A method for simulating and predicting the energy consumption during the cutting process of an intelligent processing production line

Abstract

The present invention discloses a method for simulating and predicting the energy consumption during the cutting process of an intelligent processing production line. In the present invention, the cutting speed, cutting depth, and feed rate during the cutting process of the intelligent processing production line are used as dependent variables, and the cutting energy consumption is used as an independent variable. Based on the actual production process data, with the aid of Design-expert software for simulation, a cubic polynomial fitting equation model of the mutual relationship between the cutting speed, cutting depth, feed rate, and cutting energy consumption is established; and in combination with the variance analysis table, the model is simplified. This model requires fewer data samples and has a high model accuracy, overcoming the problems of difficult data acquisition, difficult modeling, and insufficient model accuracy in the discrete processing process. The present invention utilizes the Design-expert software platform, combines the actual production process data, and conducts energy consumption simulation and prediction, improving the accuracy of the energy consumption prediction model.

Description

Technical Field

[0001] The present invention relates to an energy consumption simulation prediction method for energy consumption simulation prediction in the cutting process of an intelligent processing production line. Background Art

[0002] Reducing energy consumption is a key issue for the green and sustainable development of the manufacturing industry. Machine tools, as a typical energy-consuming equipment in the manufacturing industry, play an important role. In actual production, machine tools usually show very low efficiency, and they have significant potential for energy conservation and emission reduction. Therefore, it is necessary to analyze the influencing factors of energy consumption in the machining process of machine tools and establish an accurate energy consumption prediction model to achieve accurate evaluation and effective control of energy efficiency.

[0003] Due to the influence of the inherent uncertainty in the processing process on energy consumption and the difficulty in obtaining energy consumption data, how to model energy consumption is a huge challenge. Existing energy consumption modeling methods mainly rely on processing theory, empirical models, and intelligent algorithms. Existing processing theory and empirical models can explain the processing process, but the accuracy of the models is insufficient; existing intelligent algorithms rely too much on data sets, and unreasonable data set division and few data samples will cause great errors in modeling. In the discrete processing process, it is difficult to obtain energy consumption data. For small-sample and non-linear energy consumption data sets, a more appropriate and accurate modeling method needs to be selected to conduct energy consumption prediction research. Summary of the Invention

[0004] To solve the above problems, the present invention establishes a multiple regression prediction model of energy consumption through the response surface method, and provides a method for energy consumption simulation prediction in the cutting process of an intelligent processing production line. The method of the present invention includes:

[0005] Step 1: Taking cutting speed, cutting depth, and feed rate as test factors and cutting energy consumption as the test index, considering the mutual relationship between the test factors and the test index, using the Box-Behnken test design method in combination with the actual processing process to construct a test factor level table;

[0006] Step 2: Based on the test scheme design in Step 1, collect test data and eliminate singular samples;

[0007] Step 3: Based on the test data, use Design-expert software for simulation to establish a cubic polynomial fitting equation model of the mutual relationship between cutting speed, cutting depth, feed rate, and cutting energy consumption;

[0008] Step 4: According to the variance analysis table of the established cubic polynomial fitting equation model, analyze the significance degree of the linear term, quadratic term, cubic term, and interaction term in the model on energy consumption, eliminate non-significant terms, and simplify the cubic polynomial fitting equation;

[0009] Step Five: Analyze the established model and calculate whether its accuracy meets the requirements.

[0010] In Step One, the cutting energy consumption is mainly affected by the process parameters and the characteristics of the workpiece material. The process parameters include cutting speed, cutting depth, and feed rate. The energy consumption in this part is represented by Equation (1):

[0011] E c +E va =E spindle +E feed =f(v, a p , f t , mat) (1)

[0012] Where E c represents the cutting energy consumption, E va represents the variable energy consumption, E spindle represents the spindle motor energy consumption, E feed represents the feed axis motor energy consumption, v - cutting speed, a p - cutting depth, f t - feed rate, mat - material characteristics;

[0013] In the actual cutting process, by controlling the magnitudes of the process parameters on the operation panel of the machine tool, the machining process is further controlled. Therefore, the process parameters are used as test factors to explore the relationship with energy consumption. Thus, it is determined that:

[0014] Test factors: v - cutting speed; a p - cutting depth; f t - feed rate;

[0015] Test index: E c - cutting energy consumption.

