Co-evolutionary learning and hierarchical fuzzy control for the inverted pendulum

Stonier, R.J.; Young, Nam Mi

doi:10.1109/cec.2003.1299612

“…For the cart-pole type inverted pendulum system if each of the n=4 variables partition into five MFs then there are 625 rules in the rule base (5 4 = 625). An initial analysis of the learning of fuzzy rules in a single block fuzzy system was given in [13]. The hierarchical structure of a fuzzy logic controller results from the desire to achieve a system goal for a complex process using a divide and conquer strategy.…”

Section: Problem Statementmentioning

confidence: 99%

Design of a self-tuning hierarchical fuzzy logic controller for nonlinear swing up and stabilizing control of inverted pendulum

Shill

¹

,

Amin

²

,

Murase

³

2012

2012 IEEE International Conference on Fuzzy Systems

View full text Add to dashboard Cite

Fuzzy logic controllers suffer from rule explosion problem as the number of rules increases exponentially with the number of input variables. Although several methods have been proposed for eliminating the combinatorial rule explosion problem, none of them is fully satisfactory. In this paper, we describe a new adaptive method for the design of cascaded layer based hierarchical fuzzy system with high input dimensions. This new adaptive hierarchical architecture could be applied to dimensionality reduction in fuzzy modeling. An evolutionary algorithm based off-line leaning algorithm is employed to generate the fuzzy rules and their corresponding membership functions. The evolutionary learning paradigm is a powerful tool to tune the fuzzy logic controllers since it requires no prior knowledge about the system's behavior in order to formulate a set of functional control rules through adaptive learning. The resulting hierarchical fuzzy system has not only an equivalent approximation capability, but less number of fuzzy rules compared with the conventional fuzzy logic system. Simulation studies exhibit competing results with high accuracy that illustrating the effectiveness of the approach.

show abstract

“…Clearly, manual approaches are not desirable because they are laborious and error-prone. Conventional EC techniques are not desirable either, because when the dimension of each candidate solution increases, conventional EC techniques may quickly lose their effectiveness [143]. Therefore, utilising cooperative coevolutionary computation techniques could be a promising and suitable solution to design FLCs effectively and cooperatively.…”

Section: Motivationmentioning

confidence: 99%

“…While each sub-population is evolved, the remaining subpopulations can be held fixed. Recently CCEAs have been applied to many optimisation problems [126,143,133,69,156,92]. For example, Stonier and Young [143] used CCEAs to optimise a hierarchical fuzzy controller for the inverted pendulum.…”

Section: Coevolutionary Design Of Fuzzy Logic Controlmentioning

confidence: 99%

“…Recently CCEAs have been applied to many optimisation problems [126,143,133,69,156,92]. For example, Stonier and Young [143] used CCEAs to optimise a hierarchical fuzzy controller for the inverted pendulum. Roberts et al [133] utilised Cooperative Coevolutionary Genetic Algorithm (CCGA) for image feature construction and object detection.…”

Section: Coevolutionary Design Of Fuzzy Logic Controlmentioning

confidence: 99%

See 1 more Smart Citation

Cross-Layer Media Access Control in Wireless Body Area Networks Using Fuzzy Logic and Evolutionary Computation

Nekooei¹

0

View full text Add to dashboard Cite

<p>Over the past decade, advances in electronics, computer science, and wireless technologies have facilitated the rapid development of Wireless Body Area Networks (WBANs). WBANs consist of various sensors that are attached on or even implanted in the human body to improve health care and the quality of life. WBANs must provide high-quality communication in terms of both reliability and performance, in order to bring timely medical help to patients. Commonly used communication standard in WBANs is IEEE 802.15.4. However, due to poor channel quality in WBANs, this standard is limited in reliability and performance. To address this issue, cross-layer techniques for Media Access Control (MAC) have attracted substantial research attention in recent years. Aimed at developing cross-layer MAC technologies, Fuzzy Logic Controllers (FLCs) have been widely utilised to effectively and efficiently process information from different layers in WBANs. However, existing FLCs have mostly focused on improving communication reliability while ignoring the importance of network performance. To improve both the reliability and performance of MAC protocols in WBANs, this thesis introduces a new design of cross-layer FLC, called Cross-Layer Fuzzy logic based Backoff system (CLFB), to improve reliability by reducing the collision rate and increasing the packet delivery ratio. CLFB can also enhance the network performance in terms of throughput in WBANs while maintaining packet delays at a reasonable level by considering both channel condition and application requirements. Through the proper use of FLCs as an extension of the standard exponential back-off algorithms, CLFB is experimentally shown to achieve a high level of adaptability. This thesis also proposes an evolutionary approach to automate the design of FLCs for CLFB in WBANs. With the goal of improving network reliability while keeping the communication delay at a low level, we have particularly studied the usefulness of three coding schemes with different levels of flexibility, which enables us to represent alternative design of FLCs as candidate solutions in evolutionary algorithms. The influence of fitness functions that measure the effectiveness of each possible FLC design has also been examined carefully in order to identify useful FLCs. Moreover, we have utilised surrogate models to improve the efficiency of the design process. In consideration of practical usefulness, we have further identified two main design targets. The first target is to design effective FLCs for a specific network configuration. The second target covers a wide range of network settings. In order to examine the usefulness of our design approach, we have utilised and experimentally evaluated two popularly used evolutionary algorithms, i.e. Particle Swarm Optimisation (PSO) and Differential Evolution (DE). This thesis finally proposes a two-level control scheme at both the sensor level and the coordinator level to further improve communication quality in WBANs. The sensor-level FLC controls contention based channel access and the coordinator-level FLC controls contention free channel access. This two-level FLC architecture can effectively enhance the cooperation between sensors and the coordinator such that both the reliability and performance of the network can be significantly improved. We also studied the use of cooperative coevolutionary approach to automate the design of our twolevel control scheme. With the goal of effectively designing useful FLCs, we have particularly investigated different collaborator selection methods for our cooperative coevolutionary approach, which enable us to effectively select collaborators while evaluating the candidate FLC design in each sub-population. Specifically, we show that network knowledge can help our evolutionary design approach to select collaborators more effectively.</p>

