**4.2 Neuro fuzzy subcontractor evaluation system**

This section presents the methodology of Neuro Fuzzy subcontractor evaluation before sending to the cloud production online monitoring (CPOM). This method is successfully validated and workable for automotive rubber subcontractor. The criteria include capacity, availability, quality, delivery, productivity, back order control. This study concerns only the monthly audit and yearly evaluation. As previous mentioned, the monthly is important because it affects the daily production performance. It is weight 70% of the full mark of the total evaluation procedure. Therefore, the yearly evaluation weight 30%. The subcontractors are

method. It is not only cost reduction but also the quality improvement. The delivery plan is monitored by the SCM department of the 1st tier following with the QC inspection and packing progress in order to predict the efficiency of the supply chain management. All of the parts required on the exact due date are delivered to the 1st tier inventory which is controlled by RFID and barcode inside the centralized data based and linked to the OEM. The final step is to close the P&D order and evaluated the performance. The concept of CPOM, detailed in 4.3, system is applied to the 4 last steps of the SCM. First is the delivery plan. Then, the QC& packing work station and stock control. The 3rd operation is delivery parts to customers which is linked to the 1st tier inventory control. The last operation is to close the

**Figure 5** shows production flow for supply chain collaborative of the 1st tier and the subcontractors. Based on the functional department. There are 5 departments which are defined in the same supply chain; engineering, planning supply chain, subcontractor and QC. The P&D order starts from the planning department and sends to the supply chain department and passes to the subcontractor. The engineering department prepares molds and operation standard procedure which is obtained from the new model testing at the real environment. The mold is sent to the subcontractor and tested for completion quality. Then, the subcontractor begins the preproduction and first lot experiment. The result is recorded and sent to the

P&D order which is linked to evaluation system.

*The cycle of subcontractor in automotive supply chain management.*

*Concepts, Applications and Emerging Opportunities in Industrial Engineering*

**Figure 4.**

**218**

#### *Concepts, Applications and Emerging Opportunities in Industrial Engineering*

evaluated with the same criteria and get results in 4 levels of satisfaction; A, B, C, and D. The grade A and grade B are passed with satisfaction whereas the grade C and the grade D are fail but they can improve the performance under the time and quality condition. There are 4 criteria are measured; 1. delivery on-time; 2. on quantity of the P&M order; 3. quality of the parts without any defect; 4. Back order clearance. The calculation performance is divided into three categories; delivery (D), back order clearance (B), and quality (Q). The function is shown in the Eq. 1, Eq. 2, and Eq. 3. Each criterion is applied and evaluated by every part. The performance evaluation level is given by score.

$$P(\mathcal{D}) = \sum\_{i=1}^{i=n} \mathbf{D}i \sum\_{j=1}^{j=n} \mathbf{P}j.(\dot{\mathbf{t}}\mathbf{1} + \dot{\mathbf{t}}\mathbf{2} + \dot{\mathbf{t}}\mathbf{3} + \cdots \dot{\mathbf{t}}\mathbf{n})\tag{3}$$

but can handle the back-order clearance for the next delivery. The less back-order clearance is the more satisfaction. It is divided into 5 levels; excellent, very good, good, moderate and fails. Each level can get different scores. The 1st level is excellent which holds the back-order between 0 to 3% and get 10 marks. The 2nd, 3rd, 4th level hold the back-order clearance by 4–10%, 11–20%, and over 21% which can

*Automotive Industrial Supply Chain Performance Evaluation under Uncertain Constraints…*

½ � xjx part defect ð Þ≤3%; to get 10 marks ½ � xjx NCR clearance ð Þ≤3%; to get 5 marks

½ � xjx customer claim ð Þ≤ 3%; to get 5 marks ½ � xjx quick respond ð Þ≥ 95%; to get 5 marks

The set of quality performance shows above that is x when the x is in the various conditions. If the part defect is ≤3%, then it will get 10 marks. If the NCR clearance is ≤3%, then it will get 5 marks. If the part report inspection is ≥95%, it will get 5 marks. If the customer claim is ≤3%, it will get 5 marks. If the quick respond is

The final evaluation for subcontractor evaluation is combined the three criteria; delivery, quality and back order clearance. The total score and the satisfaction level

**Table 1** shows the total scores and the evaluation criteria. The total score is 80 marks. This is for the monthly evaluation. The remain score is for yearly audit which gives the total score is 30 marks. However, this paper concerns only the monthly evaluation. Presently, the cluster automotive rubber part is evaluated by manual. It is time consuming, high effort, slow problem solving, and cannot deal