[0016] Specifically, in Step One, in order to capture the influence of each factor on the test index, the orthogonal method is used to design the experiment, and the orthogonal table is directly used for design according to the number of test attributes and the number of levels; according to the Box - Behnken experimental design method, the levels of the test factors are selected, and each factor takes 3 - 7 levels, and then the required factor - level table is constructed.

[0017] In Step Two, after setting the test factors and the test index based on Step One, according to the machining situation of the workpiece, the factor levels in the test machining are determined. The specific process is as follows:

[0018] (2.1) Determination of the machining levels of the test factors cutting speed, cutting depth, and feed rate: According to the requirements of the Box - Behnken experimental design, the three influencing factors take 3 - 7 test levels according to the actual machining state;

[0019] (2.2) After determining the factor levels, the processing operation is carried out in the way of orthogonal experiment; the orthogonal experiment process can be designed to carry out full-factor experiment or partial-factor experiment;

[0020] (2.3) Collect the corresponding energy consumption data;

[0021] (2.4) Outlier sample rejection: During the process of collecting data, there will be some disturbing situations, which will cause individual data in the obtained data to deviate significantly from other detection values. Such "outlier samples" are rejected.

[0022] (2.5) Determine the experimental data.

[0023] Specifically, the method for judging outlier samples in step 2.4 is as follows: After obtaining the experimental data, draw a scatter plot and find the outlier points. The outlier points are isolated data points. From the distribution point of view, the outlier points are far away from other data points in the data set; if outlier points are found on the scatter plot, then judge this point as an outlier sample; when encountering an outlier sample, it is necessary to determine whether to reject or modify this sample through repeated precise experiments.

[0024] Specifically, in step three, based on the obtained experimental data, use Design-expert software to simulate the model to obtain a cubic polynomial expression of the relationship between the independent variable and the dependent variable, which is expressed by formula (2). Here, the third-order linear formula is expanded and explained:

[0025]

[0026] where x 1 , x 2 , x 3 are independent variables, b 0 is a constant term, b 1 , b 2 , b 3 are linear terms, b 4 , b 5 , b 6 , b 10 , b 11 , b 12 , b 13 , b 14 , b 15 , b 16 are interaction terms, b 7 , b 8 , b 9 are quadratic terms, b 17 , b 18 , b 19 are cubic terms, and ε is an error term;

[0027] The cutting energy consumption E is established according to formula (2). c and the experimental factors v - cutting speed, a p - cutting depth, f t - feed rate, the cubic polynomial fitting equation model is as shown in formula (3):

[0028]

[0029] Specifically, in step four, the 95% confidence level in the analysis of variance is applied to adjust the statistical significance of the polynomial model. Among them, the P - value and F - value are very important for determining the sufficiency and significance of the regression model; the magnitude of the influence of a single factor on the energy consumption in the model is determined by the F - value. The larger the F - value, the greater the influence of this factor in this model; the P - value is the basis for determining whether all terms in this model are significant:

[0030]

[0031] In the above formula, P - value represents the P - value, and F - value represents the F - value; according to the definition of the P - value, the significance of the terms is analyzed, and the terms with insignificant influence in the model are eliminated, thereby achieving the effect of simplifying the model.

[0032] Specifically, in step five, the prediction accuracy of the model is calculated, using the mean absolute percentage error MAPE and the coefficient of determination R 2 to represent.

[0033] Coefficient of determination:

[0034]

[0035] Mean absolute percentage error:

[0036]

[0037] In the formula is the true value, y i is the predicted value, is the average value, i represents the i - th sample, and n represents the total number of samples.

[0038] The beneficial effects of the present invention are as follows: Taking the cutting speed, cutting depth, and feed rate during the cutting process of the intelligent processing production line as test factors and the cutting energy consumption as the test index, based on the actual production process data, with the help of Design-expert software for simulation, a cubic polynomial fitting equation model of the mutual relationship between the cutting speed, cutting depth, feed rate, and cutting energy consumption is established; and combined with the variance analysis table, the model is simplified. This model requires fewer data samples and has high model accuracy, overcoming the problems of difficult data acquisition, difficult modeling, and insufficient model accuracy in the discrete processing process. The present invention uses the Design-expert software platform, combines the actual production process data, and conducts energy consumption simulation prediction, improving the accuracy of the energy consumption prediction model. Brief Description of the Drawings

[0039] Figure 1 is the flowchart of the present invention.

