Cable displacement prediction system of self-adaptive code spraying control system
By applying the ANFIS algorithm to replace the traditional fuzzy algorithm in the adaptive inkjet control system, the cable scrapping problem caused by the incoordination of the production rhythm of the inkjet printer and the crimping machine is solved, and the accuracy of the prediction of cable displacement is improved, achieving a more efficient coding change process and better control effect.
Patent Information
- Application Number
- CN202411770187.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-04
- Publication Date
- 2025-03-18
AI Technical Summary
In the fully automatic terminal crimping production line, the production rhythm of the inkjet printer and the crimping machine is not coordinated, resulting in the cable being printed and invalid character identification, causing the cable to be scrapped, and the use of a preset fuzzy prediction algorithm results in a large deviation range of the predicted value of the cable displacement, reducing the efficiency of transcoding.
The ANFIS algorithm is used instead of the traditional fuzzy algorithm to predict cable displacement. By analyzing the cable cutting environment and the working condition of the inkjet printer, the input feature parameters affecting the inkjet character identification printing effect are distinguished, the input feature parameters are trained, and the cable displacement is predicted using the ANFIS model.
After the model training is completed, the prediction time is greatly shortened, and the training and prediction process can be accelerated through parallel calculations, which reduces the prediction deviation value, thereby improving the prediction accuracy of the cable displacement prediction model, improving the code transfer efficiency, and reducing the deformation of the character identification after switching.
Smart Images
Figure CN119644734A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of mechanical processing and automated production control, and relates to a cable displacement prediction system for an adaptive inkjet coding control system. Background Art
[0002] With the continuous development of automobile wiring harness identification methods, inkjet coding is used by more and more automobile wiring harness companies. However, when this identification method is applied to the fully automatic terminal crimping production line, due to the special structural requirements of the production line and the control mode of the inkjet printer, the process of connecting any two groups of cables will cause the problem of incoordination between the production rhythm of the inkjet printer and the fully automatic terminal crimping machine, which will cause a section of cable to be printed with invalid character identification, eventually leading to the scrapping of the cable. The long-term existence of this problem has caused serious economic losses and waste of resources to related companies. In order to solve the problem of cable loss in enterprises, an adaptive inkjet control system has been developed to realize the function of adaptively switching the printing characters at the connection position of adjacent cable groups, that is, adjusting the production rhythm of the inkjet printer to synchronize it with the crimping machine. The preset fuzzy prediction algorithm and decision control algorithm then ensure that the position of the character identification after switching meets the requirements of the production process specifications. The advantage of this system is that the function of adaptively switching the printing identification of the inkjet printer can be realized without making major changes to the original production line composition. However, the use of the preset fuzzy prediction algorithm will result in a large deviation range of the predicted value of the cable displacement during the character switching process, which reduces the efficiency of transcoding. Summary of the invention
[0003] In order to make up for the deficiencies of the prior art, the present invention provides a cable displacement prediction system for an adaptive inkjet control system. According to the system characteristics, a method for predicting cable displacement using the ANFIS algorithm instead of the traditional fuzzy algorithm is proposed, so as to reduce the deformation of the character representation and improve the printing effect of the characters after switching. The present invention is achieved through the following technical scheme: The present invention provides a cable displacement prediction system for an adaptive inkjet control system, including analyzing the cable cutting environment and the working condition of the inkjet printer, distinguishing the input feature parameters that affect the printing effect of the inkjet printer character identification, training the input feature parameter recognition, and screening out a screening model for factors affecting the prediction of cable displacement from previous character displacement data; the filtered data is input into the ANFIS model, and the model is trained; The ANFIS model structure includes five layers: input layer, fuzzification layer, rule layer, normalization layer, and output layer; Input Layer: The input variables are: x is the input variable, and j is the number of input variables; Fuzzy layer: Fuzzy rules are used to calculate the membership of each input variable. This system uses Gaussian function as the membership function. Assume that the membership function is ,but in is the center of the membership function, is the standard deviation; n is the input parameter; k is the number of fuzzy subsets; Rules layer: At this level, IF-THEN rules based on fuzzy logic are used to form fuzzy reasoning; assuming that The form of the rule is: rule :if yes and yes And... and yes ,but ; in, is the fuzzy set of the fuzzy system; is the fuzzy system parameter; The output obtained according to the fuzzy rule, the input part, that is, the if part, is fuzzy, and the output part, that is, the then part, is certain. The fuzzy reasoning indicates that the output is a linear combination of the input; The excitation strength of each rule is determined by the membership of the input variable: represents the excitation intensity; Normalization layer: In this layer, the excitation intensity of each rule is normalized so that its sum is 1; the normalized excitation intensity for: Output layer: Calculate the output of each rule and perform a weighted sum of the outputs of all rules to get the final system output: The final output is the weighted sum of all rule outputs.
