TW201120592A - Method for optimizing generator parameters by taguchi method and fuzzy inference - Google Patents

Abstract

The present invention is to provide a method for optimizing generator parameters by Taguchi Method and Fuzzy Inference. The present invention use Taguchi Method to preliminarily find optimum parameters of a generator, and applies Fuzzy Inference to find remaining undetermined parameters. The present invention applies Fuzzy Inference to overcome a problem that Taguchi Method probably can not determine all parameters of a generator for multiple objectives, so the method in accordance with the present invention is more systematic and reliable.

Description

201120592 VI. Description of the invention: [Technical field to which the invention pertains] The present invention relates to a method for optimizing the parameters of a generator, and more particularly to a method for determining the optimum parameters of a generator by applying a Taguchi method and a fuzzy inference. [Prior Art]

According to, how to improve the power generation efficiency of generators has always been the research goal in this field, which means that high input efficiency is expected to achieve high power generation efficiency. However, in order to meet these two conflicting design goals, designers must find checks and balances. Point to improve the overall performance of the generator, and the performance of the generator is determined by the process parameters of the generator. Therefore, the optimization of the process parameters becomes the highest of the generator material. (4) The process parameters affecting the efficiency of the generator are equivalent. This makes it difficult to find the best combination of parameters. 'The decision of the past process parameters' mostly depends on the empirical calculation rules accumulated by the predecessors, and it is completed by constantly trying the mistakes and corrections, which not only consumes a lot of Manpower, cost and directly affect the delay of the production cycle, making production plants in a very disadvantageous position in this highly competitive market. In order to solve this problem, the predecessors introduced the sub-factor method, the full factor method and the partial factor method to find the optimization parameters. The above various methods have disadvantages such as insufficient systemization, low reproducibility and excessive complexity. 'Therefore, the predecessors had to find a more suitable parameter optimization method. Among them, the Taguchi method has been proved to be a very effective parameter optimization method's overview of the guide statistics, so that the operator can The number of experiments can get the best combination of parameters. Because it does not require complicated calculation process and allows multiple parameter variables at the same time, the process of finding the best parameter combination is simplified and systematic. 201120592 However, although the Taguchi method is a very effective and suitable method, but the design goal of both the power generation efficiency and the low cogging torque should be considered at the same time, it will show the deficiency of the traditional Taguchi method, which is also the beginning of the county. 'In the decision process of the eve parameters, the Taguchi method may not be able to determine all the best parameters. SUMMARY OF THE INVENTION The main purpose of this month is to provide a method for applying the Taguchi method and the modulo=inference to determine the optimal parameters of the generator. It is hoped that the design can be used to improve the optimization of the conventional generator parameters. Low sex and too complicated...etc. 2 The purpose of the present invention is as follows: The invention comprises the following steps: clamping a plurality of functional requirements and selecting appropriate generator parameters as control factors and each control factor comprises at least two different levels; Experiments with the Taguchi method are used to obtain the function of each control factor W. The data is optimized. The energy-enhanced data determines the optimal level of each control factor under each functional requirement, and the best under the functional requirements. Control sub-combination The optimal control factor for each functional requirement is combined with the optimal control factor for each functional requirement; If Han is simultaneously full, whether there is a control factor to be determined, if all control--, the parameter design is completed. If there is still a use of Taguchi, the soil i-serving factor, meaning the child, then the control is used to find the best level of the control factor to be determined by the paste. Function 2 fuzzy inference 'takes the control factor to be determined as the input variable, each *, as the input iH variable, and defines the input paste set, where the fuzzy rule output variable is loose. That is, the quasi-involvement variable 201120592 is defined; the fuzzy rule is defined, and the sub-set of each input variable is substituted into the fuzzy rule to obtain the permutation and combination of all input variable sub-sets, and the output variable result deduced by each combination; The results of the self-fuzzy inference determine the round-in variables that satisfy the functional requirements at the same time, and the optimal control factors to be determined are obtained. The optimal control factors obtained by the Q 5 Taguchi method and the fuzzy inference are used to obtain the multi-purpose optimized manufacturing parameters. . _ 纟发 a month using the Taguchi method to find the optimal parameters of the generator, using its systemization, streamlined and clear advantages to achieve the purpose of optimization with the right amount of experiments, and with fuzzy inference to overcome the Taguchi method can not be targeted The two-rejection function requires the complete parameter optimization to complement all the parameters that cannot be determined by the Taguchi method, and thus provides a system optimization, reproducibility, and parameter optimization method that can simultaneously satisfy multiple functional requirements. [Embodiment] The present invention applies the Taguchi method and fuzzy inference to determine the generator parameters, and uses software simulation and real machine experiments to prove that the generator applying the parameters determined by the present invention has better performance, and its operation flow After using the computer simulation software RMxprt to construct the model of the generator prototype, based on the model, the Taguchi method and fuzzy inference are used to discuss how to modify the parameters of the permanent magnet generator and make the cogging torque of the prototype generator. Minimize and maximize power generation efficiency. For the card, please refer to the second figure. The generator (2) includes a stator (21) and a rotor (22) pivoted in the stator (21). 21) The inner edge is distributed with stator slots (2彳彳) arranged side by side. The parameters and data of the prototype generator are shown in Table 1 and Table 2. The performance of the prototype hair extension 201120592 is shown in Table 3. .

