Chen, Joseph C "Neural Networks and Neural-Fuzzy Approaches in an In-Process Surface Roughness Recognition System for End Milling Operations" Computational Intelligence in Manufacturing Handbook Edited by Jun Wang et al Boca Raton: CRC Press LLC,2001 16 Neural Networks and Neural-Fuzzy Approaches in an In-Process Surface Roughness Recognition System for End Milling Operations 16.1 16.2 16.3 16.4 Joseph C Chen Iowa State University Introduction Methodologies Experimental Setup and Design The In-Process Surface Roughness Recognition Systems 16.5 Testing Results and Conclusions 16.1 Introduction Different machining processes produce different products with varying qualities When evaluating the quality of a finished piece, surface roughness is the most important result of the machining process to consider, because many product attributes can be determined by how well the surface finish is produced The quality of the surface finish, or surface roughness, affects several functional attributes of parts, such as surface friction, wear, reflectivity, heat transmission, porosity, coating adherence, and fatigue resistance The desired surface roughness value is usually specified for individual parts, and a particular process is selected in order to achieve the specified roughness Typically, surface roughness measurement has been carried out by manually inspecting machined surfaces at fixed intervals A surface profilometer containing a contact stylus is used in the manual inspection procedure This procedure is both time-consuming and labor-intensive In addition, a number of defective parts could be produced during the time needed to complete an off-line surface inspection, thereby creating additional production costs Another disadvantage of using surface profilometers is that they register the serious interference of extraneous vibration generated in the surrounding environment This extraneous vibration might significantly influence the accuracy of surface measurements For these reasons, researchers are seeking solutions to model the surface roughness in an online or in-process fashion ©2001 CRC Press LLC The studies of Martellotti [1941, 1945] are among the earliest that represent a major contribution to the understanding of kinematics and the mechanism of surface generation in milling processes Martellotti developed parametric equations to describe the trochoidal path that the tool follows These studies also provide approximate analytical expressions for the ideal peak-to-valley roughness generated in upand down-slab milling, and face milling Numerous other studies have explored the topography of milled surfaces Many of these focused on predicting the two- or three-dimensional shape of a milled surface under ideal and non-ideal conditions Kline et al [1982] demonstrated the effects of cutter runout on surface errors, and surface errors or dimensional inaccuracies were predicted using the cantilever beam theory for cutter runout Another study by Babin et al [1985] applied the cantilever beam theory to predict the topography of wall surfaces produced by end milling Armarego and Deshpande [1989] presented one more milling process geometry model that incorporates cutter runout to predict cutting forces Sutherland and Babin [1985] demonstrated a two-dimensional worst-case analysis of the slot floor surface However, the model for the slot floor surface significantly underpredicted surface roughness values Research by Kolarits and DeVries [1989] extended the previous model to account for varying cut geometries and feed rates This extended floor surface generation model improved prediction capabilities considerably However, the roughness parameter predictions for some of the tests were found to deviate greatly from measured values You and Ehmann [1989] developed a comprehensive model to predict the three-dimensional surface texture generated by ball end mills They also presented an algorithm for three-dimensional representations of the machined surface; however, the effect of flexibility of the cutter-workpiece system was not considered in this model Montgomery and Altintas [1991] presented the effects of the cutter-workpiece system flexibility in their force and surface prediction model in order to analyze the surface generation mechanism in peripheral milling under dynamic cutting conditions All models previously discussed represent only deterministic cutting models, but most machined surfaces exhibit interrelated characteristics of both random and deterministic components Zhang and Kapoor [1991] demonstrated the effect of random vibrations on surface roughness in the turning process These vibrations were shown to occur due to random variations in the microhardness of the workpiece material Ismail and others presented a surface generation model in milling that included both cutter vibrations and the effects of tool wear [Ismail et al., 1993] Melkote and Thangaraj [1994] presented another enhanced end milling surface texture model including the effects of radial rake and primary relief angles These three models, limited to laboratory usage or based on theoretical analysis, could not be implemented as an in-process monitoring system The findings of this literature review, in addition to communication with leading private industrial research and development laboratories in the state of Iowa (including Winnebago Co in Forest City; Delavan Inc in Des Moines; Sauer-Sundstrand Inc in Ames), point to the feasibility of in-process surface roughness recognition (ISRR) systems for implementation in the newer generation of milling machines The successful implementation of this surface roughness recognition system will enable metal cutting industries to reduce manufacturing costs by eliminating the relatively inefficient off-line quality control aspect of surface roughness inspection Therefore, reductions in manufacturing costs will increase competitiveness in worldwide markets This implication supports the development of an effective and inexpensive ISRR system The development of this system will enable implementation of adaptive control in modern manufacturing environments 16.2 Methodologies In order to provide an adaptive control mechanism, ISRR systems require two major components: (i) sensors, which receive the dynamics signal from the machining cutting processes; and (ii) an intelligent technique able to learn the dynamics of the machining system while allowing for control features to be built in The research described in this chapter employed an accelerometer to detect the dynamics mechanism of the tool and material interface This study also used two major intelligent learning ©2001 CRC Press LLC methodologies to incorporate data about the machining process through actual cuts These methodologies were also employed to construct a control system that predicts surface roughness during the execution of the machining process These two learning methodologies are artificial neural networks (ANN) and fuzzy neural (FN) systems An overview of these two approaches follows in the next section 16.2.1 Neural Networks Model Several learning methods have been developed for ANNs Many of these learning methods are closely connected with a certain network topology, with the main categorization method distinguished by supervised vs unsupervised learning Backpropagation was chosen from among various learning methods already existing in this field This approach was adopted into this research for two reasons: primarily, it is the most representative and commonly used algorithm, in addition to being relatively easy to apply; additionally, it has been consistently successful when used in practical applications [Das et al., 1996; Huang and Chiou, 1996] The backpropagation algorithm can be divided into two main processes, the process of learning and the process of recalling 16.2.1.1 The Learning Process Step Given network parameters: Set all the necessary parameters, such as the number of input neurons (i), the number of hidden layers and the number of neurons included in each hidden layer (h), the number of output neurons (j), etc Step 2: Initialize the beginning weights and biases: Set all the initial weights and biases values randomly Step 3: Load the input vector X and the target output vector T of a training example Step 4: Calculate and infer the actual output vector Y (a) Calculate the output vector H of hidden layers net h = ∑W _ xhih • Xi – θ _ h h Equation (16.1) i ( ) H h = f net h = Equation (16.2) + exp – net h (b) Infer the actual output vector Y net j = h W _ hy hj 〈 H h − θ _ y j ( ) Y j = f net j = Equation (16.3) 1 + exp Equation (16.4) − net j Step 5: Calculate the error term (a) The error term of the output layer: δ j = Y j – Y j ( (b) The error term of the hidden layer: δ h )( T – Y ) = H (1– H ) ∑W _ hy • δ h j h hi j ©2001 CRC Press LLC j j Equation (16.5) Equation (16.6) Step 6: Calculate the revised weight of the weight matrix and the revised bias of the bias vector Equation (16.7) (a) For the output layer: ∆W _ hy hj = ηδ j H h , ∆θ _ y j = –ηδ j (b) For the hidden