As mentioned from the literature review that fuzzy logic can deal with uncertainty. This paper adopts the fuzzy logic model to deal with the subcontractor performance evaluation. The set representation selects a membership function. Fuzzy set design is a triangular consisting of crisp subsets. The fuzzy linguistic terms are design in 5 levels; very high, high, moderate, low and very low. The

**Criteria Marks Score Grade Satisfaction level** Delivery 30 70–80 A Excellent Quality 45 55–69 B Good Back order clearance 10 40–54 C Moderate Total 80 < 40 D To be improved

with uncertainty circumstances and predict future events.

conditional fuzzy rules are as the follows.

**Table 1.**

**221**

*The score and evaluation criteria.*

1. If delivery is low, then performance is low

3. If delivery is high, then performance is high.

2. If delivery is moderate, then performance is moderate.

4. If delivery is very high, then performance is very high.

½ � xjx part report inspection ð Þ≥ 95%; to get 5 marks

get the marks of 7.5, 5, 2.5 respectively.

*DOI: http://dx.doi.org/10.5772/intechopen.93679*

8 >>>>>>><

>>>>>>>:

P Qð Þ¼

≥95%, it will get 5 marks.

are shown as the followings.

$$P(\mathbf{B}) = \sum\_{i=1}^{i=n} \mathbf{B}i \sum\_{j=1}^{j=n} \mathbf{P}j.(\dot{\mathbf{t}}\mathbf{1} + \dot{\mathbf{t}}\mathbf{2} + \dot{\mathbf{t}}\mathbf{3} + \cdots \dot{\mathbf{t}}\mathbf{n})\tag{4}$$

$$P(\mathbf{Q}) = \sum\_{i=1}^{i=n} \mathbf{Q}i \sum\_{j=1}^{j=n} \mathbf{P}j.(it\mathbf{1} + it\mathbf{2} + it\mathbf{3} + \cdots it\mathbf{n})\tag{5}$$

When.

P(D) = delivery performance (%), P(B) = back order performance (%), P(Q) = quality performance (%), it = items to be produced (item consists of many pieces), Di = number of items to be delivered, Bi = number of back order to be delivered, Qi = quantity to be produced, Pj = number of pieces to be produced

$$\mathbf{P}(\mathbf{D}) = \begin{cases} \begin{bmatrix} \mathbf{x}|\mathbf{x} \text{ (delivery performance)} \ge 99\%; \text{to get 15 marks} \end{bmatrix} \\ \begin{bmatrix} \mathbf{x}|\mathbf{x} \text{ (delivery performance)} = 90\text{-}98\%; \text{to get 12 marks} \end{bmatrix} \\ \begin{bmatrix} \mathbf{x}|\mathbf{x} \text{ (delivery performance)} = 80\text{-}89\%; \text{to get 9 marks} \end{bmatrix} \\ \begin{bmatrix} \mathbf{x}|\mathbf{x} \text{ (delivery performance)} = 70\text{-}79\%; \text{to get 6 marks} \end{bmatrix} \\ \begin{bmatrix} \mathbf{x}|\mathbf{x} \text{ (delivery performance)} < 70\%; \text{to get 0marks} \end{bmatrix} \end{cases}$$

The function set of P(D) shows the performance calculation of delivery which infers that the subcontractors have capability to handle the routinely received orders. The score is divided into 5 levels. The level 1 is excellent which can deliver more than 99% of the total order and pieces and get 15 marks. The level 2 is very good which can deliver between 90 and 98% of the total order and pieces and get 12 marks. The level 3 is good which can deliver between 80 and 89% of the total order and pieces and get 9 marks. The level 4 is moderate which can deliver between 70 and 79% of the total order and pieces and get 6 marks. The level 5 is fail which can deliver less than 70% of the total order and pieces and get 0 mark. The function set of the back-order release is presented in the next step.

$$\mathbf{P(B)} = \begin{cases} \begin{bmatrix} \mathbf{x}|\mathbf{x} \text{ (delivery performance)} \le 3\mathbf{\forall}; \text{to get } \mathbf{10} \text{ marks} \end{bmatrix} \\ \begin{bmatrix} \mathbf{x}|\mathbf{x} \text{ (delivery performance)} = \mathbf{4} \text{-} \mathbf{10}\mathbf{\forall}; \text{to get } \mathbf{7.5} \text{ marks} \end{bmatrix} \\ \begin{bmatrix} \mathbf{x}|\mathbf{x} \text{ (delivery performance)} = \mathbf{11} \text{-} \mathbf{20}\mathbf{\forall}; \text{to get } \mathbf{5} \text{ marks} \end{bmatrix} \\ \begin{bmatrix} \mathbf{x}|\mathbf{x} \text{ (delivery performance)} \ge \mathbf{20}\mathbf{\forall}; \text{to get } \mathbf{2.5} \text{ marks} \end{bmatrix} \end{cases}$$

