Search

Dissolved gas analysis (DGA) is vital to status evaluation of power transformers. In this study, a solid oxide fuel cell (SOFC) gas sensor was fabricated and applied to DGA. Then, a gas chromatographic system was developed on basis of the SOFC detector. From Nernst function of SOFC, a mathematical model based on oxygen consumption accumulation (OCA) has been proposed to quantify trace gases dissolved in oil. Based on the OCA of SOFC, the gas concentration can be calculated though the chromatogram directly, no calibrating gas is needed, thereby greatly simplifying the quantification process. Following peak detection based on the improved filter matching method, tests were conducted to verify the validity of the OCA method. The experimental results demonstrated that the repeatability represented by relative standard deviation is less than 0.5%; high precision can be achieved, measurement error for five feature gases are less than 10% at given concentrations, providing a competitive performance against other gas sensors with curve fitting method.

In recent years, a novel algorithm named crisscross search optimization (CSO) algorithm has been successfully applied in the conventional energy economic emission dispatch (EED) problems of pure thermal power system (PTPS) and hydrothermal generation system (HTGS). However, there still have some problems, such as slow convergence speed and low stability. To address these issues, an extended crisscross search optimization (ECSO) algorithm is proposed in this paper. The performances of the CSO algorithm are improved by an adaptive choice procedure of the extension coefficient. And a weakening equality constraint method is used in the MOEED problems for ECSO. To test the performance of the proposed algorithm, the IEEE-30 bus System (Test System-Ⅰ), the 40 generators System (Test System-Ⅱ) and the hydrothermal generation system (HTGS) (Test System-Ⅲ) are adopted. Experimental results show that the cost of economic operation and the pollutant emission with the proposed ECSO are minimum in these test systems. Further, the simulation and comparison results show the robustness of the ECSO is superior to the CSO and the other algorithms.

Accurate wind power forecasting is of great significance for power system operation. In this study, a triple-stage multi-step wind power forecasting approach is proposed by applying attention-based deep residual gated recurrent unit (GRU) network combined with ensemble empirical mode decomposition (EEMD) and crisscross optimization algorithm (CSO). In the data processing stage, the EEMD is used to decompose the wind power/speed time series and a bi-attention mechanism (BA) is applied to enhance the sensitivity of model to the important information from both time and feature dimension. In the prediction stage, a series-connected deep learning model called RGRU consisting of the residual network and GRU is proposed as the forecasting model, aiming to make full use of extracting the static and dynamic coupling relationship among the input features. In the retraining-stage, a high-performance CSO algorithm is adopted to further optimize the fully-connected layer of RGRU model so as to improve the generalization ability of the model. The proposed method is validated on a wind farm located in Spain and the experimental results demonstrate that the proposed hybrid model has significant advantage over other state-of-the-art models involved in this study in terms of prediction accuracy and stability.

Accurate forecasts of wind power generation are essential for the operation of wind farms. But for the newly developed stations, it is difficult to make accurate prediction because there are no sufficient historical data available. It will thus be interesting to explore new data augmentation and prediction modeling approach adaptive to such new-built wind farms. In this regard, a novel asexual-reproduction evolutionary neural network (ARENN) for short-term wind power prediction based on Wasserstein generative adversarial network with gradient penalty (WGANGP) and ensemble empirical mode decomposition (EEMD) is presented in this paper. To solve the dilemma that new-built wind farms lack sufficient wind power data, the WGANGP is first applied to generate realistic data with a similar distribution of real data to augment the training dataset, which is further decomposed into a series of more stable subsequences by the EEMD so as to reduce the prediction difficulty of the machine learning model. In this study, a novel ARENN prediction model is developed to make the short-term wind power prediction, in which an asexual-reproduction evolutionary approach is first proposed to optimize the neural network based on a set of different loss functions that facilitate the population of network parameters approximating to the global optimum along different error surfaces in the evolutionary process. The proposed approach is validated on the data collected from the wind farm located in Spain and the predicted results demonstrate the advantage of our proposed approach over other methods involved in this study.