layer: ∆W _ xhih = ηδ h X i , ∆θ _ h h = – ηδ h Equation (16.8) Step 7: Adjust and renew the weight matrix and the bias vector (a) For the output layer: W_hyhj = W_hyhj + ∆W_hyhj , θ _y j = θ_y j + ∆θ_ y j Equation (16.9) (b) For the hidden layer: W_xhih = W_xhih + ∆W_xhih , θ _hh = θ _hh + ∆θ_hh Equation (16.10) Step 8: Repeat steps through 7, until the energy function has converged or the specified learning cycles are completely executed 16.2.1.2 The Recalling Process Step 1: Set all the network parameters Step 2: Read the weight matrix W_xh and W_hy, and the bias vector θ_h and θ_y Step 3: Load the input vector X of a testing example Step 4: Calculate and infer the actual output Y (a) Calculate the output vector H of hidden layers net h = ∑W _ xhih • Xi – θ _ h h Equation (16.11) i ( ) H h = f net h = Equation (16.12) + exp – net h (b) Infer the actual output vector Y net j = ∑W _ hy hj • H h – θ _ y j Equation (16.13) h ( ) Y j = f net j = 1 + exp Equation (16.14) – net j 16.2.2 Fuzzy-Nets Modeling The proposed fuzzy-nets system was developed by fuzzy rules generated from sampled input–output pairs This model is built in five steps 16.2.2.1 Step 1: Divide the Input and Output Spaces into Fuzzy Regions [ ] Assume that the domain intervals of input variable xi are x i– , x i+ , and that the domain intervals of [y –, y+] Each domain interval can be divided into 2N + regions The value of output variable y are N is dynamic for different variables, and the lengths of each region can be equal or unequal Each region is denoted by ©2001 CRC Press LLC SN (Small N), S(N-1) (Small N-1), …, MD (Medium), … , LN (Large N), Equation (16.15) and then assigned a fuzzy membership function The divisions of the input and output spaces are shown in Figure 16.1, where N is for x1, and for x2 and y The width for each variable is the same µ( x1 ) S2 S1 MD L1 L2 x1− x1+ d µ( x2 ) S3 S2 S1 MD L1 L2 L3 − x2 + x2 d µ(y) S3 S2 S1 MD L1 L2 L3 y− y+ d FIGURE 16.1 The domain intervals of the input–output variables and triangular membership function In this study, the input variables are spindle speed (S), feed rate (F), depth of cut (D), and vibration average per revolution (V) The output variable is the surface roughness average value, Ra A triangular membership function specified by three parameters {a, b, c} is employed as follows:  x –a  b–a triangle x ; a ,b ,c =  c–x  c –b   ( ©2001 CRC Press LLC ) x ≤a a ≤x ≤b Equation (16.16) b≤x ≤c c≤x The spread of the input feature shown in Figure 16.1 is defined as d= x i+ – x i– (i = 1,2, , k), 2N Equation (16.17) where x i– and x i+ are the domain intervals of variable xi, xi ∈ Xi There are 2N + fuzzy regions quantifying the universe of discourse Xi The center points of each linguistic variable are (x – i ( ) ) , x i– + d ,… , x i– + N – d , x i+ Equation (16.18) Equations 16.17 and 16.18 are also used for the output variable y 16.2.2.2 Step Generate Fuzzy Rules from Given Data Pairs through Experimentation Three steps are used for generating fuzzy rules: Determine the degree of input–output data obtained from the successful experiment Assign the input–output pairs to the region with the maximum degree Obtain one rule from one pair of designated input–output data In this study, the experimental input–output pairs were [S ,F ,D ,V ,R ], i i i i i a Equation (16.19) where i denotes the number of input–output pairs A human expert examined these rules to ensure their usefulness and correctness The degrees of each data pair were determined by the function  xi – xc 1 – , xi ∈ xc , xc + x s xs   x –x c i  , xi ∈ xc – x s , xc µ x i = 1 – xs  otherwise 0,    [ ( ) ] [ ] Equation (16.20) where xc is the center of the linguistic level x, and xs is the width of the linguistic level x, equal to d After all of the input and output elements were determined, each element was assigned to the region with the maximum degree One rule from one pair of the desired input–output pair [S1,F1,D1,V1,Ra1], was assigned For example, the degree of one input–output pair: was determined by Equation 16.20 as ©2001 CRC Press LLC } ( ) { µ ( F ) = {0.9 ∈ L , 0.1 ∈ L } µ ( D ) = {0.8 ∈ L ,0.2 ∈ L1} µ (V ) = {0.2 ∈MD, 0.8 ∈S1} µ ( R ) = {0.3 ∈S2, 0.7 ∈S1} µ S = 0.7 ∈MD, 0.3 ∈S1 1 Equation (16.21) 1 a The region of each datum with a maximum degree was assigned as follows: S ∈ MD , F ∈ L , D ∈ L ,V ∈S , R1 ∈S a Equation (16.22) Rule one was obtained by IF S1 is MD and F1 is L3 and D1 is L2 and V1 is S1 THEN Ra1 is S1, Equation (16.23) where AND indicates that the conditions of the IF statement