Mustafa Demetgul

Fault diagnosis of pneumatic systems with artificial neural network algorithms

Pneumatic systems repeat the identical programmed sequence during their operation. The data was collected when the pneumatic system worked perfectly and had some faults including empty magazine, zero vacuum, inappropriate material, no pressure, closed manual pressure valve, missing drilling stroke, poorly located material, not vacuuming the material and low air pressure. The signals of eight sensors were collected during the entire sequence and the 24 most descriptive features of the data were encoded to present to the ANNs. A synthetic data generation process was proposed to train and test the ANNs better when signals are extremely repetitive from one sequence to other. Two artificial neural networks (ANN) were used for interpretation of the encoded signals. The tested ANNs were Adaptive Resonance Theory 2 (ART2), and Back propagation (Bp). ART2 correctly distinguished the perfect and faulty operations at all the tested vigilance values. It classified 11 faulty and 1 normal modes in seven or eight categories at the best vigilance values. Bp also distinguished perfect and faulty operations without even the slightest uncertainty. In less than 10 cases, it had difficulty identifying the 11 types of possible faults. The average estimation error of the Bp was better than 2.1% of the output range on the test data which was created by deviating the encoded values. The ART2 and Bp performance was found excellent with the proposed encoding and synthetic data generation procedures for extremely repetitive sequential data.

Taguchi Method-GONNS integration

Research highlights? Taguchi Method is an excellent experimental design and data analysis tool. ? Genetically Optimized Neural Network System (GONNS) is an automatic optimizer. ? Taguchi Method-GONNS integration covers all the aspects for experimentation. Determination of the optimal operating conditions from the experimental data without fitting any analytical or empirical models is very convenient for manufacturing applications. In this paper, integration of Taguchi Method and Genetically Optimized Neural Networks (GONNS) is proposed. The proposed procedure covers all the steps from experimental design to complex optimization. The feasibility of the approach was evaluated by estimating the optimal cutting conditions for the milling of Ti6Al4V titanium alloy with PVD coated inserts. The test conditions were determined by the Taguchi Method. The optimal cutting condition and influences of the cutting speed, feed rate and cutting depth on the surface roughness were analyzed with the same method. GONNS estimated that the optimal cutting conditions were very close to the Taguchi Method when the same criterion was used. GONNS was also capable to minimize or maximize one of the output parameters while the others were kept within the desired range. Study demonstrated that Taguchi Method and GONNS complement each other for creation of a robust procedure for determination of the test conditions, analysis of the quality of the collected data, estimation of the influence of each parameter on the output(s) and estimation of optimal conditions with complex optimization objective functions.

Design and testing of an efficient and compact piezoelectric energy harvester

Piezoelectric materials generate electricity when they are subjected to dynamic strain. In this paper compact size self contained energy harvesters were built by considering typical space available for AA size batteries. Each of the harvesters contains a rectifier circuit with four diodes and a capacitor. A series of piezoelectric energy harvesters with circular and square cross-sections were built and tested at different frequency and amplitude levels. On 1M? impedance digital oscilloscope, it was observed that the voltages reached to 16V (round cross-section) and 25V (square cross-section) at 50Hz frequency. The highest power output accomplished was 625µW. The outputs of both types of the harvesters were very similar at low amplitudes. However, the square cross-section facilitates better attachment of the piezoelectric elements with the harvester shell and worked efficiently at higher amplitudes without immediate failure.

Fault diagnosis on bottle filling plant using genetic-based neural network

Timely detection of the pneumatic system problems is important in industry. Many techniques have been employed to solve this problem. In this paper, Genetic Algorithm (GA) based optimal configuration of neural networks is proposed for fault diagnostic of bottle filling systems. Back-propagation is used for neural networks algorithm. The back-propagation algorithm had six inputs and one output. A fitness function was designed to the minimize execution time of ANN model by keeping the number of hidden layer(s) and nodes as low as possible while the mean square error of estimated output error is minimized. The designed GA-ANN combination and the graphical user interface (GUI) eliminate the trial and error process for selection of the fastest and most accurate configuration. The performance of the proposed system was evaluated by using experimental data collected at a pneumatic work cell which attach caps to the bottles. The sensory data was collected at normal operating conditions and a series of faults were imposed to the system such as missing bottle, attaching nonworking bottle caps at two different cylinders, two air pressure problems (insufficient and low air), and not filling water. The study demonstrated the convenience, accuracy and speed of the proposed GA-NN environment. It may also be used for training for selection of ANN configurations at various applications.

Basic computational tools and mechanical hardware for torque-based diagnostic of machining operations

In the industry, only rotary dynamometers can be used for monitoring when multiple spindles are used in machining operations. The current commercial rotary dynamometers are bulky and expensive for most machining centers. The basic hardware and computational tools proposed are for a smaller, more cost effective Torque-based Machining Monitor (TbMM). The objective of the TbMM concept is to estimate the remaining tool life, detect chatter from the torque signal inside the proposed device, and communicate with the central computer only when problems arise. The remaining tool life estimation and chatter detection algorithms of the TbMM were developed by analyzing the experimental data collected by a commercial rotary dynamometer. The mechanical hardware of the TbMM was designed to generate voltage proportional to the cutting torque using a piezoelectric composite element. The remaining tool life was estimated from the standard deviation (or variance) of the torque signal. Teager-Kaiser algorithm (TKA) based procedure detected the chatter based on the frequency estimations only from four samples at a time. The accuracy and characteristics of the signal of the mechanical component of the TbMM were found satisfactory in the estimation of machining problems such as wear and chatter. The TbMM is a good choice particularly when multiple spindles work simultaneously on the same workpiece.