— Rated power 400 W Number of poles 12 Phase number 3 Air gap 1mm Stator slot number 36 " — Core thickness 3 0mm Coil number 36 Winding mode L Connect double stack around rotor inner diameter 17mm Stator inner and outer diameter 96.5/143.5mm Magnet material to pole ratio NdFeB N33SH (0.8 times pole pitch) Table 2 Performance Residual magnetic number density Br mT (KG) Around the coercive force bHcKA / m (KOe) Internal 禀 dwarf force iHc KA / m (KOe) Maximum magnetic energy product (BH)maxKJ/m3 (MGOe) Maximum operating temperature TW °c N-35 11.7-12.1 >868 (>10.9) >955 (5) 2) 263-287 (33-36) 80 N-38 12.1-12.5 &gt ;899 (>11.3) >955 (>12) 287-310(36-39) 80 N-40 12.5-12.8 >923 (>11.6) >955 (lack 12) 318-342 (38 -41) 80 N-42 12.8-13.2 >923 (>11.6) >955 (12) 318-342 (40~43) 80 N-45 13.2 ~13.8 >876 (>11.0) > 955 (212) 342-366 (43-46) 80 N-48 13.8-14.2 >835 (>10.5) >876 (11) 366-390 (46-49) 80 N-50 14.2-14.5 &gt ;835 (>10.5) >876 (^11) 374-406 (47-51) 80 N35H 11.7 ~12.1 >868 (>10.9) >1353 (M7) 263-287 (33~36) 120 N38H 12.1-12.5 >89 9 (>11.3) >1353 (M7) 287-310 (36~39) 120 N40H 12.4-12.8 >923 (>11.6) >1353 (M7) 302-326 (38-41) 120 N42H 12.8 -13.2 >955 (>12.0) >1353 (217) 318-342 (40-43) 120 N33SH 11.3-11.7 >844 (>10.6) >1592 (^20) 247-272 (31- 34) 150 N35SH 11.7-12.1 > 876 (>11.0) >1592 (^20) 263-287 (33-36) 150 N38SH 12.1-12.5 >907 (>11.4) >1592 (^20) 287-310(36-39) 150 N40SH 12.4-12.8 >939 (>11.8) >1592 (^20) 302-326 (38~41) 150 N28UH 10.2-10.8 >764 (> 9.6) &gt ;1990 (^25) 207-231 (26~29) 180 N30UH 10.8-11.3 >812 (>10.2) >1900 (^25) 223-247 (28-31) 180 N33UH 11.3-11.7 >852 (>10.7) >1990 (^25) 247-271 (31-34) 180 201120592 ------------ Table 3 ---_ Full Load - Specification Data Resistance (ohm) 2 ~ ~〇7) ~ ------ 25.1 342 !流(Α) '~ - 14.8 167 Friction loss (wT——-2.737 1 iron loss (W) ' " 19.6483 Copper loss (W) " 10 1.2353923 Total wear (w) '一- " 123.6207923 W output power (W, ~ 372.4059 wheel power (W — — — — — — — — — — — — — — — — — — — — " 4.3061

