*4.2.1 Quality operations before running manufacturing processes*

**Figure 1** indicates the basic elements of intelligent quality management in the first phase.

There are two aspects of specification analysis like process specification and product specification. These should be compliant with one another. Intelligent systems should be able to align and compare predefined set of requirements (specification match) and intelligently analyze those to generate complete set of specifications which could be used for properly designing the processes to run defined process plans. Fuzzy logic and expert systems could very well be employed for this.

**9**

*Introduction to Intelligent Quality Management DOI: http://dx.doi.org/10.5772/intechopen.94971*

Agents equipped with machine learning systems can also help in refining process

Quality function deployment (QFD) is an effective tool to shape product specifications based on customer requirements. Supporting this function with intelligent tools may enable automatic prediction of product quality with respect to customer expectations [18]. This information would also be used in forming design and manufacturing data. When there is a change in customer demand (which is the case in most of time), the effect of that in manufacturing can be assessed. With this capability intelligent QFD may be used during manufacturing process as well. Adding new requirements, removing an existing one, changing the design attributes would be handled by the support of intelligent tools like fuzzy logic. Supplier assessment is another area where intelligent systems would be employed. Fuzzy logic theory is heavily involved in selecting best suppliers among alternatives [19]. Similarly, literature shows evolutionary modeling [20], and neural network implementations [21]. This is indirectly related to quality operations as it ensures good quality and timely material with bearable costs to be available.

Designing quality (design of experiment) is another area where machine learning and artificial intelligence can be utilized. Finding out the optimum values of the quality parameters would be possible by learning systems which can learn various parameters

There are some information systems and software platforms for handling quality standards like ISO 9000 series. The tools which can facilitate especially corrective actions and run validation process over quality implementations. Computer based intelligent decision making along this line may generate effective implementation of quality standards. Enriching these systems with intelligent capabilities

**Figure 2** indicates the basic elements of intelligent quality management in the second phase. As indicated, generating quality standards and quality assurance system is the baseline requirement for setting up intelligent quality operations during execution. This may assure the integration of quality operation for various manufacturing systems. As described by Albers et al. (2016), a comprehensive model can be devised in order to support both digitization process and sustaining

and product specifications and generate optimum designs [17].

and help optimize the quality design [13].

*Intelligent quality operations before manufacturing.*

**Figure 1.**

may empower quality assurance systems.

*4.2.2 Quality operations during execution of manufacturing processes*

*Introduction to Intelligent Quality Management DOI: http://dx.doi.org/10.5772/intechopen.94971*

*Quality Control - Intelligent Manufacturing, Robust Design and Charts*

ment activities in various ways as listed below.

standards.

knowledge sets.

manufacturing system. These are;

framework for generating an integrated intelligent quality system. As implied above, main requirement behind this framework is to set up knowledge intensive activities for assuring quality and continuous improvement. This, in turn, requires a systematic performance monitoring and assessment. The proposed framework is based on achieving business objectives which in turn needs to be aligned with the objectives of the processes. Note that, intelligent tools can support quality manage-

• Handling quality policies and objectives of any manufacturing system could be supported by intelligent tools in order to set up proper quality plans and

• Quality management related knowledge could very well be stored in knowledge bases and utilized in making quality decisions. The performance of the system will be depended upon the acquisition and application of relevant

• Data from various sensors could be gathered and interpreted for the sake of sustaining on-line monitoring of the process in terms of quality requirements.

• Design and operational reliability (ability to perform a specific function) availability (ability to keep a functioning state) maintainability (ability to be timely and easily maintained) and safety (ability not to harm people, the environment, or any assets) analysis (SARM) could be carried out intelligently for assuring engineering characteristics of products or systems. SARM is proven to be good in identifying, analyzing, evaluating, preventing, verifying and correcting the defects and hazards of any system. Intelligent tools could facilitate these analyses more effectively and efficiently for the benefit of the enterprise.

Taking above explanation into account, it would be possible to say that intelligent quality management operations are carried out in 3 different phases of any

All of these phases should be supported by intelligent tools in various nature and should be involved within given framework for the sake of integration and

**Figure 1** indicates the basic elements of intelligent quality management in the

There are two aspects of specification analysis like process specification and product specification. These should be compliant with one another. Intelligent systems should be able to align and compare predefined set of requirements (specification match) and intelligently analyze those to generate complete set of specifications which could be used for properly designing the processes to run defined process plans. Fuzzy logic and expert systems could very well be employed for this.