must all be met simultaneously in order for the result of the THEN statement to be true 16.2.2.3 Step 3: Assign a Degree to Each Rule and Resolve the Conflicting Rules If two or more rules generated in step have the same IF command but a different THEN command, then the rules conflict To resolve conflicts between the two data sets, a degree must be assigned to each rule, generated from the data pairs as ( ) ( )( )( )( )( )( ) i d Ri = µ S i µ F i µ D i µ V i µ Ra µ E i where µ(Si) µ(Fi) µ(Di) µ(Vi) µ(Rai) µ(Ei) = = = = = = Equation (16.24) the degree of the spindle speed variable, the degree of the feed rate variable, the degree of the depth of cut variable, the degree of the vibration variable, the degree of the surface roughness variable, the degree assigned by the human expert to determine the importance of this rule The following function resolved the conflict between rules: ( ) ( ) d Rk – d Rl > ε Equation (16.25) where Rk and Rl are two conflicting rules, d(Rk) and d(Rl) are the degree of rules, Rk and Rl, and ε is the user-defined parameter < ε < 0.05 Next, the rule with the maximum degree is selected If the above function cannot resolve the conflict, ε may be decreased, or two more regions to one feature of the input vector may be increased and the input–output data pairs retrained If these rules still conflict, the region number of the next input feature is extended to two more regions and then retrained These procedures are repeated until all of the conflicting problems are resolved ©2001 CRC Press LLC L2 S L1 L1 MD S1 S2 S1 S2 S1 MD L1 L2 F FIGURE 16.2 Illustration of a combined fuzzy rule base 16.2.2.4 Step Create a Combined Rule Base A combined rule base consists of two kinds of rules: rules generated from numerical data by means of steps through 3, and linguistic rules determined by a human expert As shown in Figure 16.2, the combined rule base includes two fuzzy rules: IF S is L1 and F is L1 THEN Ra is L1; IF S is MD and F is S1 THEN Ra is S1 Equation (16.26) 16.2.2.5 Step Determine a Mapping Based on the Combined Fuzzy Rule Base A defuzzification strategy is used to determine the output control y for any given input datum There are many methods for defuzzification In this study, the following centroid of area method was applied: y= { ( ) ( ) ( ) ( )} where µ io = µ S i , µ F i , µ D i , µ V i ∑ µ 0i y i ∑ µ 0i Equation (16.27) , yi = the center value of the region, and y = the output for a given input datum This is the most widely adopted defuzzification strategy today, and it is reminiscent of the calculation of expected values of probability distributions 16.3 Experimental Setup and Design Figure 16.3 shows the complete experimental setup in this research A computer numerical control (CNC) program was written to perform the end milling cutting processes The electromagnetic proximity sensor was fixed at a close distance to the spindle, as shown in Figure 16.4, and the accelerometer sensor was mounted on the vise beneath the workpiece (Figure 16.5) Rotation and vibration data were collected simultaneously by the proximity sensor and the accelerometer sensor, respectively, when the cutter had cut the workpiece at a distance of 0.35 in The main concern was to avoid the impact of initially unstable or significant vibration Figure 16.6 illustrates the two types of signals (i.e., rotation data and vibration data) These two types of data were connected to an analog-to-digital (A/D) board (the vibration data from the accelerometer sensor had to be amplified by the PCB battery power unit beforehand) and then transmitted to a 486 personal computer for further data recording, processing, and analysis ©2001 CRC Press LLC CNC Machine Center Vise Accelerometer Sensor 486 Personal Computer VM DESIGNER: Wei-Liang Proximity Sensor PCB Battery Power Unit Workpiece A/D Board FIGURE 16.3 Experimental setup A computer program was written in C language that allowed the collection of the two kinds of data transformed from analog to digital signals by an A/D converter The collection time for each run was about 0.54 s, and the runs contained 6000 rotation or vibration data from the proximity sensors or accelerometer, respectively The cutting parameters (spindle speed, feed rate, and depth of cut) were changed manually according to different cutting conditions for each run Also, after each specimen was cut, the cutting tool was cleaned to avoid chip formation or a built-up edge (BUE), which would affect the surface roughness of the following cut The tool condition was also checked