[0040] Figure 2 is the comparison diagram of the prediction curves of the energy consumption multiple linear regression model (MLR), empirical exponential model (NREM), quadratic polynomial regression model (SORSM), simplified quadratic polynomial regression model (SSORSM), cubic polynomial regression model (TORSM), and simplified cubic polynomial regression model (STORSM) established using the public data set in the embodiment of the present invention.

[0041] Figure 3 is a 3D diagram of the interaction of influencing factors obtained by software simulation in the embodiment of the present invention.

[0042] Among them, Figure 3(a) is a 3D surface diagram of the influence of the interaction between the cutting speed and the cutting depth on the energy consumption, Figure 3(b) is a 3D surface diagram of the influence of the interaction between the cutting speed and the feed per tooth on the energy consumption, and Figure 3(c) is a 3D surface diagram of the influence of the interaction between the cutting depth and the feed per tooth on the energy consumption. Detailed Embodiment

[0043] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments of the present invention will be further described in detail below in conjunction with the drawings and embodiments.

[0044] This embodiment provides a third-order response surface method for energy consumption simulation prediction during the cutting process of an intelligent processing production line. Refer to Figure 1 , and the steps are as follows:

[0045] Step 1: Taking the cutting speed, cutting depth, and feed rate as test factors and the cutting energy consumption as the test index, considering the mutual relationship between the test factors and the test index, using the Box-Behnken test design method in combination with the actual processing process, construct a test factor level table;

[0046] The quantity used to measure the test results in the test is called the test index, and the quantity that affects the test index is called the test factor. In the actual cutting process of the machine tool, the cutting speed, cutting depth, and feed rate are set by controlling the operation panel to control the entire processing process. Different values ​​of each factor have different corresponding indicators. In order to capture the impact of each factor on the test index, when conducting the experiment, it is necessary to use the orthogonal method to design the experiment. The orthogonal table can be directly used for design according to the number of attributes and the number of levels of the experiment. According to the Box-Behnken test design method, the level of the test factor is selected, and the level range should be wide enough. Each factor can take 3-7 levels, and then the required factor level table is constructed (for example, each factor takes 3 levels, which can constitute a factor level table of 3 factors and 3 levels).

[0047] Taking the cutting process of a workpiece as an example, consider v-cutting speed, a p - cutting depth and f t -The influence of three machining factors on energy consumption, each factor takes four levels, that is, the four-level-three-factor design matrix is ​​used in the experimental design, and the factor level table in the experimental design scheme is shown in Table 1. Among them, -1, 0, 1, and 2 represent the coded values ​​of the level, and the real value needs to be used for statistics in actual processing.

[0048] Table 1. Experimental factor level table (coded value)

[0049]

[0050] Step 2: Based on the experimental design in step 1, collect experimental data and eliminate outliers.

[0051] After setting the independent variables and dependent variables based on step 1, the factor levels in the experimental processing are determined according to the processing conditions of the workpiece. This is explained here based on an existing public data set.

[0052] (2.1). Determination of the machining level of the experimental factors cutting speed, cutting depth and feed rate: According to the BBD design requirements, the three influencing factors generally take 3-7 test levels according to the actual machining status.

[0053] Replace the coded values ​​in Table 1 in step 1 with the real values ​​in the actual production process and perform BBD design. The factor levels of the design are shown in Table 2.

[0054] Table 2. Factor level table (actual value)

[0055]

[0056] (2.2) After determining the levels of the test factors, the machining operation is carried out in the way of orthogonal test, and the orthogonal test process can be designed for full factorial test or fractional factorial test.

[0057] (2.3) Collect the corresponding energy consumption data.