The set clearance of back order P(B) shows the performance score of the backorder clearance. It means that the subcontractor cannot achieve the delivery plan

*Automotive Industrial Supply Chain Performance Evaluation under Uncertain Constraints… DOI: http://dx.doi.org/10.5772/intechopen.93679*

but can handle the back-order clearance for the next delivery. The less back-order clearance is the more satisfaction. It is divided into 5 levels; excellent, very good, good, moderate and fails. Each level can get different scores. The 1st level is excellent which holds the back-order between 0 to 3% and get 10 marks. The 2nd, 3rd, 4th level hold the back-order clearance by 4–10%, 11–20%, and over 21% which can get the marks of 7.5, 5, 2.5 respectively.

> P Qð Þ¼ ½ � xjx part defect ð Þ≤3%; to get 10 marks ½ � xjx NCR clearance ð Þ≤3%; to get 5 marks ½ � xjx part report inspection ð Þ≥ 95%; to get 5 marks ½ � xjx customer claim ð Þ≤ 3%; to get 5 marks ½ � xjx quick respond ð Þ≥ 95%; to get 5 marks 8 >>>>>>>< >>>>>>>:

The set of quality performance shows above that is x when the x is in the various conditions. If the part defect is ≤3%, then it will get 10 marks. If the NCR clearance is ≤3%, then it will get 5 marks. If the part report inspection is ≥95%, it will get 5 marks. If the customer claim is ≤3%, it will get 5 marks. If the quick respond is ≥95%, it will get 5 marks.

The final evaluation for subcontractor evaluation is combined the three criteria; delivery, quality and back order clearance. The total score and the satisfaction level are shown as the followings.

**Table 1** shows the total scores and the evaluation criteria. The total score is 80 marks. This is for the monthly evaluation. The remain score is for yearly audit which gives the total score is 30 marks. However, this paper concerns only the monthly evaluation. Presently, the cluster automotive rubber part is evaluated by manual. It is time consuming, high effort, slow problem solving, and cannot deal with uncertainty circumstances and predict future events.

As mentioned from the literature review that fuzzy logic can deal with uncertainty. This paper adopts the fuzzy logic model to deal with the subcontractor performance evaluation. The set representation selects a membership function. Fuzzy set design is a triangular consisting of crisp subsets. The fuzzy linguistic terms are design in 5 levels; very high, high, moderate, low and very low. The conditional fuzzy rules are as the follows.

1. If delivery is low, then performance is low

2. If delivery is moderate, then performance is moderate.

3. If delivery is high, then performance is high.

4. If delivery is very high, then performance is very high.


**Table 1.** *The score and evaluation criteria.*

evaluated with the same criteria and get results in 4 levels of satisfaction; A, B, C, and D. The grade A and grade B are passed with satisfaction whereas the grade C and the grade D are fail but they can improve the performance under the time and quality condition. There are 4 criteria are measured; 1. delivery on-time; 2. on quantity of the P&M order; 3. quality of the parts without any defect; 4. Back order clearance. The calculation performance is divided into three categories; delivery (D), back order clearance (B), and quality (Q). The function is shown in the Eq. 1, Eq. 2, and Eq. 3. Each criterion is applied and evaluated by every part. The perfor-

mance evaluation level is given by score.

*P D*ð Þ¼ <sup>X</sup> *i*¼*n*

*P B*ð Þ¼ <sup>X</sup> *i*¼*n*

*P Q*ð Þ¼ <sup>X</sup>

When.