to ensure that it was free from defects All specimens in this experiment were conducted under dry cutting conditions without coolants Coolants are not generally used in order to reduce costs and prevent tool breakage due to thermal shock Moreover, the decision to use dry cutting conditions was based on the need to isolate the correlation between cutting vibrations and surface roughness in end milling With the experimental setup complete, the next step was to develop the ISRR-ANN and -FN models In this study, identifying the parameters of the training and testing data sets was a very important factor in establishing the experimental runs These runs could not exceed the suggested cutting parameters, which were based on machine capabilities and the nature of the material composition of both the workpiece and end mill ©2001 CRC Press LLC D (less than mm) Connected to A/D board Spindle Proximity Sensor FIGURE 16.4 Top view of the proximity sensor and spindle WORKPIECE Accelerometer sensor VISE To PCB battery power CNC TABLE FIGURE 16.5 The accelerometer setup 16.3.1 Training and Testing Experiments A total of 48 specimens were cut based on the following combination of cutting parameters: four levels of spindle speed (S = 750, 1000, 1250, and 1500 rpm); four levels of feed rate (F = 6, 12, 18, and 24 in./min); and three levels of depth of cut (D = 0.01, 0.03, and 0.05 in.) After these cuts were made, all specimens were measured off-line with a stylus profilometer to obtain their roughness average Ra values The average Ra value shown in Table 16.1 is the average of three Ra measurements of each specimen During the machining process, the accelerometer sensor produces vibration data Each cut produces 13 or more revolutions of vibration data The statistical average of the vibration voltages (V), transformed from analog to digital data, is based on one revolution of the spindle This vibration average served as the fourth independent variable for the ISRR systems Three statistical averages of vibration voltage (V) data were collected per specimen to provide a better empirical representation of the training data set Therefore, a total of 492 data sets were available for training In this research, 92 data sets were selected randomly from the 492 training data In addition, 36 experimental cuts under different cutting conditions than those in the training data set were used for a ©2001 CRC Press LLC Voltage Data points p 807 776 745 714 683 652 621 590 559 528 497 466 435 404 373 342 311 280 249 218 187 156 94 125 63 -1 32 Vibration data Revolution data FIGURE 16.6 Vibration and rotation signals flexible testing data set (Table 16.1) Therefore, a total of 128 pieces of data were used to evaluate the accuracy of ISRR systems The evaluation criteria are summarized in the next section 16.3.2 Test Criteria Criteria used in this experiment to judge the predictive capabilities of the ISRR–ANN and FN systems included the percentage deviation (φi) of each testing sample, given as: φi = Ra i′ – Ra i~M , × 100%, Ra i′ Equation (16.28) where φi = percentage deviation of single sample data Ra i′ = actual Ra measured by a profilometer Ra i~M = predicted Ra generated by the ISRR systems; M indicates the predicted value using ISRR, ANN or ISRR-FN models, respectively Additionally, the overall average percentage deviation (φ) of all 128 samples is given as m φi = ∑φ i i =1 m Equation (16.29) – where φ = average percentage deviation of all sample data m = the size of testing samples; in this study m = 128 16.4 The In-Process Surface Roughness Recognition Systems In this research, two ISRR systems were developed and tested Their training and testing processes are presented in the following sections ©2001 CRC Press LLC TABLE 16.1 Experimental Results Sample No Spindle Speed (rpm) Feed Rate (ipm) Depth of Cut (in.) Roughness (Ra: µin.) 