Pnömatik yapay kas tabanlı üst ekstremite manipülatöründe agonist-antagonist kontrol
Agonist-antagonist control in a pneumatic artificial muscle-based upper limb manipulator
- Tez No: 906342
- Danışmanlar: PROF. DR. UĞUR ARİFOĞLU
- Tez Türü: Yüksek Lisans
- Konular: Elektrik ve Elektronik Mühendisliği, Electrical and Electronics Engineering
- Anahtar Kelimeler: Belirtilmemiş.
- Yıl: 2024
- Dil: Türkçe
- Üniversite: Sakarya Üniversitesi
- Enstitü: Fen Bilimleri Enstitüsü
- Ana Bilim Dalı: Elektrik-Elektronik Mühendisliği Ana Bilim Dalı
- Bilim Dalı: Elektronik Mühendisliği Bilim Dalı
- Sayfa Sayısı: 97
Özet
Son zamanlarda robotları boyama, montaj, kaynak, kaldırma, yerleştirme vs. pek çok alanda görmekteyiz. Bu uygulamalar çoğunlukla insansı robotlar tarafından yapılmaktadır. İnsansı robotlar insan hareketlerini taklit etmeyi amaçlayan robotlardır. İnsansı robotlarda genellikle esnek yapısı nedeniyle insan kasına benzediğinden pnömatik yapay kaslar (PYK'lar) kullanılmaktadır. Bu çalışmada agonist- antagonist kas çiftini oluşturan PYK hakkında ayrıntılı araştırma yapılmış PYK çeşitleri, özellikleri, avantajları, dezavantajları ve çalışma prensibi açıklanmıştır. PYK'lar bakım kolaylığı, düşük maliyeti ve yüksek kuvvetlere dayanabilmesi nedeniyle günümüzde oldukça tercih edilen bir aktüatördür. PYK'yı daha yakından incelemek amacı ile statik modeli tasarlamak için deney düzeneği hazırlanmış, elde edilen veriler ile PYK'nın statik modelinin denklemi başarı ile bulunmuş ve katalog verileri ile karşılaştırılmıştır. Statik kuvvet haritası; kuvvet, basınç ve uzunluk arasındaki ilişkiyi tanımlar. PYK'lar genellikle yüksek kuvvetlere çıkabildiğinden PYK'nın statik kuvvet haritasını oluşturmak oldukça önemlidir. PYK'nın statik kuvvet haritası için ayrı bir deney düzeneği oluşturulmuştur. Bu veriler ile kuvvet haritası oluşturulmuştur. PYK'nın kuvvet modeli matematiksel olarak modellenmiş ve akademik çalışmalar sonucu bulunan diğer kuvvet modelleri ile karşılaştırılmıştır. Agonist antagonist kaslar canlılarda eklem hareketinin oluşmasını sağlayan insanda kol ve bacaklarda bulunan birbirine zıt yönlü çalışan kaslardır. Bu çalışmada agonist-antagonist kasların çalışması hakkında açıklamalar yapılmış ve Agonist-antagonist kasların matematiksel modeli verilmiştir. agonist-antagonist kasların kontrolü hakkında literatür taraması yapılmış ve literatürde genellikle iki PYK kullanıldığı gözlemlenmiştir. Bu alanda kullanılan kontrol yöntemleri model tabanlı kontrol ve modelden bağımsız kontrol olarak iki tip olduğu görülmüştür. Kontrol stratejisi olarak Empedans kontrol, PID (Proportional, Integral, Derivative) kontrol, kayan kip kontrol (Sliding Mode Control-SMC), bulanık(fuzzy) kontrol ve Yapay sinir ağı gibi kontrol türleri kullanıldığı gözlemlenmiştir. Bu çalışmada kontrolör olarak PID Kontrolcü seçilmiştir. Model tabanlı kontrolün zorluğundan dolayı modelden bağımsız kontrol yöntemi tercih edilmiştir. Literatürdeki iki PYK kullanımı aksine maliyet açısından tasarruf sağlamak amacı ile bir PYK ve bir yay ile agonist-antagonist kasın konum kontrolü PID kontrolör kullanılarak yapılmıştır. Eklem hareketinin doğruluğu hem dijital ortamda hem de deney düzeneği üzerinde gözlemlenmiş ve kanıtlanmıştır.