δ month refers to the first figure, which is a schematic flow chart of the method for applying the Taguchi method and the fuzzy push to determine the optimal parameters of the generator, which comprises the following steps: 疋 Multiple functional requirements, and select - appropriate generator As a control factor _), 控制, each control factor contains at least two different # level 'actual operation' as follows: ... This is better (four) defines two functional requirements 'i. generator efficiency, 丨丨 · Cogging torque, and burying _ ^ 'Select the parameters that may affect the most characteristic of the generator as a control factor. Please refer to (4) Table 4, the classification is 03: 01 and 08: 〇Q, respectively. The price pole ratio (β 〇9), Β. chute width (the position of the chute, the oblique half groove width and the oblique-slot knife α are the n3〇sh, n33sh and n35sh:), °_ magnet Materials (levels are magnets of the same strength), 〇. Coil resistance (bit 201120592 40匝), E. Stator slot type (as shown in the third figure, 36匝, 38 祖, and the fifth figure respectively) The levels are 4 ΤΥΡΘ·1 (2113), Type_2 (211b) and

Type 3 (211c) groove type), "wire diameter width (levels are 〇Mm, 0.8mm and 0.9mm), g. gullet opening width (levels are 1〇, 1.1mm and 1.2mm respectively) And η. air gap average width (levels are 咖8 coffee, 0.9_ and 1.0 coffee, respectively), wherein, except for the control factor a, there are only two levels, and each control factor is set to three levels. Control factor description level 1 level 2 position 3 A slot pole ratio 03:01 08:09 B chute width no chute oblique half slot width oblique slot width C magnet material n35sh n33sh n30sh D coil number 36 38 40 E Stator Groove Type-1 Type-2 Type-3 F Wire diameter 0.7mm 0.8mm 0.9mm G Cogging opening width 1.0mm 1.1mm 1.2mm Η Air gap average width 0.8mm 0.9mm 1.0mm Note: Thick The frame line is the parameter used by the prototype generator. For the defined complex function requirements, the Taguchi method experiment (1 02) is used to obtain the functional quantified data of different levels of each control factor. The actual operation is as follows: The preferred embodiment selects the use of ^1吣8) orthogonal tables, respectively for functional requirements, the efficiency of the generator, and Can demand 丨丨. Cogging torque to do two sets of Taguchi side 201120592 method experiment 'and each group of Taguchi square table 5 and Table 6, where words 1 to 3, for example, the ratio is based on level 1, meaning That is, the first method experiment contains 18 tests, respectively, for example, the level of each control factor is the control factor A of the first test of the Arabic number (the slot ratio used in the slot test is 03: 01). The control factor A (slot pole ratio) of the first test is level 2, which is the ratio of the slot number used in the first test to 08:09, and so on.

10 201120592

Table 6

3. The functional quantized data determines the optimal level of each control factor under each functional requirement, and then constitutes the optimal combination of control factors under the functional requirements. The actual operation is as follows: Using Table 5, the respective control factors are calculated separately. The average value of the efficiency S/N of the level, such as the average S/N of the efficiency of A1 is obtained by averaging the S/N values of the first to the ninth test, and the average S/N of the efficiency of B1 is the first to the first The S/N values of the 3rd and 10th to 12th tests are averaged, and the rest of the average S/N values of the 201120592 standards are shown in Table 7.

First, the selection of the effect #S/N average is as shown in the sixth figure, according to Π, select the most efficient control factor group qA1B3, c1D2, ..., H1); using Table 6, respectively calculate each control factor The average value of the cogging torque S/N of different levels is shown in Table 8. The average value of the cogging torque S/N of each control factor is plotted as shown in the seventh figure according to Table 8. Figure, select the combination of control factors for the minimum cogging torque (A1, B3, 〇2, D2, E2, F2, G2, H3). Table 8 ABCDEFGH Level 1 35.23 36.07 36.32 35.17 36.11 35.75 36.60 36.18 Level 2 36.00 35.56 34.92 35.15 35.30 34.94 35.03 36.41 Level 3 35.22 35.61 36.53 35.43 36.15 35.21 34.26 Floating Range 0.77 0.84 1.40 1.38 0.81 1.22 1.57 2.15 Floating Ranking (Rank) 8 6 3 4 7 5 2 1 12 201120592 'Extracting the control factor (103) that is optimized for each functional requirement that is met by the combination of the best control factors for the actual operation, such as four-way rutting, The following: The most efficient combination of control factors (A1, B3, C1, D2, E2, F2' G1 'H1)' and the control of the cogging torque (A1, B3, C2, D2' E2' F2, Fantasy, H3), you can get the control factor that simultaneously maximizes efficiency and minimizes cogging torque: (ai b3 d2 e2