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 750 750 750 750 750 750 750 750 750 750 750 750 1000 1000 1000 1000 1000 1000 1000 1000 1000 1000 1000 1000 1250 1250 1250 1250 1250 1250 1250 1250 1250 1250 1250 1250 1500 1500 1500 1500 1500 1500 1500 1500 1500 1500 1500 1500 6 12 12 12 18 18 18 24 24 24 6 12 12 12 18 18 18 24 24 24 6 12 12 12 18 18 18 24 24 24 6 12 12 12 18 18 18 24 24 24 0.01 0.03 0.05 0.01 0.03 0.05 0.01 0.03 0.05 0.01 0.03 0.05 0.01 0.03 0.05 0.01 0.03 0.05 0.01 0.03 0.05 0.01 0.03 0.05 0.01 0.03 0.05 0.01 0.03 0.05 0.01 0.03 0.05 0.01 0.03 0.05 0.01 0.03 0.05 0.01 0.03 0.05 0.01 0.03 0.05 0.01 0.03 0.05 65.40 62.75 72.40 143.85 101.90 94.05 184.80 146.60 121.05 186.55 170.40 172.40 58.40 78.30 62.20 129.90 83.60 92.05 137.50 124.15 85.75 163.15 153.30 142.30 49.95 63.30 70.85 101.30 98.75 84.95 115.00 92.25 94.70 155.45 108.85 120.65 36.55 55.70 55.65 87.55 81.65 94.05 119.45 86.50 104.25 119.20 103.30 109.40 16.4.1 ISRR-ANN Model Figure 16.7 shows the structure of the ISRR-ANN system The input variables were spindle speed (S), feed rate (F), depth of cut (D), and the statistical average of the vibration voltages (V) based on one revolution In their original form, the four independent variables (S, F, D, and V) for both data sets (training and testing) could not be trained by neural networks due to the wide range of values distributed among them ©2001 CRC Press LLC Workpiece Vibration Accelerometer Sensor ISRR-ANN Machining Process Spindle Rotation Machining Parameters Ra Proximity Sensor Spindle Speed Feed Rate Depth of Cut Input Output FIGURE 16.7 Structure of the ISRR-ANN In order to make them feasible as input neurons, all values in the first four columns (i.e., input neurons) were preprocessed by normalizing, and then transformed into a value between zero and one Based on the RMS error of the training examples and testing examples, it is clear that the 4-5-1 structure had the lowest RMS-error among all the structures with one hidden layer, and 4-7-7-1 was better than the other structures with two hidden layers In the final step of development for this ANN model, two programs were written in C language to predict the Ra values for one or two hidden layers by retrieving the weighted files that resulted from the training and testing 16.4.2 ISRR-FN System The structure of the ISRR-FN, as shown in Figure 16.8, consisted of the sensing system, machining parameters, and ISRR-FN In this sensing system, an accelerometer sensor was used to measure the realtime vibration of the workpiece, a proximity sensor was used to measure the real-time rotation of the spindle of the CNC machine center, and an A/D board and interface program were applied for analogto-digital conversion with 12-bit resolution Machining parameters, such as spindle speed, feed rate, and depth of cut, were transmitted to the ISRR-FN before or during machining The primary objective was to train the fuzzy system by generating fuzzy rules from input–output pairs, and combining these generated and linguistic rules into a common fuzzy rule base After input vectors were fuzzified by the fuzzification interface, the fuzzy inference engine generated output values by inferencing these fuzzified input values based on the fuzzy rule bank Finally, the defuzzification interface determined the final prediction of Ra values based on input variables Through training, the ISRR-FN learned to detect different conditions for individual machines, build the fuzzy rule base, and infer the surface roughness, Ra All of these processes were based on the experimental data Before the fuzzy-nets training takes up the experimental data, some parameters need to be preset based on the limitations of the machine and the machining processes They are summarized as follows: The domain intervals of the input–output data pairs were assigned as follows: • Spindle speed: [500, 2000] rpm • Feed rate: [6, 42] inches per minute • Depth of cut: [0.01, 0.07] inches • Vibration: [780, 2460] microvolts • Ra : [38, 168] micro inches ©2001 CRC Press LLC Workpiece Vibration Accelerometer Sensor Fuzzy Rule Bank Machining Process Spindle Rotation Machining Parameters Proximity Sensor Fuzzifier Deffuzzifier Ra Spindle Speed Feed Rate Fuzzy Inference Engine Depth of Cut ISRR-FN FIGURE 16.8 Structure of the ISRR-FN Each domain interval was divided into 2N + regions To increase the accuracy of prediction and properly decrease the number of fuzzy rules, N was selected as follows: • N = for input variables: spindle speed, feed rate, depth of cut, and vibration • N = for the output variable Ra The ε, as shown in Equation 14.10, was assigned the value of 0.01 The degree assigned by a human expert, µ(E')is 1, since all the experiments assume the tool to be in good condition, free from chatter or tool-wear conditions during data collection 16.5 Testing Results and Conclusions 16.5.1 Testing Results After the experiment was conducted and both the ANN and FN systems were developed, testing experiments were designed to test the predictive ability of this model The two sets of testing data mentioned in Section 3.1 were used for testing the accuracy of these two systems Using the same four variables (spindle speed, feed