Data-driven adaptive wall thickness control method and system for rocket box bottom mirror image milling

By measuring the wall thickness in real time on the mirror milling equipment and using data-driven adaptive PID control to dynamically adjust the processing parameters, the problem of machining accuracy and inefficiency in traditional mirror milling processes is solved, and the high accuracy of the rocket box bottom wall thickness is achieved. High efficiency processing.

CN118905724AActive Publication Date: 2024-11-08SHANGHAI JIAOTONG UNIV

Patent Information

Application Number
CN202411175758.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-26
Publication Date
2024-11-08
Estimated Expiration
2044-08-26

AI Technical Summary

Technical Problem

The traditional mirror milling process has problems such as difficult to ensure machining accuracy, low machining efficiency and lack of real-time feedback and control in rocket box bottom processing. Especially in the conditions of large diameter and thickness ratio and high material removal rate, fixed parameter design is difficult to adapt to power. Changes in the metaphysics lead to unstable processing quality.

Method used

Adaptive wall thickness control method driven by data is adopted, by equiping wall thickness measurement devices on mirror milling equipment, measuring workpiece wall thickness data in real time, and collecting the position data of the compensation axis through the CNC system, and real-time adjustment is performed using the preset PID algorithm. , dynamically adjust the processing parameters to achieve wall thickness compensation, and use time-varying PID parameters and linear weighted average to calculate the PID parameter set to achieve adaptive control of wall thickness.

Benefits of technology

Through the real-time data-driven control method, parameters can be adjusted dynamically according to the processing conditions, the processing quality and efficiency can be improved, and the wall thickness accuracy and production efficiency can be ensured, which solves the problems of instability and inefficiency caused by parameter fixation in traditional methods.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118905724A_ABST
    Figure CN118905724A_ABST
Patent Text Reader

Abstract

The invention provides a data-driven self-adaptive wall thickness control method and system for rocket box bottom mirror image milling, and the method comprises the steps: 1, installing a wall thickness measuring device at a supporting side of mirror image milling equipment, measuring the wall thickness data of a workpiece in a machining area in real time in a machining process, and collecting the position data of a compensation shaft in real time through a numerical control system; 2, transmitting the position data of the compensation shaft to a synchronous compensation program, and utilizing preset data to drive a PID (Proportion Integration Differentiation) algorithm to carry out wall thickness compensation; and 3, according to the wall thickness deviation, the parameters and the feed rate of the preset data driving PID algorithm are adjusted in real time, so that the actual effective cutting depth of single-step compensation is increased, and the deviation between the real-time wall thickness and the target wall thickness is reduced. By collecting and analyzing data in the machining process in real time and dynamically adjusting wall thickness control parameters, accurate, efficient and self-adaptive control over the bottom wall thickness of the rocket box is achieved, and therefore the wall thickness precision and the machining efficiency are improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of mirror milling, and in particular to a data-driven adaptive wall thickness control method and system for mirror milling of rocket box bottom. Background Art

[0002] At present, mirror milling technology has become a widely used processing method in the manufacturing process of rocket box bottom. This technology improves the rigidity of the part processing area by performing follow-up support at the mirror symmetrical position of the milling processing area, so as to improve the wall thickness accuracy and surface quality of the part. The traditional mirror milling process usually adopts a fixed processing parameter design, that is, the processing parameters of the part are determined in the tool path planning stage, and the cutting is strictly carried out according to this design during the processing. However, with the improvement of the diameter-to-thickness ratio and precision requirements of the rocket box bottom, the dynamic characteristics of the part in the processing process vary greatly with time and space. This fixed parameter design and processing method has certain limitations in actual operation.

[0003] Disadvantages of existing technology:

[0004] Processing accuracy is difficult to guarantee: Traditional fixed parameter designs may be affected by various factors during the processing process, such as material removal and different stiffness in different processing areas, resulting in the wall thickness error of the actual processed parts deviating from the design requirements, affecting the final quality of the parts.

[0005] Low processing efficiency: The traditional mirror milling process is difficult to adapt to parts with large diameter-to-thickness ratio and high material removal rate. In order to improve the wall thickness accuracy, it is necessary to increase the number of wall thickness compensations in the same processing area at the expense of feed speed, which increases processing time and reduces production efficiency.

[0006] Lack of real-time feedback and control: Existing processing methods usually lack data-driven real-time feedback mechanisms, making it difficult to respond quickly to changes in the processing process, resulting in unstable processing quality.