[0004] The beneficial effects of the present invention are: In the production of wire harnesses, the traditional fuzzy algorithm is not real-time enough in predicting the displacement of cables. After the model training is completed, the prediction time of the ANFIS algorithm will be greatly shortened, and the training and prediction process can be accelerated through parallel computing. When the prediction deviation value is 3mm, the character deformation is 20%; when the prediction deviation value is 1mm, the character deformation is 13.3%. Therefore, improving the prediction accuracy of the cable displacement prediction model is conducive to improving the efficiency of code switching, while reducing the deformation of the character logo after switching, so that the system presents a better control effect. BRIEF DESCRIPTION OF THE DRAWINGS
[0005] The present invention will be further described below in conjunction with the accompanying drawings; Figure 1 is the input membership function diagram of the present invention; Figure 2 is the output membership function diagram of the present invention; Figure 3 The influence of the cable displacement prediction deviation on the marked deformation of the present invention; Figure 4 This is the ANFIS model structure diagram of the present invention. DETAILED DESCRIPTION
[0006] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below in conjunction with the accompanying drawings. In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without violating the connotation of the present invention, so the present invention is not limited by the specific embodiments disclosed below.
[0007] The accompanying drawing is a specific embodiment of the cable displacement prediction system of the adaptive inkjet control system of the present invention. Specific embodiment: During the production line debugging process, no cable group processing experiments are set for specific specifications and specific lengths. Instead, the company's production scheduling needs are used as test conditions. This does not affect the company's normal production, and is also conducive to verifying the versatility of the system's effects.
[0009] Description of the basic experimental protocol: With a sampling period of 9.5ms, the process data of 1,200 cable group processing connections under random cable specifications and random processing lengths (determined by the production line production requirements) are recorded.
[0010] In general, the speed of the cable processing process is the result of the combined effect of the servo motor control algorithm and the system disturbance (cable surface roughness, bending degree). Since the specific speed control algorithm of the motor is unknown and the system disturbance is difficult to quantitatively calculate, the cable drive system can be considered as a typical fuzzy system. First, the traditional fuzzy prediction algorithm is used to predict the cable displacement within the future command response time. It is a prediction method based on fuzzy set theory, fuzzy language variables and fuzzy logic reasoning, which imitates the fuzzy reasoning and decision-making process of people in terms of behavior.
[0011] The general steps of applying this algorithm are: first determine the input and output of the fuzzy prediction model, then formulate fuzzy rules based on the experience of operators or experts, then fuzzify the model input signal according to real-time sampling data, then complete the fuzzy reasoning process according to the fuzzy rules, obtain the fuzzified output value, and finally denormalize the output value.
[0012] The cable displacement prediction model has three inputs and one output. Input one is the current cable velocity value, which is expressed as ; Input 2 is the absolute value of the cable acceleration at the current moment, recorded as ; Input three is the distance between the current position and the end position of the next character mark, recorded as ; The output is the displacement of the cable within the command response time, recorded as u. Based on the sampled data during the cable processing process, the input is determined The domain of , in m / s; input The domain of is {-5,-2.5,0,2.5,5}, and the unit is ;enter The domain of is {0, 0.015, 0.03, 0.045}, and the unit is m; the domain of output u is {0, 0.016, 0.032, 0.048}, and the unit is m. To analyze the sampled data, the model uses a triangular membership function.
[0013] Among them, μ(x) represents the element x Membership degree: When x is between the interval [a, b), the membership increases from 0 to 1; When x is between the interval [b,c), the membership decreases from 1 to 0; When x is less than a or greater than c, the membership is 0.
[0014] The input and output membership functions are as follows: Figure 1 , Figure 2 shown.