• Phase-1: Operations before running manufacturing processes

• Phase-2: Operations during running manufacturing processes.

completeness of the whole life cycle of quality management activities.

• Phase-3: Quality Operations after manufacturing

*4.2.1 Quality operations before running manufacturing processes*

**8**

first phase.

**Figure 1.** *Intelligent quality operations before manufacturing.*

Agents equipped with machine learning systems can also help in refining process and product specifications and generate optimum designs [17].

Designing quality (design of experiment) is another area where machine learning and artificial intelligence can be utilized. Finding out the optimum values of the quality parameters would be possible by learning systems which can learn various parameters and help optimize the quality design [13].

Quality function deployment (QFD) is an effective tool to shape product specifications based on customer requirements. Supporting this function with intelligent tools may enable automatic prediction of product quality with respect to customer expectations [18]. This information would also be used in forming design and manufacturing data. When there is a change in customer demand (which is the case in most of time), the effect of that in manufacturing can be assessed. With this capability intelligent QFD may be used during manufacturing process as well. Adding new requirements, removing an existing one, changing the design attributes would be handled by the support of intelligent tools like fuzzy logic.

Supplier assessment is another area where intelligent systems would be employed. Fuzzy logic theory is heavily involved in selecting best suppliers among alternatives [19]. Similarly, literature shows evolutionary modeling [20], and neural network implementations [21]. This is indirectly related to quality operations as it ensures good quality and timely material with bearable costs to be available.

There are some information systems and software platforms for handling quality standards like ISO 9000 series. The tools which can facilitate especially corrective actions and run validation process over quality implementations. Computer based intelligent decision making along this line may generate effective implementation of quality standards. Enriching these systems with intelligent capabilities may empower quality assurance systems.

#### *4.2.2 Quality operations during execution of manufacturing processes*

**Figure 2** indicates the basic elements of intelligent quality management in the second phase. As indicated, generating quality standards and quality assurance system is the baseline requirement for setting up intelligent quality operations during execution. This may assure the integration of quality operation for various manufacturing systems. As described by Albers et al. (2016), a comprehensive model can be devised in order to support both digitization process and sustaining

#### **Figure 2.** *Intelligent quality operations during manufacturing.*

good quality in an integrated manner [22]. Similar studies indicate that, some of the methods and methodologies such visual and constructive inspection of products and processes, intelligent classifier and data acquisition systems, defect detection and fault diagnosis as well as condition monitoring systems are inevitable for real time quality management.

Making the data available through data gathering systems enables analysis and visualization of information which can be effectively used in generating evidences for quality related decision making. By detecting defects as early as possible in manufacturing line not only yields cost reduction but also eliminates causes of failures. This is in fact one of the main requirements for successful quality assurance systems.

Real time intelligent quality assurance may ensure that the smart operations are handle without interruption and sustain continuity of manufacturing operations with the required level of reliability, flexibility, and interoperability. This, in turn, increases productivity and effectiveness as well as to reduce the defects to the minimum level possible.

Real time and online quality operations handled by intelligent tools should support well-known quality tools such as quality function deployment, statistical process control, constructive and visual inspection, failure mode and effect analysis, measurement systems etc. Each of these tools could very well be enriched by self-behaving capability and reconfigured in accordance with the requirements and specification of manufacturing systems.

Statistical process control (SPC) is one of the major area for implementing artificial intelligence for sustaining good quality within a manufacturing process. Pham

**11**

*Introduction to Intelligent Quality Management DOI: http://dx.doi.org/10.5772/intechopen.94971*

on-line.

not to be overstocked.

defect free.

are implemented.

and Oztemel (1996) are long ago took the attention of the scientific community to this point by introducing an expert system (called XPC) and machine learning system for statistical process control [1]. XPC was able to construct quality control charts and perform process capability analysis in order to make sure that the process is able to comply with customer specifications, collect data on-line, plot those over the chart constructed and interpret the status of the process, perform fault detection and diagnosis, make some recommendation in case of an out of control situation. The system was also able to modify control limits and control chart used in order to embed improved process standards. It was well appreciated as the system was playing the role of a quality engineer and continuously monitoring the process