[0007] Patent document CN113909366A discloses a method for controlling the wall thickness of a Φ3350mm 2195 aluminum-lithium alloy integral bottom spinning part, which belongs to the technical field of tank shell manufacturing. The forming quality and wall thickness accuracy of the bottom spinning parts are improved by accurately designing the forming gap, accurately designing and controlling the spinning passes and the strong-normal spinning stroke, and adopting a forming process that alternates forward and reverse spinning and combines strong-normal spinning. Although the patent controls the wall thickness by accurately designing the spinning parameters and the forming process, this control method is based on fixed parameter settings and preset process steps. During the processing, once the workpiece material or processing conditions change, it may not be possible to adjust the control parameters in real time and dynamically, resulting in a decrease in processing quality. Compared with the present invention, this patent lacks a data-driven adaptive control mechanism. Summary of the invention

[0008] In view of the defects in the prior art, the object of the present invention is to provide a data-driven adaptive wall thickness control method and system for mirror milling of rocket box bottom.

[0009] The data-driven adaptive wall thickness control method for mirror milling of rocket box bottom provided by the present invention comprises:

[0010] Step 1: A wall thickness measuring device is installed on the support side of the mirror milling equipment to measure the wall thickness data of the workpiece in the processing area in real time during the processing, and the position data of the compensation axis is collected in real time through the CNC system;

[0011] Step 2: Transmit the position data of the compensation axis to the synchronous compensation program, and use the preset data to drive the PID algorithm to perform wall thickness compensation;

[0012] Step 3: According to the wall thickness deviation, the parameters and feed rate of the preset data-driven PID algorithm are adjusted in real time to increase the actual effective cutting depth of single-step compensation and reduce the deviation between the real-time wall thickness and the target wall thickness.

[0013] Preferably, the formula of the synchronous compensation procedure is:

[0014] y(t)=y(t-1)+u(t-1)+u(t-2)+ξ(t)

[0015] Where y is the wall thickness of the workpiece, u is the offset of the milling axis, and t is the time unit; ξ is a random variable used to simulate the wall thickness error caused by the contour error when machining according to the theoretical surface tool path and the random error caused by the normal deformation of the workpiece under the action of the axial cutting force. The value range of ξ is determined by the real-time wall thickness and single-step compensation in the machining data;

[0016] Replace the ordinary PID with the PID control rate with time-varying PID parameters shown in the following formula:

[0017] Δu(t)=K I (t)e(t)-K P (t)Δy(t)-K D (t)Δ 2 y(t)

[0018] Where u(t) is the control input, y(t) is the system output, e(t) = r(t) - y(t) is the error signal, r(t) is the reference signal, K P (t), K I (t), K D (t) are the time-varying P, I, and D parameters, respectively;

[0019] The time-varying PID parameters are adjusted according to the information vector, Where K(t) = [K P (t),K I (t),K D (t)], F() is a nonlinear function; the information vector r(t),y(t),…,y(tn y +1),u(t-1),…,u(tn u +1)] contains the input and output data pairs of the controlled object, n y ,n u are the orders of output and input respectively.

[0020] Preferably, the process of calculating PID parameters is:

[0021] The PID parameters determined based on the experience of field engineers are used as the initial database, and the expression is:

[0022]

[0023] Where N(0) represents the number of information vectors stored in the initial database;

[0024] The calculation of distance and selection of information vector are completed within the sampling time. The distance expression between the data at the current moment and the information vector in the database is:

[0025]

[0026] Among them, N(t) represents the number of information vectors stored in the database at the current moment, represents the lth element of the jth information vector, Indicates the lth element of the query data at the current moment, represents the maximum value among the lth elements of all information vectors stored in the database, represents the minimum value of the lth element of all information vectors stored in the database; selects k groups of data with the smallest distance to calculate PID parameters;

[0027] Calculate the PID parameters, use the k groups of data with the smallest distance, and use the linear weighted average calculation shown in the following formula to query the appropriate PID parameter set:

[0028]

[0029] Among them, w i is the weight value corresponding to the i-th information vector, calculated by the following formula:

[0030]

[0031] Preferably, the PID parameter adjustment is completed within the sampling time, specifically:

[0032] The steepest descent method is used to modify the PID parameters. The formula is:

[0033]

[0034] η=diag{η P ,η I ,η D}

[0035] Among them, η is the learning rate, J() is the error function, and the expressions of J and ε are as follows:

[0036]

[0037] ε(t)=r(t)-y(t).