rate, depth of cut, and the VAPR) as independent variables or input neurons, predicted roughness Ra values (response variable or output neuron) were generated for either the ANN or FN model Tables 16.2 and 16.3 contain the measured Ra values and their predicted Ra values, as well as the percentage deviation of both ISRR systems with the 92-item testing set and the 36-item flexible testing set, respectively Considering the percentage deviation, both models performed well in their ability to predict the roughness of machined surfaces In summary, the prediction accuracy of ISRR-FN, ISRR-ANN 4-5-1, and ISRR-ANN 4-7-7-1 models are 95.78%, 95.87%, and 99.27%, respectively 16.5.2 Conclusions The fuzzy logic and neural-networks-based ISRR models demonstrated that learning and reasoning capabilities could be used for an in-process surface roughness recognition system With better than 95% ©2001 CRC Press LLC TABLE 16.2 Testing Data Set—92 Samples Spindle Speed Feed Rate Depth of cut Vibration Ra Ra ANN 4-5-1 Ra ANN 4-7-7-1 750 750 1500 1000 750 1250 1250 1000 1500 750 1500 1000 1000 1250 1250 750 1500 1000 1500 1250 1500 1000 750 1250 1500 1250 1500 1000 1250 1500 1000 1250 1250 1500 1000 1500 750 750 750 1250 1250 750 1000 1500 1500 1000 1500 1250 1250 1000 1500 1250 1250 1500 1500 24 18 12 12 24 24 18 12 12 12 24 6 12 18 18 12 12 18 18 18 12 18 12 18 24 12 18 24 18 12 18 6 24 12 24 24 24 12 18 18 18 24 24 24 12 12 18 3 3 5 5 1 5 5 3 1 3 5 3 1 5 5 5 5 5 0.1123 0.1488 0.1162 0.1330 0.1659 0.0977 0.1764 0.1674 0.1225 0.1509 0.1249 0.1859 0.1013 0.8855 0.1418 01045 0.1509 0.1647 0.1180 0.0974 0.1232 0.1691 0.1656 0.1252 0.1203 0.1266 0.1119 0.1391 0.2122 0.2164 0.0989 0.1253 0.1700 0.1372 0.1592 0.1151 0.1699 0.0899 0.1003 0.1740 0.1575 0.1705 0.1202 0.1900 0.1053 0.1610 0.1410 0.2160 0.2196 0.1932 0.0895 0.1249 0.1331 0.1155 0.1674 187 147 82 84 171 71 121 86 94 102 88 142 78 50 100 63 104 86 94 71 82 124 121 115 88 115 56 84 95 103 62 85 92 120 124 88 121 66 66 121 99 172 163 110 82 86 104 95 109 142 56 156 85 94 120 190 155 84 91 177 67 121 105 91 104 83 146 73 44 104 68 102 105 92 67 85 108 128 116 84 117 57 91 102 106 69 93 100 125 109 85 126 63 62 121 91 168 166 107 83 105 102 102 118 144 62 155 93 92 129 186 147 82 84 171 67 121 86 94 100 87 142 76 49 101 64 104 86 94 71 82 124 121 116 84 116 55 84 94 103 63 85 92 118 124 84 121 64 63 121 98 172 162 109 82 86 104 94 108 143 56 155 85 94 119 ©2001 CRC Press LLC TABLE 16.2 (continued) Testing Data Set—92 Samples Spindle Speed Feed Rate Depth of cut Vibration Ra Ra ANN 4-5-1 Ra ANN 4-7-7-1 1500 1500 1250 1500 1250 1500 1500 1000 1500 750 1500 1500 1250 1000 1500 1500 1000 750 1250 1000 1250 750 1500 1500 1250 750 1000 1000 750 1250 1500 1250 1000 1500 750 1000 1500 12 18 12 12 18 24 12 24 12 24 12 24 24 18 18 24 24 12 18 18 24 24 18 18 18 24 12 18 18 18 12 5 3 3 1 3 5 5 3 5 5 0.1101 0.1500 0.1288 0.1084 0.1253 0.0647 0.1359 0.1187 0.1151 0.1028 0.1394 0.1233 0.1316 0.1832 0.1339 0.1374 0.1011 0.1734 0.0759 0.1723 0.2145 0.1331 0.1300 0.1513 0.1857 0.0916 0.2035 0.0979 0.1446 0.1060 0.1516 0.1216 0.1599 0.1536 0.1622 0.0979 0.1183 82 104 100 56 85 37 87 163 82 63 120 82 156 92 120 120 138 147 50 153 109 94 120 104 121 72 142 138 147 63 104 156 92 87 121 138 94 84 102 106 57 93 39 96 166 84 68 125 85 153 92 126 125 139 145 46 157 120 101 118 102 119 69 141 140 155 67 102 155 92 98 129 140 92 82 105 101 55 85 36 87 163 82 65 118 82 155 92 118 118 137 147 49 153 108 96 119 105 121 70 143 137 147 64 105 155 91 87 129 137 94 accuracy of prediction, this system could be implemented as an in-process surface roughness prediction system; however, some directions for further research could make future implementation more effective: The cutting tool used in the study was a four-flute, high-speed steel cutter Use of the model with more diverse cutter materials or tools with a different number of flutes will benefit from further investigation The workpiece material used in the study was Aluminum 6061 T6 Different workpiece materials widely used in industry, such as carbon steel, aluminum alloys 380 and 390, or alloy steel, merit further exploration in order to build an overall ISRR system capable of extensive application in actual manufacturing environments Feedback control is necessary for automated production systems The ISRR should be a closedloop system, ensuring that output of the ISRR feeds back to the CNC machine center, causing the machine to adapt to the predicted real-time surface roughness value by adjusting the feed rate of the machine table Thus, a study of interface techniques between the ISRR system and the CNC machining center is also necessary ©2001 CRC Press LLC TABLE 16.3 Flexible Testing Data — 36 Samples Sample No Spindle Speed (rpm) Feed Rate (ipm) Depth of Cut (in.) 