M. Giesselmann

Using neural networks to estimate wind turbine power generation

This paper uses data collected at Central and South West Services Fort Davis wind farm (USA) to develop a neural network based prediction of power produced by each turbine. The power generated by electric wind turbines changes rapidly because of the continuous fluctuation of wind speed and direction. It is important for the power industry to have the capability to perform this prediction for diagnostic purposes-lower-than-expected wind power may be an early indicator of a need for maintenance. In this paper, characteristics of wind power generation are first evaluated in order to establish the relative importance for the neural network. A four input neural network is developed and its performance is shown to be superior to the single parameter traditional model approach.

Comparative Analysis of Regression and Artificial Neural Network Models for Wind Turbine Power Curve Estimation

This paper examines and compares regression and artificial neural network models used for the estimation of wind turbine power curves. First, characteristics of wind turbine power generation are investigated. Then, models for turbine power curve estimation using both regression and neural network methods are presented and compared. The parameter estimates for the regression model and training of the neural network are completed with the wind farm data, and the performances of the two models are studied. The regression model is shown to be function dependent, and the neural network model obtains its power curve estimation through learning. The neural network model is found to possess better performance than the regression model for turbine power curve estimation under complicated influence factors.

Compact, high power capacitor charger

We are developing compact, high-power chargers for rapid charging of energy storage capacitors. The main application is presently rapid charging of the capacitors inside of compact Marx generators for reprated operation. Compact Marx generators produce output pulses with amplitudes above 300 kV with ns or subns rise-times. A typical application is the generation of high power microwaves. Initially all energy storage capacitors in a Marx generator are charged in parallel. During the so-called erection cycle, the capacitors are connected in series. The charging voltage in the parallel configuration is around 40-50 kV. The input voltage of our charger is in the range of several hundred volts. Rapid charging of the capacitors in the parallel configuration will enable a high pulse repetition-rate of the compact Marx generator. The high power charger uses state-of-the-art IGBTs (isolated gate bipolar transistors) in an H-bridge topology and a compact, high frequency transformer. The IGBTs and the associated controls are packaged for minimum weight and maximum power density. The packaging and device selection makes use of burst mode operation (thermal inertia) of the charger. The present charger is considerably smaller than the one presented in Giesselmann, M et al., (2001).

Rapid Capacitor Charger for 10 HZ Operation of a Low-Inductance Compact Marx Generator

We designed and build a rapid capacitor charger for 10 Hz, 500 J/shot operation of a low-inductance, compact Marx generator. The charger uses a hard-switched IGBT H-Bridge Inverter, which drives a 30 kHz, nano- crystalline step-up transformer. The transformer, in addition to the high-voltage rectifier and a trigger- transformer are contained in a section which is filled with transformer oil. The main circuit board also contains a solid-state Marx generator to trigger the main Marx generator. We also implemented a self-powered HV-feedback sensor to stop the charge process precisely at the target voltage. This new sensor greatly enhanced the rep-rated performance of the Marx by preventing pre-fires, since it enabled us to charge aggressively without overshooting the target voltage and have more time for spark-gap recovery.

Experimental and analytical investigation of a pulsed power conditioning system for magnetic flux compression generators

The pulsed power conditioning system (PPCS) is one of the key enabling technologies for using the energy output of a magnetic flux compression generator (MFCG). This paper shows the results of comprehensive experimental studies of an inductive energy storage system using an exploding wire fuse. The effects of metal oxide varistors (MOVs) for use as pulse-shaping devices are also presented. The experimental results are complemented by a comprehensive evaluation and interpretation of the results using the tools available in the professional version of MathCAD.

Rapid capacitor charger

The increased interest in radio-frequency (RF) weapons for electronic warfare has increased the demand for rapid capacitor chargers. In most cases these charger have to operate from a DC power source and typically have to charge a capacitor bank (Marx bank) at a rate of about 10 Hz. Besides the requirements for energy delivery, the space constraints are crucial. In order to fulfill both, major components have to be operated above their specifications. However, they only need to perform for a short burst with enough time between events to enable use of thermal inertia.

Extended Kalman Filter Training of Neural Networks on a SIMD Parallel Machine

The extended Kalman filter (EKF) algorithm has been shown to be advan- tageous for neural network trainings. However, unlike the backpropagation (BP), many matrix operations are needed for the EKF algorithm and therefore greatly increase the computational complexity. This paper presents a method to do the EKF training on a SIMD parallel machine. We use a multistream decoupled extended Kalman filter (DEKF) training algorithm which can provide efficient use of the parallel resource and more improved trained network weights. From the overall design consideration of the DEKF algorithm and the consideration of maximum usage of the parallel resource, the multistream DEKF training is realized on a MasPar SIMD parallel machine. The performance of the parallel DEKF training algorithm is studied. Comparisons are performed to investigate pattern and batch-form trainings for both EKF and BP training algorithms.

High Voltage Impulse Generator Using HV-IGBTs

We are reporting on a High-Voltage Impulse Generator, which consists of a step-up transformer, which is driven by new HV-IGBTs (High-Voltage Isolated Gate Bipolar Transistors). The new HV-IGBTs are individually packaged silicon-dies intended for Pulsed-Power Applications. The silicon dies are normally packaged in large modules for locomotive motor drives and similar traction applications. In our work we used the Powerex QIS4506001 discrete IGBT and the QRS4506001 discrete diode, both with a nominal rating of 4500V/60A, derived from continuous- duty applications. Our experiments have shown that the devices are capable of handling currents in excess of 1 kA during pulsed operation.

Reliability-constrained self-organization and energy management towards a resilient microgrid cluster

Microgrids, as individual controllable entities that can operate either islanded from or interconnected to main power grid, have emerged as a promising solution to improving energy efficiency and resilience to disturbance. When linked together in a self-organized manner, a cluster of microgrids can significantly enhance the reliability and power quality for critical load. With this insight, we study the self-organization and decentralized energy management of a microgrid cluster islanded from main grid after a disruptive event. In the self-organization stage, depending on the available generation resources, each microgrid decides on whether to connect to the cluster; and the microgrid energy management systems then “negotiate” on the optimal power exchange with each other in the cluster. Once the power exchange is determined, the generation and storage resources of each microgrid are managed to guarantee the energy reliability of critical loads and overall energy efficiency, through a scheduling procedure followed by a dispatch procedure. The effectiveness of the proposed method is revealed via case studies.

Rapid capacitor charging power supply for an 1800J PFN

The RCC (rapid capacitor charger) previously developed at the P3E Center [1] has been adapted to charge an 1800 J PFN (pulse forming network) for rep-rated operation. The entire automated system to test and evaluate SGTOs (Super Gate turn-off Thyristors) runs at a 1 Hertz repetition rate; thus requiring a power supply to charge the PFN within 500 ms and have a 3.6 kJ/s average power capability to allow for data acquisition and storage between shots. The hard-switching H-bridge topology with 10 kW burst mode handling capability is very well suited for this compact table top system design. The control of the RCC has been shifted to a PIC controller responsible for PFN charging. Charging parameters include: an adjustable charging time from 50 to 500 ms, high voltage monitoring with adjustable voltage level, and RCC Go/shut-off. All charging parameters are determined by the main CPU handling the automation process and are sent to the PIC controller before each PFN charging event. With the addition of forced air cooled heat-sink for the IGBT modules, enough heat can be removed to allow continuous automated operation.