[0038] Preferably, a preset algorithm is used to delete redundant data, and the algorithm implementation process is as follows:

[0039] Extract the information vector that satisfies the following conditions from other information vectors that exclude k groups of data in advance:

[0040]

[0041] From the information vectors that satisfy the above conditions, extract the information vector that satisfies the following conditions:

[0042]

[0043] Among them, K1, K2, and K3 represent K P , K I , K D ;

[0044] The above process is achieved by extracting and deleting information vectors from the database that are highly similar to the data contained in the newly generated data set.

[0045] The data-driven adaptive wall thickness control system for mirror milling of rocket box bottom provided by the present invention comprises:

[0046] Module M1: A wall thickness measuring device is installed on the support side of the mirror milling equipment to measure the wall thickness data of the workpiece in the processing area during the processing in real time, and collect the position data of the compensation axis in real time through the CNC system;

[0047] Module M2: Transmits the position data of the compensation axis to the synchronous compensation program, and uses the preset data to drive the PID algorithm for wall thickness compensation;

[0048] Module M3: Adjust the parameters and feed rate of the preset data-driven PID algorithm in real time according to the wall thickness deviation, thereby increasing the actual effective cutting depth of single-step compensation and reducing the deviation between the real-time wall thickness and the target wall thickness.

[0049] Preferably, the formula of the synchronous compensation procedure is:

[0050] y(t)=y(t-1)+u(t-1)+u(t-2)+ξ(t)

[0051] Where y is the wall thickness of the workpiece, u is the offset of the milling axis, and t is the time unit; ξ is a random variable used to simulate the wall thickness error caused by the contour error when machining according to the theoretical surface tool path and the random error caused by the normal deformation of the workpiece under the action of the axial cutting force. The value range of ξ is determined by the real-time wall thickness and single-step compensation in the machining data;

[0052] Replace the ordinary PID with the PID control rate with time-varying PID parameters shown in the following formula:

[0053] Δu(t)=K I (t)e(t)-K P (t)Δy(t)-K D (t)Δ 2 y(t)

[0054] Where u(t) is the control input, y(t) is the system output, e(t) = r(t) - y(t) is the error signal, r(t) is the reference signal, K P (t), K I (t), K D (t) are the time-varying P, I, and D parameters, respectively;

[0055] The time-varying PID parameters are adjusted according to the information vector, Where K(t) = [K P (t),K I (t),K D (t)], F() is a nonlinear function; the information vector Contains the input and output data pairs of the controlled object, n y ,n u are the orders of output and input respectively.

[0056] Preferably, the process of calculating PID parameters is:

[0057] The PID parameters determined based on the experience of field engineers are used as the initial database, and the expression is:

[0058]

[0059] Where N(0) represents the number of information vectors stored in the initial database;

[0060] The calculation of distance and selection of information vector are completed within the sampling time. The distance expression between the data at the current moment and the information vector in the database is:

[0061]

[0062] Among them, N(t) represents the number of information vectors stored in the database at the current moment, represents the lth element of the jth information vector, Indicates the lth element of the query data at the current moment, represents the maximum value among the lth elements of all information vectors stored in the database, represents the minimum value of the lth element of all information vectors stored in the database; selects k groups of data with the smallest distance to calculate PID parameters;

[0063] Calculate the PID parameters, use the k groups of data with the smallest distance, and use the linear weighted average calculation shown in the following formula to query the appropriate PID parameter set:

[0064]

[0065] Among them, w i is the weight value corresponding to the i-th information vector, calculated by the following formula:

[0066]

[0067] Preferably, the PID parameter adjustment is completed within the sampling time, specifically:

[0068] The steepest descent method is used to modify the PID parameters. The formula is:

[0069]

[0070] η=diag{η P ,η I ,η D}

[0071] Among them, η is the learning rate, J() is the error function, and the expressions of J and ε are as follows:

[0072]

[0073] ε(t)=r(t)-y(t).

[0074] Preferably, a preset algorithm is used to delete redundant data, and the algorithm implementation process is:

[0075] Extract the information vector that satisfies the following conditions from other information vectors that exclude k groups of data in advance:

[0076]

[0077] From the information vectors that satisfy the above conditions, extract the information vector that satisfies the following conditions:

[0078]

[0079] Among them, K1, K2, and K3 represent K P , K I , K D ;

[0080] The above process is achieved by extracting and deleting information vectors from the database that are highly similar to the data contained in the newly generated data set.