P Dð Þ¼

P Bð Þ¼

**220**

8 >>>><

>>>>:

8 >>>>>>><

>>>>>>>:

*i*¼**1**

*i*¼**1**

*i*¼*n*

*i*¼**1**

*Di* X *j*¼*n*

*Concepts, Applications and Emerging Opportunities in Industrial Engineering*

*Bi* X *j*¼*n*

*Qi* X *j*¼*n*

xjx delivery performance

xjx delivery performance

xjx delivery performance

xjx delivery performance

xjx delivery performance

of the back-order release is presented in the next step.

xjx delivery performance

xjx delivery performance

xjx delivery performance

xjx delivery performance

*j*¼**1**

*j*¼**1**

*j*¼**1**

P(D) = delivery performance (%), P(B) = back order performance (%), P(Q) = quality performance (%), it = items to be produced (item consists of many pieces), Di = number of items to be delivered, Bi = number of back order to be delivered, Qi = quantity to be produced, Pj = number of pieces to be produced

� �≥99%; to get 15 marks � �

� � <sup>¼</sup> <sup>90</sup>‐98%; to get 12 marks � �

� � <sup>¼</sup> <sup>80</sup>‐89%; to get 9 marks � �

� � <sup>¼</sup> <sup>70</sup>‐79%; to get 6 marks � �

� �<70%; to get 0marks � �

� �≤3%; to get 10 marks � �

� � <sup>¼</sup> <sup>4</sup>‐10%; to get 7*:*5 marks � �

� � <sup>¼</sup> <sup>11</sup>‐20%; to get 5 marks � �

� �≥20%; to get 2*:*5 marks � �

The set clearance of back order P(B) shows the performance score of the backorder clearance. It means that the subcontractor cannot achieve the delivery plan

The function set of P(D) shows the performance calculation of delivery which infers that the subcontractors have capability to handle the routinely received orders. The score is divided into 5 levels. The level 1 is excellent which can deliver more than 99% of the total order and pieces and get 15 marks. The level 2 is very good which can deliver between 90 and 98% of the total order and pieces and get 12 marks. The level 3 is good which can deliver between 80 and 89% of the total order and pieces and get 9 marks. The level 4 is moderate which can deliver between 70 and 79% of the total order and pieces and get 6 marks. The level 5 is fail which can deliver less than 70% of the total order and pieces and get 0 mark. The function set

*Pj:*ð Þ *it***1** þ *it***2** þ *it***3** þ ⋯*itn* (3)

*Pj:*ð Þ *it***1** þ *it***2** þ *it***3** þ ⋯*itn* (4)

*Pj:*ð Þ *it***1** þ *it***2** þ *it***3** þ ⋯*itn* (5)

5. If back order is very low, then performance is very high.

6. If back order is low, then performance is high.

7. If back order is moderate, then performance is moderate.

8. If back order is high, then performance is low.

9. If back order is very high, then performance is very low.

**Figure 6** shows the fuzzy logic model for performance evaluation. It consists of two inputs and two outputs. The inputs are delivery and back order whereas the outputs are performance evaluation of the delivery and back order. The model is performed by using MATLAB which is explained later. Fuzzy logic system is an inference engine using for carrying the results based on the input conditions.

**Figure 7** shows the fuzzy inference system for delivery performance evaluation consisting of fuzzy input, fuzzification, and fuzzy output. The input membership function is divided into 4 ranges; Very high (VH), High (H), Medium (M), and Low (L). The Mandani method is selected with a triangular form.

**Figure 10** shows the FIS model for the back-order performance evaluation consisting of fuzzy input, fuzzy inference system and fuzzy output. The model is referred to the **Table 2**. Similarly, **Figure 11** shows the FIS model of the delivery, the back-order clearance membership function which is divided into 4 ranges; Low (L), medium (M), high (H) and very high (VH). The triangular form is selected. The total range is between 0 and 21. The Low range is from 0, 1.5, to 3.99. The Medium range is from 3.95, 7, to 10.99. The High range is from 11, 15, and 20 and

*Automotive Industrial Supply Chain Performance Evaluation under Uncertain Constraints…*

**Figure 12** shows the output membership function, which is divided into 4 levels

of Low, Medium, High and Very High. The total range is from 0 to10. The

the Very High range is from 20.1, 21, to 28.

*Membership function input of the delivery.*

*DOI: http://dx.doi.org/10.5772/intechopen.93679*

*Membership function output of the performance.*

*The FIS model for the back-order performance.*

**Figure 9.**

**Figure 10.**

**223**

**Figure 8.**

**Figure 8** shows the fuzzy membership function of the on-time delivery using triangular form. The divided range is referred to the **Table 1** but it needs to adjust a bit in order to fit with the actual form of the supply chain sample model. The lowlevel ranges are 69.5,75, and 80. The medium-level ranges are 79, 85, and 90. The high-level ranges are 89, 94.92, and 98.92. The very high-level ranges are 98,100,110.3. The total ranges are designed between 60 to 100%.