When utilizing machines, there is and will be a need for condition monitoring and try to identify possible machine faults before they cause unbearable costs. As described in Zabinski et al. (2019), condition monitoring systems mainly deals with, monitoring and feature extraction, real-time anomaly detection and fault diagnosis [23]. Expert systems and rule based reasoning mechanisms, neural networks and deep learning methods as well as intelligent agents can be extensively used for monitoring the machines and process for possible anomalies and faults. Literature provides huge amount of implementation in these areas. Besides, as condition monitoring is the best way to minimize the probability of failure and therefore maintenance cost, making it intelligent and generating a self-behaving

By implementing AI techniques to monitor the condition of an equipment, identifying possible faults and observing the trend of the level of deterioration may allow possible actions to be taken at convenient times without interrupting production process. Generating intelligent predictive maintenance system by this way, may prevent major overhauls or increase the time to go into an overhaul. This definitely reduces significant damage that would otherwise be unavoidable. Above all, proper maintenance schedules can be generated in between the demands for having maintaining the equipment if not done automatically. Considering the amount of money spend for keeping spare parts of the machines in stock, intelligent condition monitoring may prevent undesired use of spare parts and avoid unnecessary parts

Artificial intelligent methods, especially machine learning techniques remarkably increased the effective utilization of visual inspection system for the sake of improving the quality of products and processes [25]. Note that, quality control through inspection is long being implemented in industry for locating nonconforming products or processes. Main aim is to deliver products and services to the customers in required specification (so called good quality) or

In earlier times, it was possible to inspect nonconformity by human eyes. Since the complexity and functionality of the systems are increased, it would not be possible to decide about the quality of the inspected items by human eyes. Computer support was inevitable and self-decision making which was assured by artificial intelligence was extremely useful. This is even more important today due to the nature of big data available within manufacturing areas. Two approaches

• Taking the video record of the systems and products and then analyzing those

• Real-time data collection and defect detection using image recognition and

off-line for finding if there is any abnormality.

AI techniques such as neural networks.

observer can be an important tool for predictive maintenance [24].

#### *Introduction to Intelligent Quality Management DOI: http://dx.doi.org/10.5772/intechopen.94971*

*Quality Control - Intelligent Manufacturing, Robust Design and Charts*

good quality in an integrated manner [22]. Similar studies indicate that, some of the methods and methodologies such visual and constructive inspection of products and processes, intelligent classifier and data acquisition systems, defect detection and fault diagnosis as well as condition monitoring systems are inevitable for real

Making the data available through data gathering systems enables analysis and visualization of information which can be effectively used in generating evidences for quality related decision making. By detecting defects as early as possible in manufacturing line not only yields cost reduction but also eliminates causes of failures. This is in fact one of the main requirements for successful quality assurance

Real time intelligent quality assurance may ensure that the smart operations are handle without interruption and sustain continuity of manufacturing operations with the required level of reliability, flexibility, and interoperability. This, in turn, increases productivity and effectiveness as well as to reduce the defects to the

Real time and online quality operations handled by intelligent tools should support

Statistical process control (SPC) is one of the major area for implementing artificial intelligence for sustaining good quality within a manufacturing process. Pham

well-known quality tools such as quality function deployment, statistical process control, constructive and visual inspection, failure mode and effect analysis, measurement systems etc. Each of these tools could very well be enriched by self-behaving capability and reconfigured in accordance with the requirements and specification of

**10**

systems.

**Figure 2.**

time quality management.

*Intelligent quality operations during manufacturing.*

minimum level possible.

manufacturing systems.

and Oztemel (1996) are long ago took the attention of the scientific community to this point by introducing an expert system (called XPC) and machine learning system for statistical process control [1]. XPC was able to construct quality control charts and perform process capability analysis in order to make sure that the process is able to comply with customer specifications, collect data on-line, plot those over the chart constructed and interpret the status of the process, perform fault detection and diagnosis, make some recommendation in case of an out of control situation. The system was also able to modify control limits and control chart used in order to embed improved process standards. It was well appreciated as the system was playing the role of a quality engineer and continuously monitoring the process on-line.