F2) ' Designed to achieve the lowest cogging torque and highest efficiency. 5. If there is a control factor to be determined (] 〇 4), if all the control factors have been determined, the parameter design is completed; ^ there is still a control factor to be determined, which means that the Taguchi method cannot determine all the functions that meet the requirements of each function at the same time. The control factor j' uses fuzzy inference to find the optimal level of the undetermined control factor. The actual operation is as follows: Since the Taguchi method cannot determine the optimal level of the control factor C_magnet material, g·gap opening width and H. air gap average width, it means that all the simultaneous maximization of efficiency cannot be determined. And the control factor for minimizing the need for cogging torque. Therefore, the above three control factors C, G, and H need to use fuzzy push S to find the best level. 6. Perform fuzzy inference (105), using the undetermined control factor as the input variable 'each functional requirement as the output variable, and define the fuzzy set of the input variable and the output variable'. The actual operation is as follows: The average width of the air gap (mrTl) , the opening width of the magnetic groove (mm) and the strength of the magnet are used to count the variable 'efficiency (%) and cogging torque (N*m) as the output variable'. Please refer to the eighth to twelfth figures to define the input variables and The fuzzy set of output variables is: 13 201120592 Magnet strength = {weak (n30sh), normal (n33sh), strong (n35sh)}; cogging opening width (mm) = {small, medium, large}; air gap average Width (mm) = {small, medium, large}; efficiency (%) = {inferior, slightly worse, fair, good, excellent}; cogging torque (N*m) = {excellent, good, ordinary, slightly Big, too big}. Seven's define the fuzzy rules, and substituting the sub-sets of the input variables into the fuzzy rules, and obtain the permutation and combination of all the sub-sets, and the round-out results obtained by each combination. The actual operation is as follows: Month > Shown as the fuzzy rule of the preferred embodiment; please refer to the tenth=® ~ 〆~ 斤 indication, substituting the sub-sets of each input variable into the fuzzy rule, and obtaining the arrangement of the singular & And the 仏, J output variable result inferred by each combination can be further drawn into a fuzzy rule decision diagram. 1 ^普^& _) coffee (gap opening width is 'J,) and (average width is small rushing Qiu (efficiency is slightly worse) (gear twist ifi (magnetism strong ~ - 2 moment is good); The width of the opening is small (the average width of the air gap is in the middle (four) efficiency is slightly worse X cogging twist if (magnet - 3 ϊ), the tooth width is small, the average width of the air gap is large) - (efficiency is still Can) (cogging twist if (magnet ~ ~ a 4 moment is good); paving opening width is in (four) fire air gap average width is small efficiency is acceptable) (cogging twist if (magnetic? ί — 5 ^ist) Is tlanddP^^ti is t)ihen»i is 201120592