**Figure 9** shows the membership function of the output delivery. The triangle fuzzy model is selected. The Low-level ranges of 5.25, 5.75, and 6. The Medium-level ranges 8.5, 8.75, and 9. The High-level ranges of 11.5, 11.75, and 12. The Very High-level ranges 11.5, 11.75, and 12.

**Figure 6.**

*The fuzzy logic model for performance evaluation.*

**Figure 7.** *FIS model for delivery performance.*

*Automotive Industrial Supply Chain Performance Evaluation under Uncertain Constraints… DOI: http://dx.doi.org/10.5772/intechopen.93679*

**Figure 8.**

5. If back order is very low, then performance is very high.

*Concepts, Applications and Emerging Opportunities in Industrial Engineering*

7. If back order is moderate, then performance is moderate.

9. If back order is very high, then performance is very low.

Low (L). The Mandani method is selected with a triangular form.

98,100,110.3. The total ranges are designed between 60 to 100%.

High-level ranges 11.5, 11.75, and 12.

*The fuzzy logic model for performance evaluation.*

**Figure 6.**

**Figure 7.**

**222**

*FIS model for delivery performance.*

high-level ranges are 89, 94.92, and 98.92. The very high-level ranges are

**Figure 6** shows the fuzzy logic model for performance evaluation. It consists of two inputs and two outputs. The inputs are delivery and back order whereas the outputs are performance evaluation of the delivery and back order. The model is performed by using MATLAB which is explained later. Fuzzy logic system is an inference engine using for carrying the results based on the input conditions.

**Figure 7** shows the fuzzy inference system for delivery performance evaluation consisting of fuzzy input, fuzzification, and fuzzy output. The input membership function is divided into 4 ranges; Very high (VH), High (H), Medium (M), and

**Figure 8** shows the fuzzy membership function of the on-time delivery using triangular form. The divided range is referred to the **Table 1** but it needs to adjust a bit in order to fit with the actual form of the supply chain sample model. The lowlevel ranges are 69.5,75, and 80. The medium-level ranges are 79, 85, and 90. The

**Figure 9** shows the membership function of the output delivery. The triangle fuzzy model is selected. The Low-level ranges of 5.25, 5.75, and 6. The Medium-level ranges 8.5, 8.75, and 9. The High-level ranges of 11.5, 11.75, and 12. The Very

6. If back order is low, then performance is high.

8. If back order is high, then performance is low.

*Membership function input of the delivery.*

**Figure 10** shows the FIS model for the back-order performance evaluation consisting of fuzzy input, fuzzy inference system and fuzzy output. The model is referred to the **Table 2**. Similarly, **Figure 11** shows the FIS model of the delivery, the back-order clearance membership function which is divided into 4 ranges; Low (L), medium (M), high (H) and very high (VH). The triangular form is selected. The total range is between 0 and 21. The Low range is from 0, 1.5, to 3.99. The Medium range is from 3.95, 7, to 10.99. The High range is from 11, 15, and 20 and the Very High range is from 20.1, 21, to 28.

**Figure 12** shows the output membership function, which is divided into 4 levels of Low, Medium, High and Very High. The total range is from 0 to10. The

**Figure 9.**

*Membership function output of the performance.*

**Figure 10.**

*The FIS model for the back-order performance.*


Therefore, the model is designed to add quality to be another input membership

*Automotive Industrial Supply Chain Performance Evaluation under Uncertain Constraints…*

performance. It consists of 3 inputs; delivery, back order, and quality and one output. The output is overall performance evaluation. The sugeno is selected for the

**Figure 14** shows the neuro fuzzy inference (ANFIS) system for evaluate overall

**Figure 14** shows the neural training and learning data and the output. **Table 2** shows the overall input membership functions for the ANFIS system. The range represents the minimum and maximum value of the membership function. For example, the range of input 1 (delivery) is 6 to 15. The MF1 (Low range) represents 3 points of the triangle in the base line projection. They are 3, 6, and 9 respectively. **Figure 15** shows input membership functions plots. It consists of 4 levels; low, medium, high and very high. The low ranks between 0 and 2.5. The medium ranks between 2 to 5. The high ranks between 4.75 and 7.5. The very high is fallen over

**Figure 16** shows the input membership function of the delivery. It is divided into 4 levels; low, medium, high, and very high. The low ranks from 6 to 8. The medium ranks from 8 to 12. The high ranks from 12 to 14 and the very high ranks

**Figure 17** shows the input membership function plots. It is divided into 4 levels; low, medium, high, and very high. The low starts from 0 to 11. The medium ranks

function and adopt neuro-fuzzy system to solve the problem.