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 1500 1500 1500 1500 1500 1500 1500 1500 1500 1250 1250 1250 1250 1250 1250 1250 1250 1250 1000 1000 1000 1000 1000 1000 1000 1000 1000 750 750 750 750 750 750 750 750 750 9 15 15 15 21 21 21 9 15 15 15 21 21 21 9 15 15 15 21 21 21 9 15 15 15 21 21 21 0.01 0.03 0.05 0.01 0.03 0.05 0.01 0.03 0.05 0.01 0.03 0.05 0.01 0.03 0.05 0.01 0.03 0.05 0.01 0.03 0.05 0.01 0.03 0.05 0.01 0.03 0.05 0.01 0.03 0.05 0.01 0.03 0.05 0.01 0.03 0.05 VAPR Measured Ra (µin.) ISRR-FN Ra ISRR-FN 0.0883 0.1110 0.1056 0.1464 0.1256 0.1638 0.1473 0.1787 0.1980 0.1197 0.1381 0.1202 0.1338 0.1521 0.1444 0.1300 0.1726 0.1846 0.0911 0.1226 0.1426 0.1001 0.1487 0.1598 0.1034 0.1680 0.1687 0.0931 0.1255 0.1171 0.0950 0.1514 0.1530 0.1135 0.1624 0.1659 53 74 70 110 84 99 119 102 113 80 82 92 107 97 87 129 99 105 95 97 102 129 108 95 149 145 112 109 99 95 126 122 104 178 163 150 53 74 70 110 84 99 119 102 113 80 82 92 107 97 87 129 98 105 92 96 102 129 108 92 149 145 112 109 99 95 125 122 104 178 163 150 The ISRR systems used in this study were controlled by a personal computer system To increase operation speed, minimize size, reduce costs, and enhance efficiency, the ISRR could be created in an electronic model operated by a microprocessor, fuzzy chip, memory chip, and other related circuits These kinds of techniques are important and may be required for further development of ISRR technology for the next century References Armarego, E J A and Deshpande, N P., 1989, Computerized predictive cutting models for forces in end-milling including eccentricity effects, Annals of the CIRP, 38(1), pp 45-49 Babin, T S., Lee, J M., Sutherland, J M., and Kapoor, S G., 1985, A model for end milling surface topography, Proc of 13th North American Metalworking Research Conference, pp 362-368 Das, S., Roy, R., and Chaptopadhyay, A B., 1996 Evaluation of wear of turning carbide inserts using neural networks, Int J Mach Tools Manuf., 36(7), pp 789-797 ©2001 CRC Press LLC Huang, S J and Chiou, K C., 1996 The application of neural networks in self-tuning constant force control, Int J Mach Tools Manuf., 36, pp 17-31 Ismail, F., Elbestawi, M A., Du, R., and Urbasik, K., 1993, Generation of milled surfaces including tool dynamics and wear, ASME J Eng Ind., 115, pp 245-252 Kline, W A., DeVor, R E., and Shareef, I A., 1982, The prediction of surface accuracy in end milling, ASME J Eng Ind., 104, pp 272-278 Kolarits, F M and DeVries, W., 1989, A model of the geometry of the surface generated in end milling with variable process inputs, in Mechanics of Deburring and Surface Finishing Processes, J R Stango, and P R Fitzpatrick, Eds., ASME, PED, vol 38, pp 63-78 Martellotti, M E., 1941, An analysis of the milling process, Trans ASME 12:677-700 Martellotti, M E., 1945, An analysis of the milling process, part II — Down milling, Trans ASME, 67, pp 233-251 Melkote, S N and Thangaraj, A R., 1994, An enhanced end milling surface texture model including the effects of radial rake and primary relief angles, ASME J Eng Ind., 116, pp 166-174 Montgomery, D and Altintas, Y., 1991, Machanism of cutting force and surface generation in dynamic milling, ASME J Eng Ind., 113 (1), pp 160-168 Sutherland and Babin 1985, You, S J and Ehmann, K F., 1989, Scallop removal in die milling by tertiary cutter motion, ASME J Eng Ind., 111, pp 213-219 Zhang, G M and Kapoor, S G., 1991, Dynamic generation of machined surface, part 1: Description of a random excitation system, ASME J Eng Ind., 113(3), pp 137-144 Zhang, G M and Kapoor, S G., 1991b, Dynamic generation of machined surface, part 2: Construction of surface topography, ASME J Eng Ind., 113(3), pp 145-153 ©2001 CRC Press LLC ...16 Neural Networks and Neural- Fuzzy Approaches in an In- Process Surface Roughness Recognition System for End Milling Operations 16.1 16.2 16.3 16.4 Joseph C Chen Iowa State University Introduction... of cutting force and surface generation in dynamic milling, ASME J Eng Ind., 113 (1), pp 160-168 Sutherland and Babin 1985, You, S J and Ehmann, K F., 1989, Scallop removal in die milling by... cutting models for forces in end- milling including eccentricity effects, Annals of the CIRP, 38(1), pp 45-49 Babin, T S., Lee, J M., Sutherland, J M., and Kapoor, S G., 1985, A model for end milling