[0015] In the membership function diagram, both input and output are divided into four semantic variables: small, medium, large, and maximum. Based on the sampling data in the cable processing process, a total of 64 fuzzy rules were formulated. The fuzzy reasoning adopts the MAX-MIN method. The reasoning process is: find the effective fuzzy rules according to the fuzzy rule base, and determine the membership function value of the rule in the output by taking the smaller membership function value of the input in each domain. In the same output domain of different rules, the larger membership function value is taken as the fuzzy output of the domain. After obtaining the fuzzy output, the result is defuzzified by the centroid method, and the measurement of the cable displacement in the next instruction response cycle can be obtained. The centroid method takes the centroid of the area enclosed by the membership function curve and the horizontal axis as the final output value of the fuzzy reasoning, and the output is calculated by the following formula. This method has a smoother output reasoning control. Even in response to a slight change in the input signal, the output will change.
[0016] is the center of gravity of the fuzzy output (i.e., output value); u is the value of the fuzzy set (the range of the output variable); is the membership function of the fuzzy output set.
[0017] Test results and analysis: The test sample contains 1200 sets of data. The cables processed in this process include For all specifications, the character width is set to 10mm. The printing effect of the new character after switching is shown in Table 1-1.
[0018] Table 1-1 Actual production line results Among the 1200 test data sets, there were 838 sets of character logos with no deformation, and 362 sets of character logos with varying degrees of deformation and distortion. Among the 838 sets of data with no deformation, 2 sets of switching failures were both position failures, that is, the character logo printing range exceeded the 40mm range of the cable end. Among the 362 sets of data with no deformation, 1 set of position failure, 6 sets of excessive stretching failures, that is, the actual total length of the character logo was more than 20% longer than the standard total length of the character logo, and 4 sets of excessive compression failures, that is, the actual total length of the character logo was more than 20% shorter than the standard total length of the character logo. Among them, the maximum stretching amount was 28.6%, and the maximum compression amount was 25.3%. It can be seen from the actual measurement that the character logo with a deformation of more than 20% has obviously affected the normal recognition. From the 1200 sets of data sampled from the actual production line, the deviation range of the predicted value of the cable displacement during the character switching process is 1-3mm.
[0019] The influence of prediction value deviation on character deformation is as follows Figure 3 As shown. When the predicted position of the character mark after switching is at the position shown in Figure a), considering the influence of the predicted deviation value x, it is necessary to adjust the character mark to the position shown in Figure b) through the algorithm. According to the experimental results, it is measured that when the predicted deviation value is 3mm, the character deformation is 20%; when the predicted deviation value is 1mm, the character deformation is 13.3%. Therefore, improving the prediction accuracy of the cable displacement prediction model will help improve the efficiency of code switching, while reducing the deformation of the character mark after switching, so that the system can present a better control effect.
[0020] According to the characteristics of the system, a method for predicting cable displacement using an adaptive neural fuzzy inference system, namely ANFIS model, is proposed to replace the traditional fuzzy algorithm. The ANFIS model combines the learning ability of neural networks and the reasoning ability of fuzzy logic, and processes input data and generates prediction outputs through a series of layers. ANFIS is based on the TS fuzzy inference system. By combining the advantages of fuzzy logic and neural networks, it has the ability to handle nonlinear problems, strong robustness, strong interpretability, adaptive learning ability, strong generalization ability, wide application range, integration of multiple technologies and easy implementation.
[0021] Based on a full analysis of the cable cutting environment and the working conditions of the inkjet printer, the model identifies the input feature parameters that affect the inkjet printer's character identification printing effect, trains the input feature parameter recognition, and selects the influencing factor screening model for cable displacement prediction from the previous character displacement data. The filtered data is input into the ANFIS model and the model is trained. Finally, a cable displacement prediction model based on the ANFIS algorithm is proposed.
[0022] The ANFIS model structure consists of five layers: input layer, fuzzification layer, rule layer, normalization layer, and output layer, and each layer represents different functions.
[0023] Input Layer: The input variables are: x is the input variable, and j is the number of input variables.
[0024] Fuzzy layer: Fuzzy rules are used to calculate the membership of each input variable. This system uses the most commonly used Gaussian function as the membership function; assuming that the membership function is ,but in is the center of the membership function, is the standard deviation; n is the input parameter; k is the number of fuzzy subsets.
