In this paper, a wind power probabilistic prediction optimization algorithm based on working condition identification and probabilistic prediction is proposed. Based on the historical data of power forecasting in different wind resource regions, the probability distribution of wind power forecasting error is estimated by working condition identification and kernel density function, the wind power bandwidth forecasting result is generated by the scenario sampling method, and the daily power balance plan is formulated in combination with the grid load fluctuation. The actual optimization calculation results show that the method in this paper can save the reserve capacity and peak shaving margin of the power grid, and improve the economy and safety of wind power consumption.

Addressed in this paper is the observer -based waiting event -triggered control issue in multiarea load frequency control (LFC) power system subject to random communication delay and data packet dropout. In order to maintain the grid frequency stability of the multi-area LFC power system in an open communication network, an observer -based waiting event-triggered scheme (WETS) is proposed to deal with frequency fluctuation and extend the life of control equipments. Compared with the existing event -triggered mechanism, the proposed WETS has the merits of continuous event -triggered scheme and the periodic sampling event-triggered scheme in the both of theory and application. Then, by well considering the negative factors of data packet dropout and random time delay in data transmission, a stochastic delay -dependent multi -area LFC model is established, which involves the information of observer-based WETS, random communication delay and packet dropout in a unified structure. Further, an improved Lyapunov-Krasovskii functional (LKF) is constructed by dividing the interval of transmitted delay. Based on the constructed Lyapunov functional, Park's theorem and Jensen's integral inequality, some less conservative stabilization criteria are obtained. Finally, the validity of the proposed control method is verified via numerical examples.

The control strategy of reactive power and voltage, communication between systems and the performance of the single device will affect the performance of the reactive and voltage control of the field station. On the basis of one million kilowatts of wind power base in the past three years the actual operation data and field test data, firstly analyzes the status of the wind farm without the system voltage and reactive power control system, introduces the current voltage and reactive power control system architecture and common control strategy; secondly from the station of reactive power and voltage control of the process, clarify the key factor effect of terminal voltage and reactive power control performance; the final form of the promotion of wind farm reactive voltage control performance and operation level of the measures, which is verified.

Power line inspection robots often work in the mountains, surrounded by trees and other background disturbances, while the field illumination changes significantly. The distance between the camera and the obstacles has a great influence on the scale changes of the obstacles in the image space. Therefore, accurate obstacle detection is very difficult. The deep learning methods such as Single Shot Multi-Box Detector (SSD) algorithm can be insensitive to illumination and scale changes in complex background with a much high accurate detection. But the amount of parameters of the SSD are so huge that make it very difficult to transplant to embedded systems. Aiming at the problem of stable and accurate detection of obstacles and easy to transplant to embedded systems, this paper proposes a method which significantly reduces the model parameters and improves the detection speed. In this method, the 23 convolution layers of the original SSD are simplified to 7 layers, and the Batch Normalization layer is added after each convolution layer to normalize the convolutional data. The result shows that the algorithm not only ensures the detection accuracy, but also greatly reduces the parameter quantities of the model and improves the detection speed significantly from 4.5 fps of original SSD to 15 fps of simplified SSD on Jetson TX2(an embedded system). Compared with some classical computer vision based detection algorithms, the method is more adaptable to complex environments.

由于光照变化、乘客拥挤和站外噪声干扰大等问题,现今地铁进站客流人脸检测技术精度较低.为提高人脸检测精度,本文在YOLO2轻量级目标检测算法Tiny YOLO2原有网络结构基础上,首先利用不同数目的1×1卷积层对特征图进行压缩,然后将特征图尺寸重新调整到统一大小进行级联,得到高维特征图.缩减网络最后几层卷积核数量,用1×1卷积层替换原始网络的3×3卷积层,得到更深而且更窄的人脸检测网络.改进后的网络先后在Wider Face数据集和地铁进站客流数据集上进行训练,得到最终的人脸检测模型.加载训练好的人脸检测模型对随机选取的300幅站外乘客图片进行测试.测试结果表明:本文算法相比Tiny YOLO2原始人脸检测算法,召回率提高4.2％,单幅图片检测速度提高6.5％.同时在广泛使用的人脸检测算法评测数据集FDDB上进行测试,在误检数目为200的情况下,人脸检测准确率相比Tiny YOLO2平均提高5％,比SSD检测算法提高2％,而且本文算法能够在检测速度和精度之间取得较好的平衡,有较好的泛化性.

Wind power short-term power prediction is an important index to evaluate the level of wind power operation, and it is also an important parameter to guide the safe operation of power system. Considering the singleness one-sidedness of the existing evaluation index, this paper aims to construct a comprehensive evaluation index system for regional wind power forecasting. The traditional single evaluation index is extended to the multiple evaluation system of power prediction. Then, this paper puts forward a comprehensive evaluation index of wind farm power prediction based on maximizing deviations and grey correlation analysis, which can eliminate the artificial factors to the weight distribution of multiple evaluation indexes to a large extent. Finally, the comprehensive evaluation method is applied to an application example. The results show that the comprehensive evaluation index not only can evaluate the regional wind power prediction scientifically and comprehensively, but also can guide the optimization direction of power prediction, and has good application value and promotion prospects.

This study analyses the distributed photovoltaic (PV) power system effects on common short-term load forecasting, based on the proposed power penetration index, and load data reconfiguration method, by using common short-term load-forecasting algorithm under conditions of the distributed PV power system and the power penetration index as 0, 5, 10, 15, 20, 24.46, 30, 35, 40, 50, 60 and 75%, the short-term load-forecasting accuracy obviously decreases. In this research study, the short-term power-forecasting method is proposed in order to offset the power output of distributed PV power systems, and transfer to the original load without the effects of distributed PV power systems or treated as the input information for updated load-forecasting model, this study proposes the short-term power-forecasting algorithm based on the BP neutral network for the typical distributed PV power system through training and validation of the proposed model. The power-forecasting accuracy is analysed and verified for a typical distributed PV power system, and the proposed method is effective in improving the accuracy for load forecasting.