When utilizing machines, there is and will be a need for condition monitoring and try to identify possible machine faults before they cause unbearable costs. As described in Zabinski et al. (2019), condition monitoring systems mainly deals with, monitoring and feature extraction, real-time anomaly detection and fault diagnosis [23]. Expert systems and rule based reasoning mechanisms, neural networks and deep learning methods as well as intelligent agents can be extensively used for monitoring the machines and process for possible anomalies and faults. Literature provides huge amount of implementation in these areas. Besides, as condition monitoring is the best way to minimize the probability of failure and therefore maintenance cost, making it intelligent and generating a self-behaving observer can be an important tool for predictive maintenance [24].

By implementing AI techniques to monitor the condition of an equipment, identifying possible faults and observing the trend of the level of deterioration may allow possible actions to be taken at convenient times without interrupting production process. Generating intelligent predictive maintenance system by this way, may prevent major overhauls or increase the time to go into an overhaul. This definitely reduces significant damage that would otherwise be unavoidable. Above all, proper maintenance schedules can be generated in between the demands for having maintaining the equipment if not done automatically. Considering the amount of money spend for keeping spare parts of the machines in stock, intelligent condition monitoring may prevent undesired use of spare parts and avoid unnecessary parts not to be overstocked.

Artificial intelligent methods, especially machine learning techniques remarkably increased the effective utilization of visual inspection system for the sake of improving the quality of products and processes [25]. Note that, quality control through inspection is long being implemented in industry for locating nonconforming products or processes. Main aim is to deliver products and services to the customers in required specification (so called good quality) or defect free.

In earlier times, it was possible to inspect nonconformity by human eyes. Since the complexity and functionality of the systems are increased, it would not be possible to decide about the quality of the inspected items by human eyes. Computer support was inevitable and self-decision making which was assured by artificial intelligence was extremely useful. This is even more important today due to the nature of big data available within manufacturing areas. Two approaches are implemented.


Technology enables obtaining high quality video images with certain cameras specifically configured to manufacturing line. Data collected can be cleaned in real time for the sake of locating anomalies and defects such as scratch, dent, crack, dirt, wrong print, foreign objects, undesired bubbles etc. Computer vision system replicates human vision and assessment system.

Inspection systems may require both hardware (camera, gateway, photometer, colorimeter, CPU etc.) and software (video processor, database, reasoning mechanism and learning engine, cloud services etc.). Hardware is mainly used for data collection and processing whereas software is used for recognition and interpretation. In intelligent visual inspection, the learning and interpretation capability indicates the degree of smartness.

Production part approval process (PPAP) can be supported by intelligent tools for providing evidences that all customer demand-based specifications are well understood and proper manufacturing set up is in place. It may ensure that the manufacturing process is running in expected rate and producing well accepted products by the customers. It may also assure the fitness of quality plans, support maintaining design integrity, provide early detection of problems to be sorted, prevent nonconforming parts to be processed etc. Another benefit of intelligent PPAP is to make sure that the suppliers have well understood the engineering design attributes and specifications of the products they are vendoring.

Intelligent failure mode and effect analysis (FMEA) can be very useful for finding and diagnosing process and system faults. The system may handle the information of various failures and analyze and interpret those from multiple point of view. Characteristics of different failures within product and process life cycle can be well managed. As described by Zhao et al. (2004), the system may involve a model base and knowledge base as well as functional models of the products and effect analysis of possible failures [26]. It would be possible to perform various reasoning processes depending upon the type of failure in question. FMEA of very complex system can be carried out with implementing expert knowledge of the domain as well as respective failures. This system may be coupled with Fault Tress Analysis (FTA) which is a hierarchical representation of faults enabling the analysis for the probability of a failure and its consequences. Due to uncertainty, complexity and inexact information, it is not easy to define the probability of a fault to occur. However, intelligent tools such as fuzzy logic and neural network are very capable of handling imprecise knowledge [27]. Generating intelligent FMEA and FTA may allow generating records of failures and recommends some corrective actions for improving the quality of manufacturing systems.

Creating and integrating all of the systems, some of those are listed above, may lead fully automated intelligent quality system within a smart manufacturing environment. Being able to receive data and information from various manufacturing functions and related systems without any interruption enables the designers to device a recommendation system to prevent misleading operations and deterioration from respective manufacturing plans. Recommendation could be provided for;


**13**

**5. Conclusion**

*Intelligent quality operations after manufacturing.*

**Figure 3.**

*Introduction to Intelligent Quality Management DOI: http://dx.doi.org/10.5772/intechopen.94971*

*4.2.3 Quality operations after manufacturing*

some of those are given in **Figure 3**.

warranty as well.