[0025] Rules layer: At this level, IF-THEN rules based on fuzzy logic are used to form fuzzy reasoning; assuming that The form of the rule is: rule :if yes and yes And... and yes ,but .
[0026] in, is the fuzzy set of the fuzzy system; The parameters of the fuzzy system are used; the ANFIS model is used to automatically learn the parameters of the fuzzy rules and membership functions from the input and output data through the training algorithm; The output is obtained according to the fuzzy rules. The input part (i.e., the if part) is fuzzy, and the output part (i.e., the then part) is definite. The fuzzy reasoning indicates that the output is a linear combination of the input.
[0027] The excitation strength of each rule is determined by the membership of the input variable: Indicates the excitation intensity.
[0028] Normalization layer: In this layer, the excitation intensity of each rule is normalized so that its sum is 1. for: Output layer: Calculate the output of each rule and perform a weighted sum of the outputs of all rules to get the final system output: The final output is the weighted sum of all rule outputs.
[0029] In the actual application of the algorithm, the input and output are consistent with the traditional fuzzy algorithm, and the network structure is 3-3-1. The ANFIS fuzzy neural network algorithm was verified using the Matlab tool. Among the 1,200 sets of sample data, the maximum deviation of the ANFIS fuzzy neural network prediction algorithm was 2.08mm and the minimum deviation was 0.52mm. Overall, the prediction deviation of the ANFIS fuzzy neural network is smaller than the traditional fuzzy prediction deviation (1-3mm). After adopting the new algorithm, the prediction effect is significantly improved. The new algorithm improves the accuracy of cable displacement prediction and reduces the deformation of the inkjet characters.
Claims
1. The cable displacement prediction system of the adaptive inkjet control system is characterized by: include Analyze the cable cutting environment and the working conditions of the inkjet printer, identify the input feature parameters that affect the inkjet printer's character identification printing effect, train the input feature parameter recognition, and select a screening model for factors affecting the prediction of cable displacement from previous character displacement data; the filtered data is input into the ANFIS model, and the model is trained; The ANFIS model structure includes five layers: input layer, fuzzification layer, rule layer, normalization layer, and output layer; Input Layer: The input variables are: x is the input variable, and j is the number of input variables; Fuzzy layer: Fuzzy rules are used to calculate the membership of each input variable. This system uses Gaussian function as the membership function. Assume that the membership function is ,but in is the center of the membership function, is the standard deviation; n is the input parameter; k is the number of fuzzy subsets; Rules layer: At this level, IF-THEN rules based on fuzzy logic are used to form fuzzy reasoning; assuming that The form of the rule is: rule :if yes and yes And... and yes ,but ; in, is the fuzzy set of the fuzzy system; is the fuzzy system parameter; The output obtained according to the fuzzy rule, the input part, that is, the if part, is fuzzy, and the output part, that is, the then part, is certain. The fuzzy reasoning indicates that the output is a linear combination of the input; The excitation strength of each rule is determined by the membership of the input variable: represents the excitation intensity; Normalization layer: In this layer, the excitation intensity of each rule is normalized so that its sum is 1; the normalized excitation intensity for: Output layer: Calculate the output of each rule and perform a weighted sum of the outputs of all rules to get the final system output: The final output is the weighted sum of all rule outputs.
2. The cable displacement prediction system of the adaptive inkjet control system according to claim 1 is characterized in that: The general steps of applying fuzzy algorithms include first determining the input and output of the fuzzy prediction model, then formulating fuzzy rules based on the operator or expert experience, then fuzzifying the model input signal based on real-time sampling data, and then completing the fuzzy reasoning process according to the fuzzy rules to obtain the fuzzified output value, and finally denormalizing the output value.
3. The cable displacement prediction system of the adaptive inkjet control system according to claim 1 is characterized in that: To analyze the sampled data, the model uses a triangular membership function; Among them, μ(x) represents the element x Membership degree: When x is between the interval [a, b), the membership increases from 0 to 1; When x is between the interval [b,c), the membership decreases from 1 to 0; When x is less than a or greater than c, the membership is 0.
4. The cable displacement prediction system of the adaptive inkjet control system according to claim 1 is characterized in that: In the actual application of the ANFIS model algorithm, the input and output are consistent with the traditional fuzzy algorithm, and the network structure is 3-3-1.