A thermal-based defect classification method in textile fabrics with K-nearest neighbor algorithm:

In this study, fabric defects have been detected and classified from a video recording captured during the quality control process. Fabric quality control system prototype has been manufactured and a thermal camera was located on the quality control machine. The defective areas on the fabric surface were detected using the heat difference occurring between the defective and defect-free zones. Gray level co-occurrence matrix is used for feature extraction for defective images. The defective images are classified by k-nearest neighbor algorithm. The image processing stage consists of wavelet, threshold, and morphological operations. The defects have been classified with an average accuracy rate of 96%. In addition, the location of the defect has been identified and the defect type and location are recorded during the process via specially designed image processing interface. According to the experimental results, the proposed method works effectively.

Conditioning Monitoring and Fault Diagnosis for a Servo-Pneumatic System with Artificial Neural Network Algorithms

On-line monitoring of manufacturing process is extremely important in modern manufacturing for plant safety, maximization of the production and consistency of the product quality (Song et al., 2003). The development of diagnostic systems for the industrial applications has started in early 1970s. The recent developments in the microelectronics have increased their intelligence and let them found many industrial applications in last two decades (Mendonca et al., 2009; Shi & Sepehri, 2004). The intelligent data analysis techniques are one of the most important components of the fault diagnosis methods (Uppal et al, 2002; Uppal & Patton, 2002). In this study, the faults of a pneumatic system will be monitored by using the artificial neural networks (ANN). When the speed control and magnitude of the applied force is not critical, pneumatic systems are the first choice. They are cheap, easy to maintain, safe, clean, and components are commercially available. They have even been used for precise control of industrial systems (Nazir & Shaoping, 2009; Ning & Bone, 2005). Unfortunately, their nonlinear properties and some limitations at their damping, stiffness and bandwidth characteristics avoid their widespread applications (Belforte et al., 2004; Tsai & Huang, 2008, Bone & Ning, 2007; Taghizadeh et al., 2009; Takosoglu et al., 2009). The interest for the development of diagnostic methods for pneumatic and hydraulic systems has increased in the last decade (Nakutis & Kaskonas, 2008). Researchers concentrated on the detection of the faults of the components. The condition of the pneumatic and hydraulic cylinders (Wang et al., 2004), and digitally controlled valves (Karpenko et al., 2003) were the main focus of the studies. Some of the other considered faults were leakage of the seals (Nakutis & Kaskonas, 2005, 2007; Yang, 2006; Sepasi & Sassani, 2010), friction increase (Wang et al., 2004; Nogami et al., 1995) and other

A novel thermal-based fabric defect detection technique

During the fabric production process, many defects can be occurred stemming from the unevenness in spinning, weaving, finishing processes, or from the raw materials. The fabric quality control process for the detection of these defects is carried out by specialist operators. In this paper, a new method based on the use of thermal camera for detecting these defects from the textile fabric images is presented. For identification process of defective area, fabric images were obtained by a thermal camera during the fabric flow in quality control machine that was specially designed for this experiment. Defective and defect-free regions on fabric surface were determined by thermal camera due to the thermal differences. The mentioned thermal defect detection system will eliminate the worker usage for fabric quality control process, thus it will provide a cost-effective and competitive manufacturing.

Integrated System Health Management by using the Index Based Reasoning (IBR) and Self Organizing Map (SOM) combination

Integrated System Health Management (ISHM) will be an important component of the future launching vehicles and satellites. ISHM uses large number of sensors to evaluate the condition of the entire system. It is responsible to take the necessary corrective actions and to change the course of action if the health of the system will not allow successful completion of the mission. In this paper, Self Organizing Map (SOM) is integrated into the previously introduced Index Based Reasoner (IBR). The Simulink model of the proposed IBR-SOM approach is demonstrated and performance of the system is evaluated on the experimental data.

Fingerprinting the Lamb wave signals by using S-transformation

Lamb wave method detects the defects from the propagation characteristics of the created brief harmonic signals. Generally, the defects are detected by analyzing the delays and amplitudes of the received waves. The envelopes of the sensory signals may be used to calculate the delays and amplitudes of the received signals. Sometimes, similar envelopes could be observed at different test conditions. Use of the time-frequency spectra of the s-transformation is proposed for distinction of the problems when the envelopes of the monitored signals are very similar. In the study, a beam was compressed from different points with a hydraulic crimping tool. In separate tests, the cross sectional area at the middle of the beam was reduced by opening slots. The envelopes and time-frequency spectra of the sensory signals were calculated by using the s-transformation. The difference of the time-frequency spectra successfully distinguished the test condition when the envelopes were very similar.