Some of the recommendations would be carried out by intelligent systems capable of self-improving whereas some of them would still be sorted by the human operators.

Upon completing manufacturing, the products, some quality operations need to be carried out in delivery and after sale services. There could be various activities

Performing quality predictions before and after manufacturing can be carried out intelligently in order to follow if the predicted quality can be realized or not. Intelligent post production analyzer receives not only the predicted values but also some warranty information to keep track of the products after they delivered to the customers. Products should comply with the specifications and terms of use as promised before order received. The system may define conditions subject to

Comprehensive quality analysis may also be carried out upon receiving some feedback from the customer. Feedback can be received either by satisfaction surveys and information gathered by customer information channels (i.e. on-line support through web page applications). Information provided by the customers may be a good source of updating quality standards and expectations within manufacturing systems. Analyzing the quality costs may provide some useful information for quality related operations and their effectiveness. Cost benefit analysis could direct improvements over the quality management process. Deterioration and sampling costs as well as the cost of inspection and control systems for specific products with specific attributes/specifications. Intelligent cost analyzer may collect cost data and may establish true costing and pricing strategies which would not compromise from the quality but increase the attractiveness of the products and services. Analyzer

Intelligent quality management is attracting the attention of manufacturing society as it provides much opportunities to improve quality of both products and

may generate some recommendations for quality improvements.

attributes and specifications of the products they are vendoring.

improving the quality of manufacturing systems.

• Foreseeing possible deteriorations

• Correcting errors and faults

• Preventing errors and faults over the process or products

• Preventing the shift of non-conformed products between processes.

• Preventing the delivery of non-conformed products to customers.

replicates human vision and assessment system.

indicates the degree of smartness.

Technology enables obtaining high quality video images with certain cameras specifically configured to manufacturing line. Data collected can be cleaned in real time for the sake of locating anomalies and defects such as scratch, dent, crack, dirt, wrong print, foreign objects, undesired bubbles etc. Computer vision system

Inspection systems may require both hardware (camera, gateway, photometer, colorimeter, CPU etc.) and software (video processor, database, reasoning mechanism and learning engine, cloud services etc.). Hardware is mainly used for data collection and processing whereas software is used for recognition and interpretation. In intelligent visual inspection, the learning and interpretation capability

Production part approval process (PPAP) can be supported by intelligent tools for providing evidences that all customer demand-based specifications are well understood and proper manufacturing set up is in place. It may ensure that the manufacturing process is running in expected rate and producing well accepted products by the customers. It may also assure the fitness of quality plans, support maintaining design integrity, provide early detection of problems to be sorted, prevent nonconforming parts to be processed etc. Another benefit of intelligent PPAP is to make sure that the suppliers have well understood the engineering design

Intelligent failure mode and effect analysis (FMEA) can be very useful for finding and diagnosing process and system faults. The system may handle the information of various failures and analyze and interpret those from multiple point of view. Characteristics of different failures within product and process life cycle can be well managed. As described by Zhao et al. (2004), the system may involve a model base and knowledge base as well as functional models of the products and effect analysis of possible failures [26]. It would be possible to perform various reasoning processes depending upon the type of failure in question. FMEA of very complex system can be carried out with implementing expert knowledge of the domain as well as respective failures. This system may be coupled with Fault Tress Analysis (FTA) which is a hierarchical representation of faults enabling the analysis for the probability of a failure and its consequences. Due to uncertainty, complexity and inexact information, it is not easy to define the probability of a fault to occur. However, intelligent tools such as fuzzy logic and neural network are very capable of handling imprecise knowledge [27]. Generating intelligent FMEA and FTA may allow generating records of failures and recommends some corrective actions for

Creating and integrating all of the systems, some of those are listed above, may lead fully automated intelligent quality system within a smart manufacturing environment. Being able to receive data and information from various manufacturing functions and related systems without any interruption enables the designers to device a recommendation system to prevent misleading operations and deterioration from respective manufacturing plans. Recommendation could be provided for;

**12**

Some of the recommendations would be carried out by intelligent systems capable of self-improving whereas some of them would still be sorted by the human operators.
