoupling neural network which sequentially comprises an input layer, a hidden layer and a decoupling output layer, wherein the total number of nodes of the input layer is 2n, the input is a real-time temperature sampling value of each temperature region and a temperature sampling value of the previous moment, the hidden layer adopts a sparse connection structure with a temperature coupling priori, only the connection between the corresponding nodes of the adjacent temperature regions is reserved, the connection weight of the corresponding nodes of the non-adjacent temperature regions is fixed to 0 and does not participate in training update, the nonlinear characteristic of temperature coupling is fitted through a hyperbolic tangent activation function, and a hidden layer vector is output, the total number of nodes of the decoupling output layer is n, firstly, a strict diagonal linear mapping which is output to the output layer by the hidden layer is established, trainable non-zero weight is reserved between the ith node of the hidden layer and the ith node of the output layer, the connection weight of other cross nodes is fixed to 0, and then limiting saturation activation processing with upper limit and lower limit is carried out on a linear mapping result, and finally the decoupling compensation control quantity of each temperature region is output; S23, defining a relative decoupling degree index of a single temperature zone for temperature coupling, and constructing a loss function of the decoupling neural network based on the relative decoupling degree index; S24, collecting a full-working-condition multi-temperature-zone temperature sample set in the historical production process of the wire drawing machine as training data, pre-training the constructed decoupling neural network until the loss function converges to a preset convergence threshold or reaches a set maximum training frequency, and completing the pre-training and obtaining a pre-trained decoupling neural network model; S25, superposing decoupling compensation control quantity of each temperature zone output by the decoupling neural network in real time with heating control quantity input of the corresponding temperature zone according to one-to-one correspondence of the temperature zones to generate total control input which is finally input to each temperature zone heating actuator of the wire drawing machine, substituting the total control input into the temperature coupling system model, disassembling diagonal inherent characteristic items and non-diagonal coupling characteristic items of the transfer function matrix, completely counteracting the compensation items and the coupling interference items through the decoupling compensation control quantity, and finally outputting total control input of each temperature zone heating actuator which is the independent controlled temperature set value corresponding to each temperature zone.
3. The method according to claim 1, wherein in the step S3, for the decoupled independent controlled temperature values of each temperature zone, an orthogonal mode PID controller with one-to-one binding mode parameters is constructed, and the decoupling of the control parameters is specifically implemented by establishing a unique mapping relationship between the PID three parameters and the orthogonal mode: S31, aiming at the i-th temperature zone after decoupling, taking the independently controlled temperature set value of the temperature zone as a target value Taking a real-time temperature sampling value at the moment k of the temperature zone as an output value Constructing a single-input single-output closed-loop control loop of the temperature zone, wherein k is discrete sampling time, i is temperature zone number, and n is total number of independent heating temperature zones of the wire drawing machine; S32, establishing a unique mapping binding relation between PID proportion, integral and differential three parameters and an orthogonal mode, and outputting real-time temperature at the moment k of the ith temperature zone The method is characterized in that the method is uniquely decomposed into the sum of three modal components which are orthogonal in pairs, the orthogonality is strictly constrained by setting the inner product of vectors in a sliding time window with the length of M to be 0, and the decomposition formula under a discrete domain is as follows: And (2) and , wherein, For vector inner product operations within a sliding time window, The temperature sequence vectors of the reference tracking mode, the oscillation suppression mode and the drift compensation mode in the sliding window are respectively.
4. The method for controlling the production temperature of a wire drawing machine according to claim 1, wherein in the step S4, the real-time temperature response data of each temperature zone is subjected to adaptive orthogonal modal decomposition, and the specific implementation of extracting the single-mode core feature quantity one-to-one bound with each PID parameter based on the temperature set value and the real-time output data is as follows: S41, setting a sliding time window with the length of M for an ith temperature zone, and constructing a temperature response data vector in the window: Temperature target value vector: ; S42, taking the trend of the temperature target value as a reference, and obtaining a reference tracking modal vector through least square orthographic projection The core feature quantity of the reference tracking mode is extracted to be the reference tracking error: ; S43, to residual error vector Alternating oscillation components are extracted through high-pass orthogonal projection, and oscillation suppression modal vectors are obtained The core characteristic quantity is the energy of oscillation mode ; S44, for the final residual vector The core feature quantity is drift mode accumulated quantity Where m is the cyclic variable of the sum operation.
