Danko Brezak

Cortical bone drilling and thermal osteonecrosis

BACKGROUND Bone drilling is a common step in operative fracture treatment and reconstructive surgery. During drilling elevated bone temperature is generated. Temperatures above 47°C cause thermal osteonecrosis which contributes to screw loosening and subsequently implant failures and refractures. METHODS The current literature on bone drilling and thermal osteonecrosis is reviewed. The methodologies involved in the experimental and clinical studies are described and compared. FINDINGS Areas which require further investigation are highlighted and the potential use of more precise experimental setup and future technologies are addressed. INTERPRETATION Important drill and drilling parameters that could cause increase in bone temperature and hence thermal osteonecrosis are reviewed and discussed: drilling speed, drill feed rate, cooling, drill diameter, drill point angle, drill material and wearing, drilling depth, pre-drilling, drill geometry and bone cortical thickness. Experimental methods of temperature measurement during bone drilling are defined and thermal osteonecrosis is discussed with its pathophysiology, significance in bone surgery and methods for its minimization.

Tool wear estimation using an analytic fuzzy classifier and support vector machines

A new type of continuous hybrid tool wear estimator is proposed in this paper. It is structured in the form of two modules for classification and estimation. The classification module is designed by using an analytic fuzzy logic concept without a rule base. Thereby, it is possible to utilize fuzzy logic decision-making without any constraints in the number of tool wear features in order to enhance the module robustness and accuracy. The final estimated tool wear parameter value is obtained from the estimation module. It is structured by using a support vector machine nonlinear regression algorithm. The proposed estimator implies the usage of a larger number and various types of features, which is in line with the concept of a closer integration between machine tools and different types of sensors for tool condition monitoring.

Global positioning of robot manipulators with mixed revolute and prismatic joints

The existing controllers for robot manipulators with uncertain gravitational force can globally stabilize only robot manipulators with revolute joints. The main obstacles to the global stabilization of robot manipulators with mixed revolute and prismatic joints are unboundedness of the inertia matrix and the Jacobian of the gravity vector. In this note, a class of globally stable controllers for robot manipulators with mixed revolute and prismatic joints is proposed. The global asymptotic stabilization is achieved by adding a nonlinear proportional and derivative term to the linear proportional-integral-derivative (PID) controller. By using Lyapunovs direct method, the explicit conditions on the controller parameters to ensure global asymptotic stability are obtained.

Tool wear monitoring using radial basis function neural network

This work considers the application of radial basis function neural network (RBFNN) for tool wear determination in the milling process. Tool wear, i.e., flank wear zone widths, have been estimated in two phases using two types of RBFNN algorithms. In the first phase, RBFNN pattern recognition algorithm is used in order to classify tool wear features in three wear level classes (initial, normal and rapid tool wear). On behalf of these results, in the second phase, RBFNN regression algorithm is utilized to estimate the average amount of flank wear zone widths. Tool wear features were extracted in time and frequency domain from three different types of signals: force, acoustic emission and nominal currents of feed drives.

Passive Finite-Dimensional Repetitive Control of Robot Manipulators

In this paper, a new class of finite-dimensional repetitive controllers for robot manipulators is proposed. The global asymptotic stability is proved for the unperturbed system. The passivity-based design of the proposed repetitive controller avoids the problem of tight stability conditions and slow convergence of the conventional, internal model-based, repetitive controllers. The passive interconnection of the controller and the nonlinear mechanical systems provides the same stability conditions as the controller with the exact feed-forward compensation of robot dynamics. The simulation results on a three degrees of freedom spatial manipulator illustrate the performances of the proposed controller.

