This paper describes new approach to the synthesis of a multivariable control problem for a power system with the field flux decays of AC generator. The power system dynamics can be formulated as a class of nonlinear system in which are contained the product-type nonlinearities between phase-angle and field-flux linkage of the generator, and the dynamical system is represented by ordinary differential equations divided into linear and nonlinear terms. Moreover we have introduced the measurable nonlinear function values consisting of the terminal voltage and electrical power output of the generator as an output vector. In this paper, a feedback control law is derived by minimizing the finite-time integral performance index with a quadratic form of state vector, control vector and also nonlinear function vector of system. In this theoretical scheme, it is shown that the existence conditions of the control law are concerned closely with the minimal realization problem. Through the approach described above, the feedback controller for a synchronous generator with field and governor dynamics connected to an infinite busbar by transmission line is synthesized by using the optimization method. The simulation results are given numerically for damping large oscillations associated with transient conditions in this system.
Surge voltages evaluated by a digital analysis method, for example Electromagnetic Transients Program (EMTP) and by an analog analysis method (Transient Network Analyzer: TNA) are compared with experimental results on a scaled model of a gas insulated line. From the results, it is confirmed that the analog analysis method has a satisfactory accuracy and its computing speed is quite high. The error caused by the lumped element modeling of the gas insulated line is comparatively small, and the line is not necessary to be treated as a distributed parameter line. The lumped element model is advantageous to understand the mechanism of the transient characteristic of the induced transint voltage. A simple equivalent circuit of the gas insulated line is proposedand its accuracy is confirmed by a comparison with the calculated result obtained by an accurate line model. The transient overvoltages of the tank and across the insulated flange can be analytically calculated using the equivalent model.
A new control concept for Solid State Phase-Shifter (SSPS) connected to synchronous machines is presented. This concept enables a rotary condenser to yield the active power with fast response from the flywheel effect in the rotating speed. The relations between the power and energy from the flywheel effect and the phase shift angle are presented. The experimental results using a simulator show good agreement with the theoretical energy yield. The present study shows that the rotary condenser connected to the main network with SSPS can significantly increase the stability of the main network system. As an example, some cases of one transmission line faults in double transmission system are demonstrated. The computer simulations show a possibility of dynamic power control using the SSPS and flywheel effect of the rotary condenser as an energy storage.
For more rational operation and control of the future electric power system which would continue to expand in size and complexity, an autonomous decentralized control system is proposed as one of the new control architecture. Under the autonomous decentralized control system, each generating station and substation behaves independently and cooperatively without being controlled by the central station. Besides the inherent fault tolerant characteristics of the architecture, flexible and rapid control depending on the circumstances can be achieved. In this paper, the fundamental logic of preventive control under autonomous decentralized system is developed. It is based on the bidding method in which each station proposes own output taking its own situation and cooperation with others into consideration. The economic load dispatch including preventive control can be performed based on the bidding rule adopting the equivalent incremental generating cost curve in which the security constraints are incorporated. Application of the logic to a model system revealed satisfactory control performance for preventing the overload of transmission lines and the undervoltage.
The Hokkaido-Honshu DC transmission line, the first DC transmission line in Japan, has been successfully operated for the past 11 years since its operation started in December, 1979 with 125kV and its operating voltage was increased to 250kV in June, 1980. This DC line, which interconnects the Hokkaido and Mainland of Japan, is located at a distance ranging from 0.5km to 12km from the seacoast. Under such circumstances, the authors have been concerned about insulator contamination flashover and the corrosion of hardware, especially pin, due to the electrolysis. Thus a series of periodical ESDD measurements and investigation of the insulator hardware corrosion have been conducted. The investigation results show that the original design ESDD values are adequate and the effectiveness of zinc sleeve against the electrolytic corrosion of pin has been clarified. Reduction in weight of zinc sleeve is not only due to the electrolysis but also due to the natural corrosion in the corrosive atmosphere near the seacoast.
In this paper, we proposed a model which permits us to locate the fault point of transmission line in more detail by analyzing the distribution patterns of the current induced in overhead ground wires. Different two neural networks share the fault location processings. One determines a fault section divided by current sensors using overall current distribution pattern. The other determines a fault subsection within the predetermined section by the first neural network using the local current distribution pattern alone. This model is established to follow the procedure of human experts to locate a fault point. This procedure was applied to three cases of data obtained from a commercial line. It was confirmed that the subsectionalized resolution of this model is improved by 3_??_4 times better than the conventional system or the model using only a 3-layer neural network. By detailed simulations with a model transmission line, learning time has also been reduced by about one-20th while maintaining the accuracy and generalization performance of this model. Finally, we showed a method to obtain the best range resolution for given sensor intervals and thus defined the performance boundary of our model.