Özet (Çeviri)
Nowadays, with the advancement of technology, artificial systems designed to imitate biological systems are increasingly increasing. These artificial systems have the potential to make things easier for humans, optimize industrial processes, and lead to significant advances in healthcare. PAMs (pneumatic artificial muscles) are artificial muscles that work with air or gas pressure and are designed to imitate the movement of natural human muscles. They are often used in robotic systems, biomechanical applications, industrial automation and rehabilitation devices. In this study, in chapter 2, detailed research about PAM was made and PAM types, features, advantages, disadvantages and working principle were explained. There are many types of PAMs. In this study, braided, pleated, netted, embedded and Festo PAM types were examined. The general features of these PAMs are as follows: They work similar to natural muscles, act like a single-acting cylinder, can shorten to approximately 20%-35% of their current length, can be used as a spring at constant pressure or constant volume, and have initial forces up to ten times higher than the same type of pneumatic cylinder. PAMs are an important technology used in many applications and offer a number of advantages. The main advantages of PAMs are: They are generally made of lightweight materials, which provides portability and flexibility of use. PAMs can produce a large force in a small volume, providing powerful and effective movement. Because air or gas is used, PAMs are generally environmentally friendly. PAMs do not require lubrication and are easy to maintain since they have few moving mechanical parts. They can be used in dirty and dusty environments. It does not leak, there is no need for vacuum or air evacuation. These advantages of PAMs contribute to their preference in many industrial and commercial applications and to pioneering various technological innovations. While PAMs have many advantages, they also have disadvantages. Here are the main disadvantages of PAMs: Model-based control of PAMs is difficult and complex control algorithms may be needed in cases where precise control is required. It has high hysteresis. In contraction and relaxation states, the same shortening value does not occur at the same pressure. It may produce noise caused by air movement during operation, which can be a problem especially in indoor environments or applications that require quiet operation. The range of motion is limited. It has a slower response time compared to hydraulic or electric systems. Rubber material may undergo disinformation after a while. It is not resistant to welding spatters and sharp fragmented environments. These disadvantages do not mean that pneumatic artificial muscles may not be suitable in certain applications, but they are factors that should be considered in the design and application processes. Depending on the application requirements, these drawbacks may be minimized or tolerable.The working principle of PAMs aims to imitate the movement of human muscles in a simple but effective way. With the change of air or gas pressure, the flexible sheath of the muscle expands or contracts, allowing the desired movement to occur. PAMs swell and shorten during contraction, that is, under pressure. It also contracts and lengthens when external force is applied. We can say that this principle forms the basis of pneumatic systems used in many industrial and commercial applications. In Chapter 3, we wanted to find out the static characteristics of PAMs. For this, Festo's DMSP-20-200N RM-RM PAMs was used. This PAM has an inner diameter of 20 mm and a length of 200 mm. It can shorten up to 25% of its maximum length. The static characteristic represents the amount of muscle shortening relative to pressure. An experimental setup was prepared for this. In the experimental setup, a string encoder was connected to the PAMs and the output of the encoder was connected to the data logger card of National Instrument. By adjusting the pressure between 0-6 bars in the Daisy Lab application, the amount of shortening was found separately for both contraction and relaxation of the muscle. These data were transferred to the Matlab environment and equations were found with the curve fitting method both in the cftool environment and with the fminserch command. Two types of equations were obtained. One of these is the Fourier equation. Although this equation is very close to the experimental results, a polynomial equation was also found as an alternative since it is a complex equation. The data obtained from the equations are presented graphically. Additionally, the experimental results and catalog data were compared. As a result of the comparison, it was seen that the experimental data did not match the catalog data exactly. There may be many reasons why the test results are not exactly the same as the catalogue, some of which may be air leakage in the experimental setup, rubber being a material that can stretch and undergo deformation, PAM not being a linear actuator, PAMs hysteresis being high, and by taking the average of PAMs contraction-relaxation states. It was thought that this might be due to comparison. In Chapter 4, we wanted to find the static force model of PAM. PAMs are generally used because they produce high force, so it is important to find the force characteristic. PAM force can be measured in two different ways: isometric force measurement, that is, the length of the PAM is kept constant while the pressure increases, or isobaric force measurement, that is, the length of the PAM is variable while the pressure is maintained. Isobaric force measurement was used in this study. An experimental setup was prepared to find the relationship between force, length and pressure. DMSP-20-200N RM-RM, National Instruments data logger card, pressure regulator, pressure indicator, load cell indicator, linear ruler, DC motor, Load cell were used in the experiment. The aim of the experiment is to read the force values by keeping the pressure of the DMSP-200N RM-RM constant, increasing and decreasing the throttling ratio, creating a static force map and testing the force model to be found. For this, the pressure value to be given to PAM from the Daisy Lab application interface was adjusted to 6 bars with 0.5 increments starting from 0.5 bar, and the muscle length was adjusted to 21% reduction starting from 0% reduction rate and 3% increments. Separate force values were found for each pressure and throttling ratio. Two different force