Drill wear monitoring in cortical bone drilling

Medical drills are subject to intensive wear due to mechanical factors which occur during the bone drilling process, and potential thermal and chemical factors related to the sterilisation process. Intensive wear increases friction between the drill and the surrounding bone tissue, resulting in higher drilling temperatures and cutting forces. Therefore, the goal of this experimental research was to develop a drill wear classification model based on multi-sensor approach and artificial neural network algorithm. A required set of tool wear features were extracted from the following three types of signals: cutting forces, servomotor drive currents and acoustic emission. Their capacity to classify precisely one of three predefined drill wear levels has been established using a pattern recognition type of the Radial Basis Function Neural Network algorithm. Experiments were performed on a custom-made test bed system using fresh bovine bones and standard medical drills. Results have shown high classification success rate, together with the model robustness and insensitivity to variations of bone mechanical properties. Features extracted from acoustic emission and servomotor drive signals achieved the highest precision in drill wear level classification (92.8%), thus indicating their potential in the design of a new type of medical drilling machine with process monitoring capabilities.

A comparison of feed-forward and recurrent neural networks in time series forecasting

Forecasting performances of feed-forward and recurrent neural networks (NN) trained with different learning algorithms are analyzed and compared using the Mackey-Glass nonlinear chaotic time series. This system is a known benchmark test whose elements are hard to predict. Multi-layer Perceptron NN was chosen as a feed-forward neural network because it is still the most commonly used network in financial forecasting models. It is compared with the modified version of the so-called Dynamic Multi-layer Perceptron NN characterized with a dynamic neuron model, i.e., Auto Regressive Moving Average filter built into the hidden layer neurons. Thus, every hidden layer neuron has the ability to process previous values of its own activity together with new input signals. The obtained results indicate satisfactory forecasting characteristics of both networks. However, recurrent NN was more accurate in practically all tests using less number of hidden layer neurons than the feed-forward NN. This study once again confirmed a great effectiveness and potential of dynamic neural networks in modeling and predicting highly nonlinear processes. Their application in the design of financial forecasting models is therefore most recommended.

GPU implementation of the feedforward neural network with modified Levenberg-Marquardt algorithm

In this paper, an improved Levenberg-Marquardt-based feedforward neural network, with variable weight decay, is suggested. Furthermore, parallel implementation of the network on graphics processing unit is presented. Parallelization of the network is achieved on two different levels. First level of parallelism is data set level, where parallelization is possible due to inherently parallel structure of the feedforward neural networks. Second level of parallelism is Jacobian computation level. Third level of parallelism, i.e. parallelization of optimization search steps, is not implemented due to the variable weight decay, which makes third level of parallelism redundant. Suggested weight decay variation enables the compromise between higher accuracy with oscillations on one side and stable, but slower convergence on the other. To improve learning speed and efficiency, modification of random weight initialization is included. Testing of proposed algorithm is performed on two real domain benchmark problems. The results obtained and presented in this paper show effectiveness of proposed algorithm implementation.

Passive internal model based repetitive control of robot manipulators

In this paper, a new class of globally stable finite dimensional repetitive controller for robot manipulator is proposed. The passivity based design of the proposed repetitive controller avoid the problem of tight stability conditions and slow convergence of the conventional, internal model based, repetitive controllers. The passive interconnection of the controller with nonlinear mechanical systems provide stability margin which is the same as stability margin of the controller with exact feed-forward compensation of robot dynamics. The simulation results illustrate the convergence properties of the proposed controller

Initial conditions optimization of nonlinear dynamic systems with applications to output identification and control

The paper presents a gradient-based algorithm for initial conditions optimization of nonlinear multivariable systems with boundary and state vectors constraints. The algorithm has a backward-in-time recurrent structure similar to the backpropagation-through-time (BPTT) algorithm, which is mostly used as a learning algorithm for dynamic neural networks. It is shown that dynamic parameter optimization problem can be formulated as the initial conditions optimization problem. Further, it is shown that output parameter identification and output controller design problems can be formulated as dynamic parameter optimization problem. The effectiveness of the proposed algorithm is demonstrated on the problem of output identification and control of a nonlinear two-mass torsional system.