7.25. The total length is scoped between 0 and 10.

from over 14. The total length is bounded from 6 to 15.

*The neuro-fuzzy model applied for overall performance evaluation.*

*ANFIS of the neuro-fuzzy performance evaluation model.*

ANFIS rules of inference.

*DOI: http://dx.doi.org/10.5772/intechopen.93679*

**Figure 13.**

**Figure 14.**

**225**

#### **Table 2.**

*Case of subcontractor neuro-fuzzy performance evaluation.*

#### **Figure 11.**

*Membership function input for back order performance.*

**Figure 12.**

*Membership function output performance evaluation.*

membership function of the Low range is from 2.45, 2.5, to 2.75. Membership function of the Medium range is from 4.5,4.75, and 5. The membership function of the High range is from 7.12, 7.22, to 7.47. The membership function of the Very High range is from 9.5, 10, to 13.33.

**Figure 13** shows the FIS model is considered and evaluated individually. The ultimate goal of the fuzzy -based system is to get the overall performance evaluation. However, the overall performance has to include quality of the produce part. *Automotive Industrial Supply Chain Performance Evaluation under Uncertain Constraints… DOI: http://dx.doi.org/10.5772/intechopen.93679*

Therefore, the model is designed to add quality to be another input membership function and adopt neuro-fuzzy system to solve the problem.

**Figure 14** shows the neuro fuzzy inference (ANFIS) system for evaluate overall performance. It consists of 3 inputs; delivery, back order, and quality and one output. The output is overall performance evaluation. The sugeno is selected for the ANFIS rules of inference.

**Figure 14** shows the neural training and learning data and the output. **Table 2** shows the overall input membership functions for the ANFIS system. The range represents the minimum and maximum value of the membership function. For example, the range of input 1 (delivery) is 6 to 15. The MF1 (Low range) represents 3 points of the triangle in the base line projection. They are 3, 6, and 9 respectively.

**Figure 15** shows input membership functions plots. It consists of 4 levels; low, medium, high and very high. The low ranks between 0 and 2.5. The medium ranks between 2 to 5. The high ranks between 4.75 and 7.5. The very high is fallen over 7.25. The total length is scoped between 0 and 10.

**Figure 16** shows the input membership function of the delivery. It is divided into 4 levels; low, medium, high, and very high. The low ranks from 6 to 8. The medium ranks from 8 to 12. The high ranks from 12 to 14 and the very high ranks from over 14. The total length is bounded from 6 to 15.

**Figure 17** shows the input membership function plots. It is divided into 4 levels; low, medium, high, and very high. The low starts from 0 to 11. The medium ranks

**Figure 13.** *The neuro-fuzzy model applied for overall performance evaluation.*

**Figure 14.** *ANFIS of the neuro-fuzzy performance evaluation model.*

membership function of the Low range is from 2.45, 2.5, to 2.75. Membership function of the Medium range is from 4.5,4.75, and 5. The membership function of the High range is from 7.12, 7.22, to 7.47. The membership function of the Very

**[Input1] [Input2] [Input3]**

*Concepts, Applications and Emerging Opportunities in Industrial Engineering*

[0 2.5 5]

[2.5 5 7.5]

[5 7.5 10]

[7.5 10 12.5]

Name = 'input2' Range = [2.5 10] NumMFs = 4 MF1 = 'in2mf1':'trimf',

MF2 = 'in2mf2':'trimf',

Name = 'input3' Range = [1 40] NumMFs = 4 MF1 = 'in3mf1':'trimf',

MF2 = 'in3mf2':'trimf',

MF3 = 'in3mf3':'trimf',

MF4 = 'in3mf4':'trimf',

[12 1 14]

[1 14 27]

[14 27 40]

[27 40 53]

MF3 = 'in2mf3':'trimf',

MF4 = 'in2mf4':'trimf',

Name = 'input1' Range = [6 15] NumMFs = 4 MF1 = 'in1mf1':'trimf',

MF2 = 'in1mf2':'trimf',

MF3 = 'in1mf3':'trimf',

MF4 = 'in1mf4':'trimf',

*Case of subcontractor neuro-fuzzy performance evaluation.*

*Membership function input for back order performance.*

[3 6 9]

[6 9 12]

[9 12 15]

[12 15 18]

**Table 2.**

**Figure 11.**

**Figure 12.**

**224**

**Figure 13** shows the FIS model is considered and evaluated individually. The ultimate goal of the fuzzy -based system is to get the overall performance evaluation. However, the overall performance has to include quality of the produce part.