values were found for muscle contraction and relaxation. Feedback position control was made to adjust the throttle ratios. For position control, the position value read from the linear ruler was compared with the reference position value selected by the user and transmitted to the PID controller. The controller signal was transferred to the DC motor. When the desired position was reached, the engine stopped working and the force values were read from the load cell indicator or the computer interface.Experimental results were tabulated and a static force map was drawn in Matlab environment for DMSP-20-200N RM-RM. According to these results, it was observed that the PAM force varies significantly with length, the maximum force can only be applied at the initial length, and the resulting force decreases as the contraction rate of the PAM is increased. Force static models available in the literature were researched and added to the study. The new static force model was created using the stress equations affecting pressure vessels. The equation found was compared with the experimental results and the error rate was calculated. The error rate found was compared with existing static force models and was found to have better performance. In Chapter 5, the working principle of the agonist-antagonist system, its mathematical model and PID control with a PAM and a spring are explained. Agonist-antagonist systems refer to groups of muscles that work reciprocally. In this system, there is a pair of muscles: agonist and antagonist muscles. The agonist muscle represents the muscle that moves a joint in one direction, while the antagonist muscle represents the muscle that moves the same joint in the opposite direction. For example, the biceps and triceps muscles form an agonist-antagonist pair. The biceps bends the elbow (flexion), while the triceps muscle straightens it (extension). The operation of the two PAM agonist antagonist system is as follows; Initially, the length of the two muscles is adjusted to the same pressure, reaching half of the total shortening amount. To change the joint angle, pressure is applied to one of the muscles and the pressure of the other muscle is reduced by the same amount. It was desired to find a mathematical model of the agonist-antagonist PAM pair. For this, a phenomenological model was used. In other words, PAM is likened to elements that do not actually exist but can define it. With this approach, PAM is modeled with spring, damping coefficient and contraction force and the equations are given in the thesis. PID controller was used to control the agonist-antagonist system. PID controller consists of proportional controller, integral controller and differential controller. The working principle of this controller is as follows: The difference between the desired value, that is, the reference value and the current output, is calculated. This is an error signal. When the error signal comes to the PID controller, PID multiplies this error by a determined coefficient, takes its integral and derivative. The signal from the controller is sent to the system and a new output signal is obtained. Again, the new output value is compared with the input and the same operations are performed. There are many methods to adjust PID coefficients. Some of these are manual tuning, Ziegler-Nichols, Loop tuning software and frequency response methods. It was desired to control the position of the agonist-antagonist system with PID using a PAM and a spring. In the literature, two PAMs are generally used for this movement. While some of the studies in the literature on the control of the agonist-antagonist system with PAM adopt a model-based control approach, in some others a model-independent control approach is used in order to avoid the complexity and difficulty of modeling. In this study, the model-independent control method was used and PID coefficients were adjusted by manual adjustment method. For this purpose, an experimental setup was set up. Encoder, DMSP 20-200N RM-RM PAM, spring, National Instrument data logger card, pressure regulator and pressure gauge materials were used in the experiment. The encoder is set to zero for the situation where PAM is unpressurized. As the PAM pressure is increased, the angle of the encoder increases, and at the maximum shortening of the PAM (at 6 bar), the encoder can show an angleof up to 120 degrees. MATLAB/SIMULINK program was used for control. The circuit for the system was designed in the SIMULINK program. With this circuit, a reference variable signal was created, position information was read from the encoder with the analog input block, and the signal was sent to the pressure regulator with the analog output block. In this way, the system was tried to be controlled in real time. PID coefficients were calculated through experiments carried out via Simulink, and PID position control of the agonist-antagonist system was successfully carried out using a PAM and a spring.
Benzer Tezler
- Doğadan esinli interaktif bina kabuğu tasarımı için örüntüye dayalı bir model
A model for nature inspired interactive building envelope design based on pattern
HÜLYA ORAL
Yüksek Lisans
Türkçe
2015
Mimarlıkİstanbul Teknik ÜniversitesiBilişim Ana Bilim Dalı
PROF. DR. GÜLEN ÇAĞDAŞ
- Yapay kaslar kullanarak hava araçlarında konum kontrolü
Location control in air vehicles using artificial muscles
MUSTAFA BAKIR
Yüksek Lisans
Türkçe
2019
Havacılık MühendisliğiErciyes ÜniversitesiSivil Havacılık Ana Bilim Dalı
DR. ÖĞR. ÜYESİ MUSTAFA SOYLAK
- Eklemeli imalatla üretilmiş pnömatik yapay kas ile el rehabilitasyon ortezi mekatronik sistem tasarımı
Mechatronic system design of a hand rehabilitation orthosis with additive manufactured pneumatic artificial muscle
EMRE TUĞBERK GÜLNERGİZ
Yüksek Lisans
Türkçe
2020
Mekatronik Mühendisliğiİstanbul Gedik ÜniversitesiMekatronik Mühendisliği Ana Bilim Dalı
DOÇ. DR. SAVAŞ DİLİBAL
- Pnömatik yapay kaslı 3 serbestlik dereceli robot kolun yapay sinir ağları ile konum kontrolü
Position control of pneumatic artificial muscle 3 degrees of freedom robot arm with artificial neural network
KÜBRA TÜRE
Yüksek Lisans
Türkçe
2023
Makine MühendisliğiZonguldak Bülent Ecevit ÜniversitesiMakine Mühendisliği Ana Bilim Dalı
PROF. DR. HAMDİ ALPER ÖZYİĞİT
- Aktif ayak-ayak bileği ortezin tasarımı ve veri güdümlü öngörülü kontrolü
Design and data-driven predictive control of an active ankle-foot orthosis
OSMAN ÜLKİR
Doktora
Türkçe
2019
Mekatronik MühendisliğiMarmara ÜniversitesiMekatronik Mühendisliği Ana Bilim Dalı
DOÇ. DR. ERKAN KAPLANOĞLU