show abstract

“…Clearly, manual approaches are not desirable because they are laborious and error-prone. Conventional EC techniques are not desirable either, because when the dimension of each candidate solution increases, conventional EC techniques may quickly lose their effectiveness [143]. Therefore, utilising cooperative coevolutionary computation techniques could be a promising and suitable solution to design FLCs effectively and cooperatively.…”

Section: Motivationmentioning

confidence: 99%

Cross-Layer Media Access Control in Wireless Body Area Networks Using Fuzzy Logic and Evolutionary Computation

Nekooei¹

0

View full text Add to dashboard Cite

<p>Over the past decade, advances in electronics, computer science, and wireless technologies have facilitated the rapid development of Wireless Body Area Networks (WBANs). WBANs consist of various sensors that are attached on or even implanted in the human body to improve health care and the quality of life. WBANs must provide high-quality communication in terms of both reliability and performance, in order to bring timely medical help to patients. Commonly used communication standard in WBANs is IEEE 802.15.4. However, due to poor channel quality in WBANs, this standard is limited in reliability and performance. To address this issue, cross-layer techniques for Media Access Control (MAC) have attracted substantial research attention in recent years. Aimed at developing cross-layer MAC technologies, Fuzzy Logic Controllers (FLCs) have been widely utilised to effectively and efficiently process information from different layers in WBANs. However, existing FLCs have mostly focused on improving communication reliability while ignoring the importance of network performance. To improve both the reliability and performance of MAC protocols in WBANs, this thesis introduces a new design of cross-layer FLC, called Cross-Layer Fuzzy logic based Backoff system (CLFB), to improve reliability by reducing the collision rate and increasing the packet delivery ratio. CLFB can also enhance the network performance in terms of throughput in WBANs while maintaining packet delays at a reasonable level by considering both channel condition and application requirements. Through the proper use of FLCs as an extension of the standard exponential back-off algorithms, CLFB is experimentally shown to achieve a high level of adaptability. This thesis also proposes an evolutionary approach to automate the design of FLCs for CLFB in WBANs. With the goal of improving network reliability while keeping the communication delay at a low level, we have particularly studied the usefulness of three coding schemes with different levels of flexibility, which enables us to represent alternative design of FLCs as candidate solutions in evolutionary algorithms. The influence of fitness functions that measure the effectiveness of each possible FLC design has also been examined carefully in order to identify useful FLCs. Moreover, we have utilised surrogate models to improve the efficiency of the design process. In consideration of practical usefulness, we have further identified two main design targets. The first target is to design effective FLCs for a specific network configuration. The second target covers a wide range of network settings. In order to examine the usefulness of our design approach, we have utilised and experimentally evaluated two popularly used evolutionary algorithms, i.e. Particle Swarm Optimisation (PSO) and Differential Evolution (DE). This thesis finally proposes a two-level control scheme at both the sensor level and the coordinator level to further improve communication quality in WBANs. The sensor-level FLC controls contention based channel access and the coordinator-level FLC controls contention free channel access. This two-level FLC architecture can effectively enhance the cooperation between sensors and the coordinator such that both the reliability and performance of the network can be significantly improved. We also studied the use of cooperative coevolutionary approach to automate the design of our twolevel control scheme. With the goal of effectively designing useful FLCs, we have particularly investigated different collaborator selection methods for our cooperative coevolutionary approach, which enable us to effectively select collaborators while evaluating the candidate FLC design in each sub-population. Specifically, we show that network knowledge can help our evolutionary design approach to select collaborators more effectively.</p>

show abstract

Co-evolutionary learning and hierarchical fuzzy control for the inverted pendulum

Cited by 8 publications

References 10 publications

Design of a self-tuning hierarchical fuzzy logic controller for nonlinear swing up and stabilizing control of inverted pendulum

Design of a self-tuning hierarchical fuzzy logic controller for nonlinear swing up and stabilizing control of inverted pendulum

Cross-Layer Media Access Control in Wireless Body Area Networks Using Fuzzy Logic and Evolutionary Computation

Cross-Layer Media Access Control in Wireless Body Area Networks Using Fuzzy Logic and Evolutionary Computation

Contact Info

Product

Resources

About