Table 9 - continued 6 if (magnet strength is weak) and (gear opening width is in) and (air gap average width is large) then (efficiency is acceptable) (window torque is good); 7 if (magnet The strength is weak) and (the cogging opening width is large) and (the air gap average width is small) then (efficiency is acceptable) (the cogging torque is good); · 8 if (the magnet strength is weak) and ( The slot opening width is large) and (the air gap average width is in) then (efficiency is acceptable) (the cogging torque is normal); 9 if (the magnet strength is weak) and (the slot opening width is is) and (air gap average width is large) then (efficiency is acceptable) (cogging torque is normal); 10 if (magnetism is ordinary) and (gear opening width is small) and (air gap average width is small) Then (efficiency is good) (cogging torque is excellent); 11 if (magnet strength is normal) and (gear opening width is small) and (air gap average width is) then (efficiency is excellent) (cogging Torque is excellent); 12 if (magnet strength is normal) and (gear opening width is small) and (air gap average width is large) then (efficiency is excellent) (cogging torque is good); 13 if (magnet Strength is normal) and (the slot opening width is in) and (the air gap average width is small) t Hen (efficiency is excellent) (cogging torque is good); 14 if (magnet strength is normal) and (gear opening width is) and (air gap average width is) then (efficiency is good) (cogging torque) Is excellent); 15 if (magnet strength is normal) and (gear opening width is in) and (air gap average width is large) then (efficiency is good) (cogging torque is good); 16 if (magnet strength Is normal)and (the cogging opening width is large) and (the air gap average width is small) then (efficiency is good) (cogging torque is good); 17 if (magnet strength is normal) and (gear opening width is Large)and (air gap average width is) then (efficiency is good) (cogging torque is good); 18 if (magnet strength is normal) and (gear opening width is large) and (air gap average width is large )then (efficiency is acceptable) (cogging torque is good); 19 if (magnet strength is strong) and (gear opening width is small) and (air gap average width is small) then (efficiency is good) ( Cogging torque is normal); 20 if (magnet strength is strong) and (gear opening width is small) and (air gap average width is) then (efficiency is good) (cogging torque is good); 21 if (magnet strength is strong) and (gear opening width is small) and (air gap Average width is large) then (efficiency is acceptable) (cogging torque is normal); 22 if (magnet strength is strong) and (gear opening width is in) and (air gap average width is small) then (efficiency) Is still) (cogging torque is normal); 23 if (magnet strength is strong) and (gear opening width is in) and (air gap average width is) then (efficiency is acceptable) (cogging torque) Is ordinary); 24 if (magnet strength is strong) and (gear opening width is in) and (air gap average width is large) then (efficiency is acceptable) (cogging torque is slightly larger); r 15 201120592 25 Table 9 - continued if (magnet strength is strong) and (gear opening width is large) and (air gap average width is small) then (efficiency is slightly worse) (cogging torque is slightly larger); 26 if ( The magnet strength is strong) and (the cogging opening width is large) and (the air gap average width is in the middle) then (the efficiency is slightly worse) (the cogging torque is normal); 27 if (the magnet strength is strong) and ( The slot opening width is large) and (the air gap average width is large) then (efficiency is slightly worse) (the cogging torque is slightly larger); M*· _ • w-一_> - VIII, self-fuzzy inference results , determine the input variables that satisfy both functional requirements (1 06), and get The optimal control factor to be determined, the actual operation is as described in φ: Observe the fuzzy rule decision diagram, extract the input variable combination that simultaneously satisfies the maximum efficiency and minimize the cogging torque, and obtain the optimized control factor·· The best combination of control factors is (C2, G1, H2). 9. Combine the best control factor (1 07) obtained by Taguchi method and fuzzy inference to obtain multi-purpose optimized manufacturing parameters: the best combination of parameters is (A1, B3, C2, D2, E2, F2, G1, H2), meaning that the optimal parameters of the generator are: • Control factor A: slot ratio is 03:01; control factor B: chute width is oblique groove width; control factor C: magnet material is n33sh; D: the number of turns of the coil is 38匝; the control factor E: the stator slot type is Type-2; the control factor F: the wire diameter is 0.8 mm; the control factor G: the slot opening width is 1.0 mm; the control factor Η: The average air gap width is 0.9 mm. 16 201120592 After modifying the generator prototype according to the optimized parameters determined in this embodiment, the simulation is performed by RMxprt software. The simulation results are shown in the table. Comparing Table 3 and Table 10, the results show that the output efficiency is not optimized before. 75.07% increased to 84.7528% 'The cogging torque was reduced from 0.3016 (N*m) before the original unoptimized to 0.2616 (N*m). Table 10 Full Load Specifications 祎 Resistance (ohm) 2 Voltage (V) 26.9147 Current (A) 14.1324 Friction Loss (W) 1 .737 1 Iron Loss - (W ) 12.6483 Copper Loss (W ) 54.0434 1 63 5 Total Loss (W 68.4288 1 63 5 Output power (W ) 3 80.3693063 Input power (W ) 448.798 1 226 Efficiency (%) 84.752873 75 Cogging torque (N .m ) 0.26 1 6 Power factor 1 Synchronous speed (rpm) 1100 Rated torque (N .m ) 3 .896 1 and according to the above optimized parameters for real machine production, and test its performance output data as shown in Table 11, observe the data when the speed is llOO (RPM), compared with the simulation results of Table 10 It is found that the cogging torque error of the actual machine test and the simulation result is about 3.11%, the error of the output voltage is about 〇YQ%, the error of the output current is about 6.28%, and the error of the efficiency is about 3. .... ... 17 201120592 The error is within acceptable limits, thus demonstrating the feasibility and accuracy of the present invention.