5. The method of claim 4, wherein the step S5 is based on the extracted core feature quantity of the corresponding mode, and the complete decoupling self-adaptive tuning of the three parameters of proportion, integral and differential is completed, and the specific implementation of the improved temperature controller comprises the following steps: the self-adaptive setting of the update proportion of the piecewise function is adopted as follows: , wherein, And updating integral self-adaptive setting for the basic value of the proportionality coefficient to be: wherein As the basis value of the integral coefficient, The maximum calibration value of the accumulated quantity of the drift compensation modes under the rated working condition of the ith heating temperature zone, Taking 1 when the output quantity of the controller does not reach the upper limit of the control quantity of the heating actuator of the wire drawing machine, taking 0 when the output quantity of the controller reaches the upper limit, and updating differential self-adaptive setting to be: , wherein, As the basic value of the differential coefficient, In order to take the maximum value of the current oscillation mode energy and the minimum threshold value, ensure that the denominator is always a non-zero effective value, And finally, improving the output of the temperature controller to be: 。
6. The method for controlling the production temperature of the wire drawing machine according to claim 1, wherein the improved elite retention variation mechanism in the step S6 is characterized in that population individuals are divided into three levels of elite individuals, common individuals and inferior individuals after being sequenced according to fitness values, differential retention and variation rules are executed aiming at the characteristics of the individuals at different levels, elite retention is firstly executed after each round of iteration is completed to lock a high-quality parameter gene which is suitable for temperature control of the wire drawing machine, differential variation operation is executed according to the levels to update the population, and an optimal control parameter set is output after iteration is carried out to preset termination conditions, so that real-time control is carried out on a temperature region corresponding to the wire drawing machine.
7. The method for controlling the production temperature of the wire drawing machine according to claim 6, wherein the specific rules of hierarchical division and differential variation are that individuals with the top 10% of the fitness value rank after each round of iteration are divided into elite individuals, the elite individuals are directly and completely reserved for the next round of iteration without any variation operation, the individuals with the middle 30% to 70% of the rank are divided into common individuals, the common individuals are subjected to slight single-parameter random variation, any one of three parameters of PID is only randomly adjusted, local optimization precision is enhanced while the high-quality genes of a main body are reserved, the individuals with the 20% of the rank are divided into inferior individuals, full-parameter reset variation is performed on the inferior individuals, and new individuals conforming to the stable boundaries of the parameters are completely regenerated.

Description

Production temperature control method of wire drawing machine Technical Field The invention belongs to the technical field of intelligent control, and particularly relates to a production temperature control method of a wire drawing machine. Background The wire drawing machine is core equipment in wire drawing production of metal wires and high polymer materials, the temperature control precision of a multi-temperature-zone heating system directly determines the molding quality, mechanical property and production stability of products, and the high-precision temperature regulation and control of temperature zones has become the core requirement of a high-end wire drawing production process. Therefore, independent temperature adjustment needs to be carried out on each temperature zone, and therefore, accurate and uncoupled control of the temperatures of the multiple temperature zones becomes a key for researching and developing a temperature control system of the wire drawing machine. Disclosure of Invention The invention provides a production temperature control method of a wire drawing machine aiming at the technical problems in the background technology. In order to achieve the above purpose, the technical scheme adopted by the invention comprises the following steps: S1, acquiring temperature data of multiple temperature areas of a target wire drawing machine, and determining a set range of temperature control of each temperature area; s2, based on temperature data of multiple temperature areas, decoupling the coupling interference of each temperature area by improving a neural network decoupling algorithm to obtain independent controlled temperature values of each temperature area; S3, constructing an orthogonal mode PID controller with mode parameters bound one by one according to the independent controlled temperature values of the decoupled temperature areas, and constructing a unique mapping relation between PID three parameters and the orthogonal mode to realize decoupling of control parameters; S4, carrying out self-adaptive orthogonal mode decomposition on the real-time temperature response data of each temperature zone, and extracting single-mode core feature quantity which is bound with each PID parameter one by one based on a temperature set value and real-time output data; s5, based on the extracted core characteristic quantity of the corresponding mode, the coupling-free self-adaptive setting of the three parameters of proportion, integral and differential is completed, and an improved temperature controller is obtained; S6, adopting a sparrow search algorithm, improving an elite retention variation mechanism to globally optimize control parameters of an improved temperature controller, obtaining an optimal control parameter set, and regulating and controlling a corresponding temperature region of the wire drawing machine in real time. Preferably, in the step S2, based on the temperature data of the multiple temperature areas, decoupling processing is performed on the coupling interference between the temperature areas by improving a neural network decoupling algorithm, so as to obtain the specific implementation of the controlled temperature value of each temperature area independently: S21, setting a temperature control system of the wire drawing machine to comprise n independent heating temperature areas, taking heating control quantity of each temperature area as system input and taking a real-time temperature sampling value of each temperature area as system output, and constructing a multi-input multi-output system model reflecting temperature coupling characteristics among temperature areas in a discrete domain; S22, constructing a decoupling neural network which sequentially comprises an input layer, a hidden layer and a decoupling output layer, wherein the total number of nodes of the input layer is 2n, the input is a real-time temperature sampling value of each temperature region and a temperature sampling value of the previous moment, the hidden layer adopts a sparse connection structure with a temperature coupling priori, only the connection between the corresponding nodes of the adjacent temperature regions is reserved, the connection weight of the corresponding nodes of the non-adjacent temperature regions is fixed to 0 and does not participate in training update, the nonlinear characteristic of temperature coupling is fitted through a hyperbolic tangent activation function, and a hidden layer vector is output, the total number of nodes of the decoupling output layer is n, firstly, a strict diagonal linear mapping which is output to the output layer by the hidden layer is established, trainable non-zero weight is reserved between the ith node of the hidden layer and the ith node of the output layer, the connection weight of other cross nodes is fixed to 0, and then limiting saturation activation processing with upper limit and lower limit is carried out on a linear mapping resu