[0081] Compared with the prior art, the present invention has the following beneficial effects:

[0082] (1) This method provides data support for the data-driven control method by collecting and analyzing the data in the processing process in real time. The processing parameters are dynamically adjusted based on the results of data processing to achieve adaptive control of wall thickness. This adaptive control strategy can make timely adjustments according to changes in the processing conditions to ensure the processing quality of the processing process.

[0083] (2) By adopting a data-driven adaptive wall thickness control method, the present invention successfully solves the problems of unstable processing quality and low processing efficiency caused by fixed processing parameters and reliance on experience in the traditional processing process, and improves the processing quality and processing efficiency by optimizing the processing parameters in real time. BRIEF DESCRIPTION OF THE DRAWINGS

[0084] Other features, objects and advantages of the present invention will become more apparent from the detailed description of non-limiting embodiments made with reference to the following drawings:

[0085] Figure 1 The basic architecture diagram of the mirror milling data-driven adaptive PID control method. DETAILED DESCRIPTION

[0086] The present invention is described in detail below in conjunction with specific embodiments. The following embodiments will help those skilled in the art to further understand the present invention, but are not intended to limit the present invention in any form. It should be noted that, for those of ordinary skill in the art, several changes and improvements can also be made without departing from the concept of the present invention. These all belong to the protection scope of the present invention.

[0087] Example 1

[0088] The present invention provides a data-driven adaptive wall thickness control method for mirror milling of rocket box bottom, and the specific technical scheme is as follows:

[0089] 1) A wall thickness measuring device is installed on the support side of the mirror milling equipment to measure the wall thickness data of the workpiece in the processing area during the processing in real time, and the position data of the compensation axis is collected in real time through the CNC system;

[0090] 2) The collected data is transmitted to the wall thickness compensation model and the data is processed using a preset data-driven PID algorithm;

[0091] The wall thickness compensation model is:

[0092] y(t)=y(t-1)+u(t-1)+u(t-2)+ξ(t)

[0093] Among them, y is the wall thickness of the workpiece, u is the milling axis offset, and ξ is a random variable, which is used to simulate the random error caused by factors such as the wall thickness error caused by the contour error during theoretical surface tool path processing and the normal deformation of the workpiece under the action of axial cutting force. The value range of ξ is determined by the real-time wall thickness and single-step compensation amount in the processing data.

[0094] Replace the ordinary PID with a PID control rate with time-varying PID parameters as shown in the following formula:

[0095] Δu(t)=K I (t)e(t)-K P (t)Δy(t)-K D (t)Δ 2 y(t)

[0096] Where u(t) is the control input, y(t) is the system output, e(t) = r(t) - y(t) is the error signal, r(t) is the reference signal, K P (t), K I (t), K D (t) are the time-varying P, I, and D parameters respectively.

[0097] Information Vector Contains the input and output data pairs of the controlled object, n y ,n u are the orders of output and input respectively, and the time-varying PID parameters are adjusted according to the information vector, namely Where K(t) = [K P (t),K I (t),K D (t)], F() is a nonlinear function.

[0098] The controller design process is mainly divided into the following steps:

[0099] (1) Generate the initial database:

[0100] The PID parameters determined based on the experience of the field engineer are used as the initial database, that is, the initial database shown in the following formula is generated:

[0101]

[0102] Wherein, N(0) represents the number of information vectors stored in the initial database.

[0103] (2) Calculate the distance and select the information vector:

[0104] The distance between the current data and the information vector in the database is as follows:

[0105]

[0106] Among them, N(t) represents the number of information vectors stored in the database at the current moment, represents the lth element of the jth information vector, Indicates the lth element of the query data at the current moment, represents the maximum value among the lth elements of all information vectors stored in the database, It represents the minimum value among the lth elements of all information vectors stored in the database. Then, k groups of data with the smallest distance are selected to calculate the PID parameters.