Table 11 Speed Resistance View Voltage Current Output Power Turn In Power Efficiency CRPM) (ohm) ((Ψτη) (V) (I)--- m (W) (%) 100 2 0.56 1.38 0.82 1.1316 5.8643 19.296416 200 2 0.86 4.06 2.19 8.8914 18.0117 49.364353 300 2 1.19 6.76 3.6 24.336 37.3849 65.095764 400 2 1.56 9.4 4.92 46.248 65.3450 70.775039 600 2 2.26 14.47 6.746 97.61462 141.9998 73.304532 700 2 2.62 17 8.81 149.77 192.0558 77.982517 800 2 2.96 19.5 10.11 197.145 247.9761 79.501591 900 2 3.29 22 11.4 250.8 310.0749 80.883675 1000 2 3.61 24.3 12.6 306.18 378.0379 80.991859 1100 2 3.93 26.7 13.89 370.863 452.7031 81.921901 1200 2 4.23 29.1 15.08 438.828 531.5570 82.555206 [Simple description of the diagram] The first picture: the block diagram of the creation. The second picture: for the hair Schematic diagram of the motor. The third figure is a schematic diagram of the Type-1 stator slot of the generator. The fourth picture is a schematic diagram of the Type-2 stator slot of the generator. Figure 5: Type-3 stator slot for the generator Schematic diagram 18 201120592 Intent sixth picture: Rose for each control factor Rate S/Ν average value Figure 7 Value diagram: Cogging torque S/Ν average for each control factor

Figure 8: Schematic diagram of the fuzzy set of magnet strength. The ninth figure is a fuzzy set of the width of the slot opening. Figure 10: Schematic diagram of the fuzzy set of the average width of the air gap. Dijon Figure: Schematic diagram of the fuzzy set of power generation efficiency. Figure 12: Schematic diagram of the fuzzy set of cogging torque. Thirteenth figure: A fuzzy rule decision diagram. [Explanation of main component symbols] (2) Generator (21) Stator (211) (211a) (211b) (211c) Stator slot (22) Rotor

Claims (1)

201120592 VII. Scope of application: • A method of applying the 〇 method and fuzzy inference to determine the number of generators, which includes the following steps: m 疋 复 plural functional requirements, and select the appropriate generator control factor, and each control factor contains At least two different levels; for the defined complex functional requirements, the Taguchi method is used to quantify the data to different levels of each control factor; The self-powered monthly data determines the good level of each control factor under each functional requirement: and then the optimal control factor combination under the demand of the cylinder; the optimal control factor combination for comparing the functional requirements, extracting each The control factor that optimizes the functional requirements; , / a judgment whether there is a control factor to be determined, if all the control factors have been determined, then the parameters are designed; if there is still a control factor to be determined, it means that the method can not be determined by the field σ method ^All control factors that satisfy the requirements of each function' use the leakage inference theory to find the undetermined control factor & the optimal level; & 仃 仃 仃 推 , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , And defining the fuzzy set σ of the input variable and the output variable, wherein the subset of the fuzzy rule is the order of the input variables; I ^ fuzzy rule, and sub-sets the input variables into the fuzzy rule' Input array combination of variables, and output variable results inferred by each combination; The number of input variables' functional requirements of optimum control of the factor to be determined;, -. 5 The optimal control factor obtained by the Taguchi method and the fuzzy inference can be used to optimize the manufacturing parameters of 201120592 to multi-purpose. 2. The application of the Taguchi method and the fuzzy inference to determine the optimal parameters of the generator as described in the scope of claim 1 The method wherein the functional requirements include a generator efficiency and a cogging torque, the control factor including a slot ratio, a chute width, a magnet material, a number of turns of the coil, a stator slot shape, a wire diameter, and a slot opening twist And the average width of the air gap. _ VIII, schema · (such as the next page) twenty one