Although machine learning methods have been widely applied in the wind power prediction field, they are not suitable for building the prediction model of a new-built wind farm because of no sufficient historical data. In this study, a novel deep transfer learning approach is proposed for addressing the few-shot learning problem in multi-step ahead wind power prediction. In the pre-training stage, several convolutional neural networks (CNNs) in parallel are separately connected to the long short-term memory network (LSTM), thus forming a unique serio-parallel CNNs-LSTM (CL) feature extractor. The CL utilizes the CNNs and LSTM to extract both the meteorological and temporal feature information of the neighboring wind farms for facilitating the prediction modeling of the source wind farm. In the transfer-training stage, a transfer strategy is designed to transfer partial network parameters of a well-trained CL feature extractor to construct the prediction model of the target wind farm. In addition, a personalized-training strategy is implemented by using crisscross optimization (CSO) to retrain the parameters of fully-connected layer. The proposed method is validated on a swarm of wind farms located in China and the experimental results show its obvious superiority over the non-transfer models involved in this study.

Large-scale integration of wind energy into electric grid is restricted by its inherent intermittence and volatility. So the increased utilization of wind power necessitates its accurate prediction. The contribution of this study is to develop a new hybrid forecasting model for the short-term wind power prediction by using a secondary hybrid decomposition approach. In the data pre-processing phase, the empirical mode decomposition is used to decompose the original time series into several intrinsic mode functions (IMFs). A unique feature is that the generated IMF1 continues to be decomposed into appropriate and detailed components by applying wavelet packet decomposition. In the training phase, all the transformed sub-series are forecasted with extreme learning machine trained by our recently developed crisscross optimization algorithm (CSO). The final predicted values are obtained from aggregation. The results show that: (a) The performance of empirical mode decomposition can be significantly improved with its IMF1 decomposed by wavelet packet decomposition. (b) The CSO algorithm has satisfactory performance in addressing the premature convergence problem when applied to optimize extreme learning machine. (c) The proposed approach has great advantage over other previous hybrid models in terms of prediction accuracy.

This paper formulates the optimal power flow (OPF) problem with the consideration of minimizing many objective functions including the basic fuel cost, fuel cost with valve-point effects, transmission active power loss, basic fuel cost with transmission active power loss as well as basic fuel cost with voltage deviation. To solve the OPF problem, a novel crisscross search based grey wolf optimizer (CS-GWO) is proposed, in which the hunting operation in GWO is firstly modified by introducing a greedy mechanism and then the horizontal crossover operator is added to refine the first three ranking wolves. In addition, the vertical crossover operator is applied to maintain the population diversity so as to prevent the premature convergence, which provides a unique mechanism for GWO to get rid of dimensional local optimum. The cooperation of last two operators can accelerate convergence speed and avoid falling into dimensional local optimum of hunting process. The proposed CS-GWO is validated on IEEE 30-bus system and IEEE 118-bus system. The experimental results demonstrate the CS-GWO has obvious advantage over the original GWO and the other state-of-art heuristic algorithms, especially in large-scale system.

As one of important optimization problems in power system, economic dispatch (ED) with multiple fuel options is characterized by high non-convexity, non-linearity and discontinuity. The combined action of multiple fuel options and valve-point effects increases the degree of difficulty to solve the ED problem. In this paper, a recently developed heuristic algorithm called crisscross optimization algorithm (CSO) is attempted to address the large-scale and non-convex ED problem with both multiple fuel options and valve-point effects taken into account. The proposed CSO method solves the ED problem through horizontal crossover and vertical crossover. The former searches for the new solutions within a half population of hyper-cubes by adopting a cross-border search approach while the latter provides a unique mechanism to prevent from the premature convergence problems based on the concept of dimensional local minimum. Both operators alternatively generate moderation solutions which are subsequently updated by an elite selection strategy. The proposed method is validated on six test systems consisting of 10–640 generating units and compared with other state-of-the-art methods in the literature. The results show that CSO yields higher quality solutions especially for solving large-scale ED problems with multiple fuel options.