[0107] (3) Calculate PID parameters:

[0108] Using the k sets of data selected in step (2), the appropriate PID parameter set is queried using the linear weighted average calculation shown in the following formula:

[0109]

[0110] Among them, w i is the weight value corresponding to the i-th information vector, which can be calculated by the following formula:

[0111]

[0112] (4) PID parameter adjustment:

[0113] When the PID parameters calculated in step (3) cannot achieve suitable control performance, these control parameters need to be updated and stored in the database, that is, the PID parameters need to be adjusted to reduce the control error. In order to modify the PID parameters, the steepest descent method is used as shown below:

[0114]

[0115] η=diag{η P ,η I ,η D}

[0116] Among them, η is the learning rate, J() is the error function, and the expressions of J and ε are as follows:

[0117]

[0118] ε(t)=r(t)-y(t)

[0119] (5) Delete redundant data:

[0120] When implementing the proposed data-driven method in a real system, there is a constraint that the calculation between steps (2) and (4) must be completed within the sampling time, and storing redundant data in the database will cause the computer to waste time that could be used for the process. Therefore, in order to avoid storing too much data, an algorithm is used to remove redundant data. The algorithm is divided into the following two steps:

[0121] a. Extract the information vector that satisfies the following conditions from other information vectors that exclude k groups of data in advance:

[0122]

[0123] b. From the information vectors that satisfy the first condition, extract the information vector that satisfies the following conditions:

[0124]

[0125] Among them, K1, K2, and K3 represent K P , K I , K d The above process is achieved by extracting and deleting information vectors from the database that are highly similar to the data contained in the newly generated data set.

[0126] By processing the input and output data, the PID parameters for calculating the cutting depth can be adjusted synchronously in real time to achieve adaptive control of the wall thickness.

[0127] 3) By optimizing the processing parameters, that is, adjusting the PID parameters and feed rate of the cutting depth calculation in real time according to the above-mentioned control method and online processing data such as wall thickness deviation, the actual effective cutting depth of the single-step compensation can be increased to improve the efficiency of the wall thickness single-step compensation, and the deviation between the real-time wall thickness and the target wall thickness can be quickly reduced to improve the wall thickness accuracy. It can also reduce the frequency of reducing the feed speed due to excessive wall thickness deviation in the synchronous compensation program, thereby indirectly improving the processing efficiency.

[0128] 4) In order to facilitate the operation and management of operators, the method also provides a friendly user interface and operation interface. Operators can view processing data and processing parameters and monitor the processing process through the interface.

[0129] Here is one possible implementation:

[0130] High-precision sensors are installed on processing equipment to monitor key parameters in the processing process in real time, such as workpiece wall thickness, cutting force, cutting temperature, etc. These sensors transmit the collected data in real time to the control system.

[0131] After receiving the data from each sensor, the control system processes it in real time. Through the above data-driven PID control method, the control system can use adaptive control strategies to adjust the processing parameters according to the state changes in the processing process, such as changes in the wall thickness of the workpiece and fluctuations in the cutting axis. By dynamically adjusting the cutting depth PID parameters, cutting speed and other parameters, high-precision and efficient control of the wall thickness can be achieved.

[0132] At the same time, the control system also has the ability to self-learn and optimize. By continuously collecting and analyzing processing data, the control system can gradually optimize the control strategy and parameter settings to improve processing quality and efficiency.

[0133] Since the mirror milling mechanical processing system is relatively complex, it is difficult to accurately model it. In addition, since the processing object of the mirror milling is a thin-walled curved surface part with extremely poor rigidity, it is very easy to produce deformation induced by cutting force and supporting force during the processing, resulting in the dynamic parameters of the entire system constantly changing with time and space, further increasing the difficulty of controlling the mirror milling system. These factors lead to poor results of traditional control methods. In order to solve the problem that the mirror milling system is difficult to accurately model and the dynamic parameters change with time and space, resulting in poor control effects of traditional model-based control methods, the present invention adopts a data-driven adaptive PID control method. The present invention mainly utilizes the control method's ability to learn and adaptively adjust control parameters under different dynamic characteristics of the system to improve controller performance. Figure 1 The basic architecture of the mirror milling data-driven adaptive PID control method is shown.

[0134] Example 2

[0135] The present invention also provides a data-driven adaptive wall thickness control system for mirror milling of rocket box bottom. The data-driven adaptive wall thickness control system for mirror milling of rocket box bottom can be realized by executing the process steps of the data-driven adaptive wall thickness control method for mirror milling of rocket box bottom. That is, those skilled in the art can understand the data-driven adaptive wall thickness control method for mirror milling of rocket box bottom as a preferred implementation of the data-driven adaptive wall thickness control system for mirror milling of rocket box bottom.