A new method was attempted in order to measure characteristics of partial discharge (PD) for an active 9.5km long-distance 275 kV XLPE cable line with insulating joints. It was found that the detection of PD and the calibration of detection sensitivity were carried out safely and easily. The proposed method includes (1) detection of PDs and injection of calibration pulses using two pairs of metal electrodes laid on PVC jackets with anti-corrosion layers separated by a joint sleeve sectionalizing insulator, (2) measurement of detection sensitivity with distance using calibration pulses injected from an adjacent insulating joint, (3) detection of PDs with a high S/N ratio measured at a frequency corresponding to the lowest noise level, (4) usage of rf cores attached at cross bonding leads and wires for sheath current limiters to reduce the crosstalk of signal and noise through these lines at insulating joints. Using the combination of these technique, the highest detection sensitivity was 1 pC in the vicinity of an insulating joint, and was 15 pC at 617m distant from the adjacent joint. The sensitivity obtained by the conventional PD measurement using low frequency method was at most 1, 000 pC.
Investigations are made on the simulation of reignition high-frequency arc extinction that occur during interrupting inductive currents with SF6 gas circuit breakers. This simulation was conducted by employing a combination of the Mayr-type arc equation and the circuit equation. The simulations conventionally conducted by using the Mayr-type arc equation of large short-circuit current interruption pertained in most cases to the period in the neighborhood of one current zero point. In this report, however, simuiation was made on high-frequency current which passed the zero point a number of times. The gist of our report can be summarized as follows: (1) A test circuit was prepared by simulating reignition and high-frequency arc extinction, and tests are conducted by using a 300 kV SF6 gas model circuit breaker. (2) High-frequency current and arc conductance were measured, and from the measurement results, the arc parameters (arc time constants and arc losses) of the Mayr-type arc equation were read. (3) The phenomenon of high-frequency arc extinction was calculated through simulating the test circuits, by applying the obtained arc parameters to the Mayr-type are equation. The occurrence limits of high-frequency arc extinction were found to closely correspond to the actual measurement results. (4) Studies were made on those clarified arc parameter values as well as on respective effects on high-frequency arc extinction by frequency of high-frequency current, transient recovery voltage, and attenuation of high-frequency current.
This paper describes the water tree diagnoses of 6.6 kV class XLPE cables by DC component method. This method measures the DC current component by energizing the cable with AC high voltage, therefore, it would be suitable for on-line diagnoses. The DC component is considered to appear due to the rectifying effect of water trees inside the insulation. In the laboratory test, the DC component showed a good correlation with the degradation. However, almost no correlation was seen in the on-site measurement. The authors found that the cause of this bad correlation is galvanic effect inside the jacket. This generates the noise current through the grounding circuit when the jacket resistivity is low. Based on this finding, the equivalent circuit of on-site measurement was proposed. Considering the equivalent circuit, two methods were proposed to discriminate the true DC component out of noise current. One is to by-pass the noise current from the measurement apparatus, and the other is to statistically estimate the true DC component by using the transition in the jacket resistivity. The on-site test were carryed out to confirm the applicability of these methods.
This paper proposes a new testing method for detecting a local short circuit cell which causes unusual high temperature rise and unusual cell voltage rise of the adjoing cell at the operating stack. The method utilizes lower cell voltage of the local short circuit cell than that of normal cells when small constant DC current is supplied for some time. As low cell voltage of the short circuit portion spread all over the cell, the test after stacking can be carried out by measuring cell voltages of exposed edge part of each cell at the stack side. In order to increase detecting sensitivity, low testing temperature is recomended. At the test in air, current density and the duration of current flow must be strictly limited within the level which the cell voltage does not saturate or carbon dioxide will generate from the electrode due to oxidation of carbon. At the test in nitrogen, detecting sensitivity can increase by using higher current density and longer duration of current flow than that at the test in air without any oxidation of the electrode. This method is shown to have enough sensitity to detect harmful level of local short circuit resistance.