High range is from 9.5, 10, to 13.33.

*Membership function output performance evaluation.*

**Figure 15.**

*The input membership function of the back-order clearance.*

**Figure 19** shows neural training and learning. There are 3 inputs and 1 output. The training is set at 1000 iteration. There are 64 data. The final output is closely 70.

*Automotive Industrial Supply Chain Performance Evaluation under Uncertain Constraints…*

**Figure 20** shows the relation surface of the 3 criteria in 3 dimensions. It is found from the relation that the quality is the most influence to the performance whereas

**Figure 21** shows the rule viewer for the overall performance evaluation model for subcontractor. The data collection from the actual operation. The subcontractor data are inputted into the model and carried out with the result. As the sample above, the score of input 1 (delivery performance) given 15 marks meaning that the delivery performance is very high (99–100%). The 2nd input of the back-order clearance is received 10 marks meaning that this subcontractor performed very well. The percentage of back-order clearance is 1–3%. The 3rd input (input 3) is quality of product and service to the OEM agent and this subcontractor received 40 marks. The quality includes the products delivered quality to the OEM agent.

The error is 4.546 <sup>10</sup><sup>6</sup>

**Figure 18.**

**Figure 19.**

**227**

*Neural training and learning.*

.

*Rule viewer output performance of the delivery on time.*

*DOI: http://dx.doi.org/10.5772/intechopen.93679*

the delivery is lower following with the backorder clearance.

**Figure 16.** *The input membership function of the delivery.*

**Figure 17.** *The input membership function of the quality.*

from 11 to 23. The high ranks from 22 to 34. The very high ranks from over 33. The total length bounded from 0 to 45.

**Figure 18** shows the rule viewer for the example of the output of the delivery performance. There are four ranges of the input membership function. The example range is in the high which indicates 95.7 and the performance indicates very high. The given score is 11.7 marks in the 15 marks.

*Automotive Industrial Supply Chain Performance Evaluation under Uncertain Constraints… DOI: http://dx.doi.org/10.5772/intechopen.93679*

**Figure 18.** *Rule viewer output performance of the delivery on time.*

**Figure 19** shows neural training and learning. There are 3 inputs and 1 output. The training is set at 1000 iteration. There are 64 data. The final output is closely 70. The error is 4.546 <sup>10</sup><sup>6</sup> .

**Figure 20** shows the relation surface of the 3 criteria in 3 dimensions. It is found from the relation that the quality is the most influence to the performance whereas the delivery is lower following with the backorder clearance.

**Figure 21** shows the rule viewer for the overall performance evaluation model for subcontractor. The data collection from the actual operation. The subcontractor data are inputted into the model and carried out with the result. As the sample above, the score of input 1 (delivery performance) given 15 marks meaning that the delivery performance is very high (99–100%). The 2nd input of the back-order clearance is received 10 marks meaning that this subcontractor performed very well. The percentage of back-order clearance is 1–3%. The 3rd input (input 3) is quality of product and service to the OEM agent and this subcontractor received 40 marks. The quality includes the products delivered quality to the OEM agent.

**Figure 19.** *Neural training and learning.*

from 11 to 23. The high ranks from 22 to 34. The very high ranks from over 33. The

**Figure 18** shows the rule viewer for the example of the output of the delivery performance. There are four ranges of the input membership function. The example range is in the high which indicates 95.7 and the performance indicates very high.

total length bounded from 0 to 45.

*The input membership function of the quality.*

*The input membership function of the delivery.*

**Figure 15.**

**Figure 16.**

**Figure 17.**

**226**

*The input membership function of the back-order clearance.*

*Concepts, Applications and Emerging Opportunities in Industrial Engineering*

The given score is 11.7 marks in the 15 marks.

#### **Figure 20.** *The relation surface of the 3 criteria in 3 dimensions.*

#### **Figure 21.**

*The rule viewer for the performance evaluation model.*

It measures as ppm (part per million). The rank score is A, B, C, D. A is given by the ppm more than 90%. B is more than 80%. C is more than 70% and D is less than 70%. In addition, the quality is measured by defects, NCR (non-conformance record), back order clearance, communication, and claim from customer.

**Table 3** shows validation of the ANFIS comparison with actual performance calculation. They are 13 subcontractors are studied (A-K). The column #2 is given marks that calculated by manual. The column #3 and column #4 are also the given marks are calculated by manual. The column #5 and column #6 are determined by the fuzzy inference method. The column #7 and column 8 show the comparison between the actual performance and the neuro fuzzy subcontractor performance. It is found that the neural fuzzy system performs effectively with error 0%.