[0136] The data-driven adaptive wall thickness control system for mirror milling of rocket box bottom provided by the present invention comprises:

[0137] Module M1: A wall thickness measuring device is installed on the support side of the mirror milling equipment to measure the wall thickness data of the workpiece in the processing area during the processing in real time, and collect the position data of the compensation axis in real time through the CNC system;

[0138] Module M2: Transmits the position data of the compensation axis to the synchronous compensation program, and uses the preset data to drive the PID algorithm for wall thickness compensation;

[0139] Module M3: Adjust the parameters and feed rate of the preset data-driven PID algorithm in real time according to the wall thickness deviation, thereby increasing the actual effective cutting depth of single-step compensation and reducing the deviation between the real-time wall thickness and the target wall thickness.

[0140] The formula of the synchronous compensation procedure is:

[0141] y(t)=y(t-1)+u(t-1)+u(t-2)+ξ(t)

[0142] Where y is the wall thickness of the workpiece, u is the offset of the milling axis, and t is the time unit; ξ is a random variable used to simulate the wall thickness error caused by the contour error when machining according to the theoretical surface tool path and the random error caused by the normal deformation of the workpiece under the action of the axial cutting force. The value range of ξ is determined by the real-time wall thickness and single-step compensation in the machining data;

[0143] Replace the ordinary PID with the PID control rate with time-varying PID parameters shown in the following formula:

[0144] Δu(t)=K I (t)e(t)-K P (t)Δy(t)-K D (t)Δ 2 y(t)

[0145] Where u(t) is the control input, y(t) is the system output, e(t) = r(t) - y(t) is the error signal, r(t) is the reference signal, K P (t), K I (t), K D (t) are the time-varying P, I, and D parameters, respectively;

[0146] The time-varying PID parameters are adjusted according to the information vector, Where K(t) = [K P (t),K I (t),K D (t)], F() is a nonlinear function; the information vector Contains the input and output data pairs of the controlled object, n y ,n u are the orders of output and input respectively.

[0147] The process of calculating PID parameters is:

[0148] The PID parameters determined based on the experience of field engineers are used as the initial database, and the expression is:

[0149]

[0150] Where N(0) represents the number of information vectors stored in the initial database;

[0151] The calculation of distance and selection of information vector are completed within the sampling time. The distance expression between the data at the current moment and the information vector in the database is:

[0152]

[0153] Among them, N(t) represents the number of information vectors stored in the database at the current moment, represents the lth element of the jth information vector, Indicates the lth element of the query data at the current moment, represents the maximum value among the lth elements of all information vectors stored in the database, represents the minimum value of the lth element of all information vectors stored in the database; selects k groups of data with the smallest distance to calculate PID parameters;

[0154] Calculate the PID parameters, use the k groups of data with the smallest distance, and use the linear weighted average calculation shown in the following formula to query the appropriate PID parameter set:

[0155]

[0156] Among them, w i is the weight value corresponding to the i-th information vector, calculated by the following formula:

[0157]

[0158] PID parameter adjustment is completed within the sampling time, specifically:

[0159] The steepest descent method is used to modify the PID parameters. The formula is:

[0160]

[0161] η=diag{η P ,η I ,ηD}

[0162] Among them, η is the learning rate, J() is the error function, and the expressions of J and ε are as follows:

[0163]

[0164] ε(t)=r(t)-y(t).

[0165] Use a preset algorithm to delete redundant data. The algorithm implementation process is as follows:

[0166] Extract the information vector that satisfies the following conditions from other information vectors that exclude k groups of data in advance:

[0167]

[0168] From the information vectors that satisfy the above conditions, extract the information vector that satisfies the following conditions:

[0169]

[0170] Among them, K1, K2, and K3 represent K P , K I , K D ;

[0171] The above process is achieved by extracting and deleting information vectors from the database that are highly similar to the data contained in the newly generated data set.

[0172] Those skilled in the art know that, in addition to implementing the system, device and its various modules provided by the present invention in a purely computer-readable program code, it is entirely possible to implement the same program in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers and embedded microcontrollers by logically programming the method steps. Therefore, the system, device and its various modules provided by the present invention can be considered as a hardware component, and the modules included therein for implementing various programs can also be considered as structures within the hardware component; the modules for implementing various functions can also be considered as both software programs for implementing the method and structures within the hardware component.

[0173] The above describes the specific embodiments of the present invention. It should be understood that the present invention is not limited to the above specific embodiments, and those skilled in the art can make various changes or modifications within the scope of the claims, which does not affect the essence of the present invention. In the absence of conflict, the embodiments of the present application and the features in the embodiments can be combined with each other arbitrarily.