Wind speed forecasting is of great significance for wind farm management and safe integration into electric power grid. As wind speed is characterized by high autocorrelation and inherent volatility, it is difficult to predict with a single model. The aim of this study is to develop a new hybrid model for predicting the short wind speed at 1 h intervals up to 5 h based on wavelet packet decomposition, crisscross optimization algorithm and artificial neural networks. In the data pre-processing phase, the wavelet packet technique is used to decompose the original wind speed series into subseries. For each transformed components with different frequency sub-bands, the back-propagation neural network optimized by crisscross optimization algorithm is employed to predict the multi-step ahead wind speed. The eventual predicted results are obtained through aggregate calculation. To validate the effectiveness of the proposed approach, two wind speed series collected from a wind observation station located in the Netherlands are used to do the multi-step wind speed forecasting. To reduce the statistical errors, all forecasting methods are executed 50 times independently. The results of this study show that: (1) the proposed crisscross optimization algorithm has significant advantage over the back-propagation algorithm and particle swarm optimization in addressing the prematurity problems when applied to train the neural network. (2) Compared with the previous hybrid models used in this study, the proposed hybrid model consistently has the minimum mean absolute percentage error regardless of one-step, three-step or five-step prediction.

Founded on bounded rationality and limited information, evolutionary game theory can well describe the evolution rule of population behavior and predict individuals’ decision-making behavior. Therefore, this paper focuses on the general N-population multi-strategy evolutionary games, and uses them to investigate the generation-side long-term bidding issues in electricity market. First, the long-term equilibrium characteristics of typical two-population and three-population two-strategy evolutionary game scenarios are thoroughly investigated through theoretical analysis and dynamic simulation, where novel relative net payoff parameters are completely defined for these games in engineering. Research shows that the long-term evolutionary stable equilibria are only determined by the relative net payoff parameters, so that an expected evolutionary stable equilibrium can be obtained by adjusting these parameters. Second, the modeling idea of general N-population multi-strategy evolutionary games is elaborated based on replicator dynamics. In the case study, the evolutionary stable equilibrium of generation-side long-term bidding is investigated for a supply-side market involving different generator populations. This case effectively verifies the evolutionary dynamics of the general N-population multi-strategy evolutionary games built in this paper. Lastly, future investigations of evolutionary game theory are prospected. Overall, this paper explores the long-term equilibrium characteristics of the general N-population multi-strategy evolutionary games, which can provide some inspirations and theoretical reference for researches on complex long-term dynamic interactive decision-making problems of group participants with bounded rationality in some relevant fields, especially in the economics, management, and engineering fields.

Short-term hydrothermal generation scheduling (SHGS) considering various hydraulic and electric constraints is a complex non-convex optimization problem. The coupling connection of cascaded reservoirs and the valve-point effects of thermal units greatly increase the difficulty of finding optimal solution. This paper presents an efficient solution to the SHGS problem by using a novel crisscross optimization (CSO) algorithm, which generates the optimal scheduling results by applying two distinctive search operators, i.e., horizontal crossover and vertical crossover. The horizontal crossover is used as a global optimizer that can reduce the search blind spots in complex solution space through a cross-border search strategy. The vertical crossover is used to address the premature convergence problem by applying a unique dimensional crossover mechanism. Both search operators take turns to generate moderation solutions and CSO always maintains a population of historically best solutions by using a greedy strategy so as to speed up the convergence speed. To investigate the CSO’s performance on the SHGS problem, three test systems widely adopted in the literature are used to do the short-term hydrothermal generation scheduling. The results reveal that the proposed algorithm has good robustness and outperforms other state-of-the-art methods in terms of solution quality.

This paper proposes a method to identify the equipment anomalies based on convolution neural network, aiming at the weak-light situation inside the cable tunnel. This paper has proposed a method to identify the equipment anomalies with weak light situation inside the cable tunnel, based on the convolution neural network. On the basis of the gray image, this method adds the Sobel operator to enhance edge-preprocessing effect and start training through Convolution Neural Network (CNN). The convergence criteria is the Loss Function related with the Weight Parameter W and Bias Parameter b. The convergence method is the one named Backpropagation, which updates the parameters each time to reduce the loss. The fast operating speed of full connection layer can help getting the direct classification result of equipment status in the image. Based on the experimental analysis of the internal images of the Zhuhai tunnel, it can be seen that this method is suitable for the dark and chaotic environment of the tunnel. Additionally, it has a high recognition rate for the image segmentation of the lighting equipment and a high accuracy for the classification of the abnormal situation of the image.