[0058] (2.4) Singular sample rejection: During the process of collecting data, there will be some disturbing situations, which will cause some individual data in the obtained data to deviate significantly from other detected values. A singular sample refers to the "extreme value" that appears in the sample, and the data value looks extremely large or extremely small, and its distribution significantly deviates from the rest of the observed values. Outlier analysis is to test whether there are unreasonable data in the data. In data analysis, neither the existence of outliers can be ignored, nor can outliers be simply removed from the data analysis. After obtaining the test data in the present invention, a scatter plot is drawn. If it is found that there are outliers on the scatter plot (an outlier is an isolated data point. From the perspective of distribution, the outlier is far away from other data points in the data set), then this point is judged as a singular sample. If a singular sample is encountered, it is necessary to determine whether to reject or modify this sample through repeated precise tests.

[0059] (2.5) Determination of test data: Here, the full factorial experimental design method is adopted, and the obtained data is shown in Table 3.

[0060] Table 3. Test data

[0061]

[0062] Step 3: Analysis of test data. With the help of Design-expert software for simulation, a cubic polynomial fitting equation model of the mutual relationship between cutting speed, cutting depth, feed rate and cutting energy consumption is established.

[0063] Based on the actual machining data in Step 2, with the help of Design-expert software for simulation calculation, a cubic polynomial fitting equation model of the relationship between test factors and energy consumption index is established as shown in Formula (6).

[0064]

[0065] where v - cutting speed; a p - cutting depth; f t - feed rate; E c - cutting energy consumption.

[0066] Step 4: According to the analysis of variance table of the established cubic polynomial fitting equation model, analyze the significant degrees of the linear term, quadratic term, cubic term and interaction term in the model on the energy consumption, reject the non-significant terms, and simplify the cubic polynomial fitting equation.

[0067] Obtain the analysis of variance table corresponding to the cubic polynomial fitting equation model, as shown in Table 4:

[0068] Table 4. Analysis of Variance Table for Cubic Polynomial Model

[0069] Source of variation Sum of squares Degree of freedom Mean square F value P value Model 1.412E+006 19 74310.72 52.83 <0.0001 v - Cutting speed 26505.31 1 26505.31 18.84 <0.0001 <![CDATA[a p - Depth of cut]]> 20421.50 1 20421.50 14.52 0.0004 <![CDATA[f t - Feed rate]]> 2063.32 1 2063.32 1.47 0.2323 <![CDATA[v a p > 20419.84 1 20419.84 14.52 0.0004 <![CDATA[v f t > 24232.06 1 24232.06 17.23 0.0001 <![CDATA[a p f t > 5725.99 1 5725.99 4.07 0.0498 <![CDATA[v 2 > 6484.10 1 6484.10 4.61 0.0373 <![CDATA[a p 2 > 941.64 1 941.64 0.67 0.4177 <![CDATA[f t 2 > 7384.01 1 7384.01 5.25 0.0268 <![CDATA[v a p f t > 1569.18 1 1569.18 1.12 0.2966 <![CDATA[v 2 a p > 15010.05 1 15010.05 10.67 0.0021 <![CDATA[v 2 f t > 7164.06 1 7164.06 5.09 0.0290 <![CDATA[v a p 2 > 635.69 1 635.69 0.45 0.5049 <![CDATA[v f t 2 > 4424.00 1 4424.00 3.15 0.1013 <![CDATA[a p 2 f t > 87.78 1 87.78 0.062 0.8039 <![CDATA[a p f t 2 > 2344.10 1 2344.10 1.67 0.2035 <![CDATA[v 3 > 11784.62 1 11784.62 8.38 0.0059 <![CDATA[a p 3 > 941.88 1 941.88 0.67 0.4176 <![CDATA[f t 3 > 746.74 1 746.74 0.53 0.4701 Residual 61892.43 44 1406.65 Total 1.474E+006 63

[0070] The results of the analysis of variance table show that the F-value of the model is 52.83 and the P-value is less than 0.0001, indicating that the model is highly significant. According to the magnitude of the F-value, in this cubic model, the order of the influence of single factors on energy consumption is cutting speed (18.84) > cutting depth (14.52) > feed rate (1.47). In addition, the results of this regression analysis show that the cutting speed (v) and the cutting depth (a p ), the term va p , vf t , a p f t , v 2 , f t , v 2 a p , v 2 , f t , v 3 have a significant impact on energy consumption. The feed rate (f t ) is an essential item in the actual experiment and needs to be included in the simplified model. Therefore, the model is simplified by removing other items with insignificant effects, and the simplified cubic polynomial model is shown in Equation (7).