Claims

1. A data-driven adaptive wall thickness control method for mirror milling of rocket box bottom, characterized in that: include: Step 1: A wall thickness measuring device is installed on the support side of the mirror milling equipment to measure the wall thickness data of the workpiece in the processing area in real time during the processing, and the position data of the compensation axis is collected in real time through the CNC system; Step 2: Transmit the position data of the compensation axis to the synchronous compensation program, and use the preset data to drive the PID algorithm to perform wall thickness compensation; Step 3: According to the wall thickness deviation, the parameters and feed rate of the preset data-driven PID algorithm are adjusted in real time to increase the actual effective cutting depth of single-step compensation and reduce the deviation between the real-time wall thickness and the target wall thickness.

2. The data-driven adaptive wall thickness control method for rocket box bottom mirror milling according to claim 1 is characterized in that: The formula of the synchronous compensation procedure is: y(t)=y(t-1)+u(t-1)+u(t-2)+ξ(t) Where y is the wall thickness of the workpiece, u is the offset of the milling axis, and t is the time unit; ξ is a random variable used to simulate the wall thickness error caused by the contour error when machining according to the theoretical surface tool path and the random error caused by the normal deformation of the workpiece under the action of the axial cutting force. The value range of ξ is determined by the real-time wall thickness and single-step compensation in the machining data; Replace the ordinary PID with the PID control rate with time-varying PID parameters shown in the following formula: Δu(t)=K I (t)e(t)-K P (t)Δy(t)-K D (t)Δ 2 y(t) Where u(t) is the control input, y(t) is the system output, e(t) = r(t) - y(t) is the error signal, r(t) is the reference signal, K P (t), K I (t), K D (t) are the time-varying P, I, and D parameters, respectively; The time-varying PID parameters are adjusted according to the information vector, Where K(t) = [K P (t),K I (t),K D (t)], F() is a nonlinear function; the information vector Contains the input and output data pairs of the controlled object, n y ,n u are the orders of output and input respectively.

3. The data-driven adaptive wall thickness control method for rocket box bottom mirror milling according to claim 2 is characterized in that: The process of calculating PID parameters is: The PID parameters determined based on the experience of field engineers are used as the initial database, and the expression is: Where N(0) represents the number of information vectors stored in the initial database; The calculation of distance and selection of information vector are completed within the sampling time. The distance expression between the data at the current moment and the information vector in the database is: Among them, N(t) represents the number of information vectors stored in the database at the current moment, represents the lth element of the jth information vector, Indicates the lth element of the query data at the current moment, represents the maximum value among the lth elements of all information vectors stored in the database, represents the minimum value of the lth element of all information vectors stored in the database; selects k groups of data with the smallest distance to calculate PID parameters; Calculate the PID parameters, use the selected k groups of data with the smallest distance, and use the linear weighted average calculation shown in the following formula to query the appropriate PID parameter set: Among them, w i is the weight value corresponding to the i-th information vector, calculated by the following formula:

4. The data-driven adaptive wall thickness control method for rocket box bottom mirror milling according to claim 3 is characterized in that: PID parameter adjustment is completed within the sampling time, specifically: The steepest descent method is used to modify the PID parameters. The formula is: η=diag{η P ,or I ,or D } Among them, η is the learning rate, J() is the error function, and the expressions of J and ε are as follows: ε(t)=r(t)-y(t).

5. The data-driven adaptive wall thickness control method for rocket box bottom mirror milling according to claim 4 is characterized in that: Use a preset algorithm to delete redundant data. The algorithm implementation process is as follows: Extract the information vector that satisfies the following conditions from other information vectors that exclude k groups of data in advance: From the information vectors that satisfy the above conditions, extract the information vector that satisfies the following conditions: Among them, K1, K2, and K3 represent K P , K I , K D ; The above process is achieved by extracting and deleting information vectors from the database that are highly similar to the data contained in the newly generated data set.

6. A data-driven adaptive wall thickness control system for mirror milling of rocket box bottom, characterized in that: include: Module M1: A wall thickness measuring device is installed on the support side of the mirror milling equipment to measure the wall thickness data of the workpiece in the processing area during the processing in real time, and collect the position data of the compensation axis in real time through the CNC system; Module M2: Transmits the position data of the compensation axis to the synchronous compensation program, and uses the preset data to drive the PID algorithm for wall thickness compensation; Module M3: Adjust the parameters and feed rate of the preset data-driven PID algorithm in real time according to the wall thickness deviation, thereby increasing the actual effective cutting depth of single-step compensation and reducing the deviation between the real-time wall thickness and the target wall thickness.