With the improvement of industrialization, people’s requirements for power supply quality also increase year by year. Therefore, how to ensure the normal supply of electricity has become a very concerned issue for power maintenance personnel. In this paper, we use the image captured by the cable tunnel inspection system to locate the cable connector in the image and map it to the infrared image based on the convolution neural network method. Based on the Analysis of the temperature of the cable connector in time, the monitoring system is able to make an alarm on the abnormal situation which can maintain the safety of power supply and extend the service life of the cable. Aiming at the small number of sample images collected, a transfer learning method is adopted, which reduces the training intensity and ensures a better positioning and recognition effect.

In this paper, we propose an inspection scheme based on leader-follower formation for mobile sensors robots to accomplish the task of inspecting a cable tunnel to detect potentially danger and provide warning mechanism. We adopt leader-follower formation control to achieve formation and artificial potential field method to avoid the occasional appeared obstacle. The leader possesses abilities of mission planning and formation changing to conduct the entire inspection and the followers simply track the movement of the leader. Unlike the usual tunnel inspection with a single robot, the strategy we proposed in this paper has advantages of wider inspection range, higher accuracy and higher tolerance through the utilization of multi robots system. Simulation results are presented to validate the effectiveness of the proposed strategy.

As cogeneration plays an increasingly important role in energy utilization, the combined heat and power economic dispatch (CHPED) becomes an important task in power system operation. In this paper, a novel crisscross optimization (CSO) algorithm is implemented to solve the large scale CHPED problem, which is a challenging non-convex optimization problem with a large number of local minima. The feature of applying CSO to address the CHPED problem lies in two interacting operators, namely horizontal crossover and vertical crossover. The horizontal crossover searches for the new solutions within a half population of hyper-cubes with a large probability while in their respective peripheries with a decreasing probability. The vertical crossover provides a effective mechanism for those stagnant dimensions of a population to escape from premature convergence. The combination of both gifts CSO with a powerful global search ability. The effectiveness of the proposed method is validated on six cogeneration systems with different characteristics. The numeric results demonstrates that the proposed CSO method achieves much better results than other methods reported in the literature. To investigate the robustness and applicability of CSO in large power system, two new systems with 96 and 192 units by duplicating the system of case 4 two times and four times are also studied. The results obtained substantiates the suitability of CSO for large-scale constrained CHPED problem.

Wind power forecasting is crucial for the economic dispatch and operation of power system. In this study, a novel hybrid wind power prediction approach is proposed by applying a cascaded deep learning model to extract the implicit meteorological and temporal characteristics of each subseries generated by a two-layer of mode decomposition method. In the proposed model, the empirical mode decomposition is employed to decompose the original time series into a set of intrinsic mode functions (IMFs) and the variational mode decomposition is applied to further decompose the IMF1 sub-layers into several sub-series because of the irregular feature of IMF1. To make use of the coupling relationship between wind power sub-layer, wind speed sub-layer and wind direction, convolutional neural network is used to extract the implicit features of these relationship and then long short-term memory utilizes these features as inputs and further extract the temporal correlation hidden features in each time sub-series. The eventual predicted results are obtained by superimposing the predicted values of all subsequences. The experimental results illustrate that: (a) The prediction performance is obviously improved when the proposed two-layer of decomposition is considered. (b) To achieve better prediction accuracy, it is proven to be an effective way to apply convolutional neural network and long short-term memory to extract the implicit meteorological relationship and the temporal correlation characteristic hidden in each decomposed time sub-series, respectively. (c) The proposed hybrid model outperforms other hybrid models involved in this study and shows a promising prospect in the short-term wind power prediction.

It is often long time-consuming for applying the traditional genetic algorithm to the PID optimization of hydro-turbine governor. To address such issue, a multi-agent based distributed genetic algorithm (MAGA) is proposed. Based on establishment of the non-linear digital simulation model of generating units, a distributed mobile computing platform implementing MAGA is built using agent middle ware, on which, the optimization for the self-adaptive PID governor is performed under different operating conditions. The simulation results show that the proposed MAGA can not only obtain good optimization performance compared with the conventional genetic algorithm, but shorten the optimization time significantly. Apparently, the proposed method provides a new perspectives and solutions to solve the optimization problem of complex control system.