[0071]

[0072] Step Five: Analyze the established model and calculate whether its accuracy meets the requirements.

[0073] To verify the performance of the present invention, a multiple linear regression model of energy consumption (MLR), an empirical exponential model (NREM), a quadratic polynomial regression model (SORSM), and a simplified quadratic polynomial regression model (SSORSM) are established, as shown in Formulas (8), (9), (10), and (11). The prediction accuracy of each model is calculated using the mean absolute percentage error MAPE and the coefficient of determination R 2 . Among them, the lower the value of MAPE, the higher the accuracy of the model; the closer R 2 is to 1, the higher the accuracy of the model. The results are shown in Table 5. The prediction curves of each model are as Figure 2 shown. By comparison, the accuracy of the established cubic polynomial regression model is the best. The interaction between factors is shown in Figures 3(a), (b), and (c) respectively. Combining the P-values in the analysis of variance table, it can be seen that the interaction between factors has a significant impact on energy consumption.

[0074] MLR model:

[0075] E c =-357.4059 + 3.2924v + 173.6875a p + 292.075f t (8)

[0076] NREM model:

[0077] E c = 1.99844v 1.10634 a p 0.21311 f t 0.12635 (9)

[0078] SORSM model:

[0079]

[0080] SSORSM model:

[0081]

[0082] Table 5, MAPE and R 2

[0083] MLR NREM SORSM SSORSM TORSM STORSM MAPE 0.0778 0.0833 0.0504 0.0521 0.0367 0.0449 <![CDATA[R 2 > 0.8474 0.8593 0.9287 0.9223 0.9580 0.9498

[0084] In this embodiment, taking the cutting process as the object, the functional relationship between the test factors and the energy consumption is mined, and an energy consumption model is established. After performing Steps 1 to 5, a cubic polynomial regression model of the relationship between the test factors and the energy consumption is obtained, and the model is simplified in combination with the analysis of variance table. Compared with the established multiple linear regression model of energy consumption (MLR), empirical exponential model (NREM), quadratic polynomial regression model (SORSM), and simplified quadratic polynomial regression model (SSORSM), the prediction accuracy of the cubic polynomial regression model is 95.8%, which is higher than that of other models.

[0085] It can be seen that the model established by the present invention not only requires fewer data samples, but also has high model accuracy, overcoming the problems of difficult data acquisition, difficult modeling, and insufficient model accuracy in the discrete processing process.

[0086] Some steps in the embodiments of the present invention can be implemented by software, and the corresponding software program can be stored in a readable storage medium, such as an optical disc or a hard disk.

[0087] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for simulating and predicting energy consumption in the cutting process of an intelligent machining production line. It is characterized in that The following steps are involved: Step 1: Taking cutting speed, cutting depth and feed rate as test factors and cutting energy consumption as test index, considering the relationship between test factors and test indexes, using Box-Behnken test design method combined with actual machining process, constructing the test factor level table; Step 2: Based on the experimental design of step 1, collect experimental data and eliminate outliers; Step 3: Based on the experimental data, the Design-expert software is used for simulation to establish a cubic polynomial fitting equation model of the relationship between cutting speed, cutting depth, feed rate and cutting energy consumption; Step 4: According to the variance analysis table of the established cubic polynomial fitting equation model, analyze the significance of the impact of the linear term, quadratic term, cubic term and interaction term in the model on energy consumption, eliminate non-significant terms, and simplify the cubic polynomial fitting equation; Step 5: Analyze the established model and calculate whether its accuracy meets the requirements; In step 1, the cutting energy consumption is mainly affected by the process parameters and the characteristics of the workpiece material. The process parameters include cutting speed, cutting depth, and feed rate. The energy consumption of this part is expressed by formula (1): E c +E va =E spindle +E feed =f(v,a p ,f t ,mat) (1) Among them, E c represents the cutting energy consumption, E va represents the variable energy consumption, E spindle represents the spindle motor energy consumption, E feed represents the feed axis motor energy consumption, v - cutting speed, a p - cutting depth, f t - feed rate, mat - material characteristics; In the actual cutting process, the processing is controlled by controlling the size of each process parameter on the machine tool operation panel. Therefore, the process parameters are used as experimental factors to explore the relationship with energy consumption, and thus determine: Test factors: v - cutting speed; a p - cutting depth; f t - feed rate; Test index: E c - Cutting energy consumption; In step 3, based on the obtained experimental data, the model is simulated using the Design-expert software to obtain a cubic polynomial expression of the relationship between the independent variable and the dependent variable, which is expressed by formula (2). Here, the third-order linear formula is expanded and explained: where x 1 , x 2 , x 3 is the independent variable, b 0 is the constant term, b 1 , b 2 , b 3 is the linear term, b 4 , b 5 , b 6 , b 10 , b 11 , b 12 , b 13 , b 14 , b 15 , b 16 is the interaction term, b 7 , b 8 , b 9 is the quadratic term, b 17 , b 18 , b 19 is the cubic term, and ε is the error term; Establish the cutting energy consumption E according to formula (2) c and the experimental factors v - cutting speed, a p - cutting depth, f t - feed rate is shown in the cubic polynomial fitting equation model as formula (3):