7. The data-driven adaptive wall thickness control system for mirror milling of rocket box bottom according to claim 6 is characterized in that: The formula of the synchronous compensation procedure is: y(t)=y(t-1)+u(t-1)+u(t-2)+ξ(t) Where y is the wall thickness of the workpiece, u is the offset of the milling axis, and t is the time unit; ξ is a random variable used to simulate the wall thickness error caused by the contour error when machining according to the theoretical surface tool path and the random error caused by the normal deformation of the workpiece under the action of the axial cutting force. The value range of ξ is determined by the real-time wall thickness and single-step compensation in the machining data; Replace the ordinary PID with the PID control rate with time-varying PID parameters shown in the following formula: Δu(t)=K I (t)e(t)-K P (t)Δy(t)-K D (t)Δ 2 y(t) Where u(t) is the control input, y(t) is the system output, e(t) = r(t) - y(t) is the error signal, r(t) is the reference signal, K P (t), K I (t), K D (t) are the time-varying P, I, and D parameters, respectively; The time-varying PID parameters are adjusted according to the information vector, Where K(t) = [K P (t),K I (t),K D (t)], F() is a nonlinear function; the information vector Contains the input and output data pairs of the controlled object, n y ,n u are the orders of output and input respectively.

8. The data-driven adaptive wall thickness control system for mirror milling of rocket box bottom according to claim 7 is characterized in that: The process of calculating PID parameters is: The PID parameters determined based on the experience of field engineers are used as the initial database, and the expression is: Where N(0) represents the number of information vectors stored in the initial database; The calculation of distance and selection of information vector are completed within the sampling time. The distance expression between the data at the current moment and the information vector in the database is: Among them, N(t) represents the number of information vectors stored in the database at the current moment, represents the lth element of the jth information vector, Indicates the lth element of the query data at the current moment, represents the maximum value among the lth elements of all information vectors stored in the database, represents the minimum value of the lth element of all information vectors stored in the database; selects k groups of data with the smallest distance to calculate PID parameters; Calculate the PID parameters, use the selected k groups of data with the smallest distance, and use the linear weighted average calculation shown in the following formula to query the appropriate PID parameter set: Among them, w i is the weight value corresponding to the i-th information vector, calculated by the following formula:

9. The data-driven adaptive wall thickness control system for mirror milling of rocket box bottom according to claim 8 is characterized in that: PID parameter adjustment is completed within the sampling time, specifically: The steepest descent method is used to modify the PID parameters. The formula is: η=diag{η P ,or I ,or D } Among them, η is the learning rate, J() is the error function, and the expressions of J and ε are as follows: ε(t)=r(t)-y(t).

10. The data-driven adaptive wall thickness control system for mirror milling of rocket box bottom according to claim 9 is characterized in that: Use a preset algorithm to delete redundant data. The algorithm implementation process is as follows: Extract the information vector that satisfies the following conditions from other information vectors that exclude k groups of data in advance: From the information vectors that satisfy the above conditions, extract the information vector that satisfies the following conditions: Among them, K1, K2, and K3 represent K P , K I , K D ; The above process is achieved by extracting and deleting information vectors from the database that are highly similar to the data contained in the newly generated data set.

Citation Information

Patent Citations

  • Method for controlling wall thickness of phi 3350 mm 2195 aluminum lithium alloy integral box bottom spinning part

    CN113909366A

  • Double five-axis mirror image synchronous machining system and mirror image synchronous compensation milling method

    CN110560758A

  • Alignment method applied to box bottom scalloped segment self-adaptive compensation machining

    CN114054816A

  • Alignment method for self-adaptive compensation machining of overall box bottom with ultra-large diameter-thickness ratio

    CN115647438A

  • Equal-wall-thickness precision machining method for whole box bottom with ultra-large diameter-thickness ratio

    CN115647744A

Cited By

  • Integral box bottom mirror image milling quality analysis method

    CN119149958A

  • Intelligent control system and full-automatic edge milling machine thereof

    CN119927291A

  • Equal-wall-thickness processing device for large-diameter seamless stainless steel pipe

    CN120696487A

  • Device for equalizing wall thickness of large-diameter seamless stainless steel pipe

    CN120696487B

  • Bidirectional cooperation and adaptive control method and system driven by wall thickness follow-up measurement data

    CN121199764A