**Subcontractors**

**229**

A B C D E F G H

I J K L M **Table 3.** *The validation*

 *of the ANFIS comparison*

 *with actual* 

*performance.*

15

 7.5

 35

14.9

15

 10

 35

14.9

15

 10

 40

14.9

15

15

15

 10

 20

14.9

 10

 20

14.9

 10

 30

14.9

15

 10

 35

14.9

15

 10

 20

14.9

15

 10

 40

14.9

12

 10

 40

11.7

15

 10

 20

14.9

15

 10

 40

14.9

12

 7.5

 40

11.7

7.3 9.84 9.84 9.84 9.84 9.84 9.84 9.84 9.84 9.84 9.84 9.84

7.3

59 64.74 44.74 61.54 64.74 44.74 59.74 54.74 44.74 44.74 64.74 59.74

57.2

 **Delivery**

 **Back order Quality**

 **Fuzzy logic delivery**

 **Fuzzy logic back order Actual** 

**performance**

 **Neuro fuzzy** 

**subcontractor**

59.00 64.74 44.74 61.54 64.74 44.74 59.74 54.74 44.74 44.74 64.74 59.74 57.20

**performance**

 **Error%**

0

*DOI: http://dx.doi.org/10.5772/intechopen.93679*

0

0

0

0

0

0

*Automotive Industrial Supply Chain Performance Evaluation under Uncertain Constraints…*

0

0

0

0

0

0


#### *Automotive Industrial Supply Chain Performance Evaluation under Uncertain Constraints… DOI: http://dx.doi.org/10.5772/intechopen.93679*

**Table 3.** *Thevalidation*

 *of the ANFIS comparison*

 *with actual* 

*performance.*

It measures as ppm (part per million). The rank score is A, B, C, D. A is given by the ppm more than 90%. B is more than 80%. C is more than 70% and D is less than 70%. In addition, the quality is measured by defects, NCR (non-conformance record), back order clearance, communication, and claim from customer.

**Figure 20.**

**Figure 21.**

**228**

*The relation surface of the 3 criteria in 3 dimensions.*

*Concepts, Applications and Emerging Opportunities in Industrial Engineering*

*The rule viewer for the performance evaluation model.*

**Table 3** shows validation of the ANFIS comparison with actual performance calculation. They are 13 subcontractors are studied (A-K). The column #2 is given marks that calculated by manual. The column #3 and column #4 are also the given marks are calculated by manual. The column #5 and column #6 are determined by the fuzzy inference method. The column #7 and column 8 show the comparison between the actual performance and the neuro fuzzy subcontractor performance. It

is found that the neural fuzzy system performs effectively with error 0%.


**Table 4** shows the fuzzy rules of the ANFIS performance prediction. The 3 first columns are input and 4th column is output. The column 5–7 are generated by the ANFIS. The last column is the performance prediction. There are 64 cases or rules. If a rule is changed meaning that the performance will be changed. So that, the

*Automotive Industrial Supply Chain Performance Evaluation under Uncertain Constraints…*

*DOI: http://dx.doi.org/10.5772/intechopen.93679*

This section explains the cloud computing online monitoring system as shown in the **Figure 22**. The CPOM system links the production process and monitor, and evaluation. The production monitoring and control starts at the P&D orders which are received from the 1st tier company every 15 days. It is sometimes that the uncertain orders are given to the subcontractors to increased or decreased quantity of the parts. The production plan is assigned based on the availability of the factory. Capacity is presently checked as well as inventory, back order quantity, and raw material stock on hand. CIM system calculates the availability and actual capacity

performance can be predicted.

*The fuzzy rules of the ANFIS performance prediction.*

**Table 4.**

**4.3 Cloud production online monitoring (CPOM) system**


*Automotive Industrial Supply Chain Performance Evaluation under Uncertain Constraints… DOI: http://dx.doi.org/10.5772/intechopen.93679*

**Table 4.**

*Concepts, Applications and Emerging Opportunities in Industrial Engineering*

*The fuzzy rules of the ANFIS performance prediction.*

**Table 4** shows the fuzzy rules of the ANFIS performance prediction. The 3 first columns are input and 4th column is output. The column 5–7 are generated by the ANFIS. The last column is the performance prediction. There are 64 cases or rules. If a rule is changed meaning that the performance will be changed. So that, the performance can be predicted.
