Zhenyou Zhang

Condition-based Maintenance (CBM) is a maintenance policy that take maintenance action just when need arises with real-time condition monitoring. Intelligent CBM means a CBM system is capable of understanding and making maintenance decisions without human intervention. To achieve this objective, it is needed to detect current conditions of mechanical and electrical systems and predict the fault of the systems accurately. What’s more, the maintenance scheduling need to be optimized to reduce the maintenance cost and improve the reliability, availability and safety based on the results of fault detection and prediction.Data mining is a computational process of discovering patterns in large data sets involving methods at the intersection of artificial intelligence, machine learning, statistics, and database systems. The goal of the data mining is to extract useful information from a data set and transform it into an understandable structure for further use.This thesis develops framework of Intelligent Fault Diagnosis and Prognosis System (IFDPS) for CBM based on Data Mining Techniques. It mainly includes two tasks: the one is to detect and predict the condition of the equipment and the other is to optimize maintenance scheduling accordingly. It contains several phases: sensor selection and its placement optimization, signal processing and feature extraction, fault diagnosis, fault prognosis and predictive maintenance scheduling optimization based on results of fault diagnosis and prognosis. This thesis applies different data mining techniques containing Artificial Neural Network such as Supervised Back-Propagation (SBP) and Self-Organizing Map (SOM), Swarm Intelligence such as Particle Swarm Optimization (PSO), Bee Colony Algorithm (BCA) and Ant Colony Optimization (ACO), and Association Rule (AI) in most of these phases.The outcomes of the thesis can be applied in mechanical and electrical system in industries of manufacturing, wind and hydro power plants.

Wind energy is one of the fast growing sources of renewable power production currently and there is a great demand to reduce the cost of operation and maintenance to achieve competitive energy price in the market especially for offshore wind farms. An offshore wind farm usually comprises a large number of turbines and thus needs a number of service vessels for maintenance. It is already a complicated task to plan the schedule and route for each of the vessels on a daily basis, dealing with several constraints, such as weather window and maintenance demand, at the same time. Even more challenging is to find an optimal solution. This paper propose a method, i.e. Duo Ant Colony Optimization (Duo-ACO), to improve the utilization of the maintenance resources, specifically the efficient scheduling and routing of the maintenance fleet and thus reduce the operation and maintenance (O&M) cost. The proposed metaheuristic method can help operator to avoid a time-consuming process of manually p...

This paper proposes a new algorithm combining particle swarm optimization (PSO) and simulated annealing (SA) to solve hot strip mill scheduling problem which is formulated as a multi-objective optimization problem with multi-constraints which process schedule constraints, production means constraints, and energy consumption constraints into account. The constraints in hot mill processes are analyzed in three aspects—process schedule, product means, and energy consumption. According to actual experience, Hybrid Vehicle Routing Problem (HVRP) model is established to solve the scheduling problem which is a combination of variable fleet vehicle routing problem, prize collection vehicle routing, and capacitated vehicle routing problem. In this new algorithm, the PSO algorithm is redefined and modified by introducing metropolis criterion of SA twice, which is used to update two extremes of particle swarm, including the individual optimal solution and the global optimal solution to improve the convergence precision and the convergence rate. The proposed method has been applied in a hot mill line to verify its availability and feasibility by comparing the manual scheduling result.

Condition-based Maintenance (CBM) is a maintenance policy that take maintenance action just when need arises with real-time condition monitoring. Intelligent CBM means a CBM system is capable of understanding and making maintenance decisions without human intervention. To achieve this objective, it is needed to detect current conditions of mechanical and electrical systems and predict the fault of the systems accurately. What’s more, the maintenance scheduling need to be optimized to reduce the maintenance cost and improve the reliability, availability and safety based on the results of fault detection and prediction.Data mining is a computational process of discovering patterns in large data sets involving methods at the intersection of artificial intelligence, machine learning, statistics, and database systems. The goal of the data mining is to extract useful information from a data set and transform it into an understandable structure for further use.This thesis develops framework of Intelligent Fault Diagnosis and Prognosis System (IFDPS) for CBM based on Data Mining Techniques. It mainly includes two tasks: the one is to detect and predict the condition of the equipment and the other is to optimize maintenance scheduling accordingly. It contains several phases: sensor selection and its placement optimization, signal processing and feature extraction, fault diagnosis, fault prognosis and predictive maintenance scheduling optimization based on results of fault diagnosis and prognosis. This thesis applies different data mining techniques containing Artificial Neural Network such as Supervised Back-Propagation (SBP) and Self-Organizing Map (SOM), Swarm Intelligence such as Particle Swarm Optimization (PSO), Bee Colony Algorithm (BCA) and Ant Colony Optimization (ACO), and Association Rule (AI) in most of these phases.The outcomes of the thesis can be applied in mechanical and electrical system in industries of manufacturing, wind and hydro power plants.

Wind energy is one of the fast growing sources of renewable power production currently and there is a great demand to reduce the cost of operation and maintenance to achieve competitive energy price in the market especially for offshore wind farms. An offshore wind farm usually comprises a large number of turbines and thus needs a number of service vessels for maintenance. It is already a complicated task to plan the schedule and route for each of the vessels on a daily basis, dealing with several constraints, such as weather window and maintenance demand, at the same time. Even more challenging is to find an optimal solution. This paper propose a method, i.e. Duo Ant Colony Optimization (Duo-ACO), to improve the utilization of the maintenance resources, specifically the efficient scheduling and routing of the maintenance fleet and thus reduce the operation and maintenance (O&M) cost. The proposed metaheuristic method can help operator to avoid a time-consuming process of manually planning the scheduling and routing.