2. The method for simulating and predicting energy consumption in the cutting process of an intelligent processing production line according to claim 1, It is characterized in that In the step 1, in order to capture the influence of various factors on the test indicators, the orthogonal method is used to design the test, and the orthogonal table is used to design directly according to the number of attributes and the number of levels of the test; according to the Box-Behnken test design method, the levels of the test factors are selected, and each factor takes 3-7 levels, thereby constructing the required factor level table.

3. The method for simulating and predicting energy consumption in the cutting process of an intelligent machining production line according to claim 1, It is characterized in that In the step 2, after the test factors and test indicators are set based on the step 1, the factor levels in the test processing are determined according to the processing conditions of the workpiece. The specific process is as follows: (2.1). Determination of the machining level of the experimental factors cutting speed, cutting depth and feed rate: According to the requirements of Box-Behnken experimental design, the three influencing factors take 3-7 test levels according to the actual machining status; (2.2). After determining the factor levels, the processing operation is carried out using the orthogonal test method; (2.3) Collect corresponding energy consumption data; (2.4). Singular sample rejection: During the process of data collection, there may be some disturbances, which may result in individual data that significantly deviates from other measured values in the obtained data. Such "singular samples" are rejected; (2.5). Determination of test data.

4. The method for simulating and predicting the energy consumption during the cutting process of an intelligent processing production line according to claim 3, characterized in that, the method for judging singular samples in step 2.4 is: after obtaining the test data, draw a scatter plot and find the outlier. The outlier is an isolated data point. From the perspective of distribution, the outlier is far away from other data points in the data set; if an outlier is found on the scatter plot, then this point is judged as a singular sample; when encountering a singular sample, it is necessary to determine whether to reject or modify this sample through repeated precise tests.

5. The method for simulating and predicting the energy consumption during the cutting process of an intelligent processing production line according to claim 3, characterized in that, in step 2.2, the orthogonal test process can be designed to conduct a full factorial experiment or a fractional factorial experiment.

6. The method for simulating and predicting the energy consumption during the cutting process of an intelligent processing production line according to claim 1, characterized in that, in step four, the 95% confidence level in the analysis of variance is used to adjust the statistical significance of the polynomial model. The P-value and F-value are important for determining the sufficiency and significance of the regression model; the magnitude of the influence of a single factor on the energy consumption in the model is determined by the F-value. The larger the F-value, the greater the influence of this factor in this model; the P-value is the basis for determining whether all terms in this model are significant: in the above formula, P-value represents the P-value, and F-value represents the F-value; the significance of the terms is analyzed according to the definition of the P-value, and the terms with insignificant influence in the model are removed, thereby achieving the effect of simplifying the model.

7. The method for simulating and predicting the energy consumption during the cutting process of an intelligent processing production line according to claim 1, characterized in that, The prediction accuracy of the calculation model in the fifth step is represented by the Mean Absolute Percentage Error (MAPE) and the coefficient of determination R 2 as described 8. The method for simulating and predicting the energy consumption during the cutting process of an intelligent processing production line according to claim 7, characterized in that, Coefficient of determination: Absolute percentage error: where is the true value, y i is the predicted value, is the average value, i represents the i-th sample, and n represents the total number of samples.

Images