More and more renewable power with large capacity has been incorporated into power system. The high fluctuation and randomness features of wind power present a number of challenges for the safety grid. ‘Grid friendly’ wind farm with storage device can reduce the volatility impact of grid safety. The optimal output strategy based on least-square polynomial fitting was established for the wind farm with the storage device. The results showed separate order correspond to separate smoothing effect. In the case of a 49.5MW wind farm, the third order polynomial model achieved the optimal output simulation of wind farm with storage device. In contrast to other numerical value filter methods, least-square polynomial fitting method has fast computation speed, and can calculate the slope of optimal output power in real time.

Accurate wind power forecast is an important method for solving the utilization problem of new energy. Forecast evaluation results have been applied to the dynamic dispatch of power systems that utilize large-scale wind power. As the starting point in optimizing regional forecast evaluation, this study first gathered fully diverse power forecast evaluation indexes that are based on a traditional index. Second, a comprehensive evaluation method for regional wind power forecast was proposed using principal component analysis and the information entropy calculation method. Finally, the proposed method was used to evaluate the regional wind power forecast. The case study revealed comprehensive evaluations and increased scientific weight allocation. Results confirmed the correctness and rationality of the proposed method, which can serve as a reference for power systems.

Wind power has been the most rapid developed new energy power form in recent years, its volatility, intermittent, and randomness have serious impact on the safe operation of power grid. Therefore the accurate prediction of the wind power is an important safeguard and reference to guide the new energy power system. Based on principal component analysis (PCA) and entropy method (EM), the paper expands the traditional single evaluation index and proposes a new comprehensive evaluation index. The results of the experiment show that this index is scientific and comprehensive, and can eliminate the human factor on index weight distribution.

In view of assembled MW-SCALE pitch doubly-fed wind turbine, due to its structure and more complex electronic control system, the original motivation of wind conditions through simulation, testing doubly-fed wind generator, wind speed converter in different operating conditions performance, and power curve. Testing machine and the fan controlling the normal communications, and complete the master's control tasks, inspection MW-SCALE doubly-fed wind turbine operation mode, test the converter control system and its compatibility. Test a variety of fault conditions in the wind turbine, converter basic protection. Through the wind turbine, gear box, converter, control system of joint test (except tower, blades), completed on the generator set design, manufacture, assembly quality inspection, while in the performance of the entire assessment test machine safety and protection of property, and problems found to be eliminated to protect the unit under test to factory quality standards.

The multistep prediction and information granulation prediction in the wind speed prediction were discussed. First, the least squares support vector machine was used for modeling method. Then, wavelet packet transform was used for data processing method. On the basis of wavelet decomposition, high frequency part was analyzed and the prediction accuracy was improved with the method. Last, the modeling method was used for multistep prediction and information granulation prediction. Many cases showed that multistep prediction is able to get the wind speed curve, which applies to regional energy scheduling including wind farms. Information granulation prediction can deal with redundant data and get more accurate characteristic value of wind speed, which applies to analyze the unit characteristics of different wind farms or different units.

Power quality problems could cause enormous amounts of economic cost to electricity consumers. Taking voltage dips as an example, this paper presents a general overview of methods to quantify economic impact of power quality problems. It also presents in detail several ways how to collect power quality economic data. At last, suggestions on how such researches could be done in China are represented.

To realize voltage sag source localization in distribution network, the paper proposes a function fitting method based on the least squares. Establish a voltage distance function in response to fault distance changes by the line voltage. According to the voltage distance function, combine with the bus voltage after fault to find out likely fault section and distance. Through the sorting algorithm to sort all possible results, weaken the effect of pseudo fault point on the judgment result. Finally the simulation verifies the effectiveness of the method.

It is well known that large-capacity wind power, as a type of strong fluctuations and random power, has an impact on grid safety. Due to this situation, accurate wind speed forecast plays an important role in reducing the impact of wind power on the grid. In this paper, we discuss the short-term wind speed forecast problem based on the wavelet packet transform and least squares support vector machine (LS-SVM). Firstly, high-frequency and low-frequency signals of wind speed are analyzed by the wavelet packet algorithm. Then, optimal wavelet packet transform is selected by minimum entropy principle. Based on these, short-term wind speed forecast model is established by LS-SVM. As an application of the proposed method, a case study with the actual data of a wind farm is presented to show the efficiency and accuracy compared with the previous results.

With the increasing number of wind farms in power systems, the scheduling of a single wind farm needs to be improved. For this end, this paper proposes an optimal short-term load dispatch strategy for a single wind farm. Firstly, considering the large number of wind units and the high dimensionality of the scheduling solutions, we analyze the unit load characteristics, from which we extract the unit load characteristic matrix, and then classify the wind power units with the FCM fuzzy clustering algorithm. Secondly, we define the running loss indicator and action loss indicator. Based on the prediction of wind power and the load instructions, we establish a unit commitment model in wind farm, and solve the model using a combination of the fuzzy clustering algorithm and genetic algorithm, which overcomes the difficulty of the high dimensionality of the solution in the wind farm scheduling problem, to obtain the optimal scheduling strategy. Finally, through the simulation of the scheduling strategy for a 45 MW wind farm, we demonstrate the feasibility and effectiveness of the proposed strategy.

Wind power industry developed rapidly in recent years. wind power is a type of power with randomness and fluctuation. Accurate wind speed forecasting can reduce the impact of wind power. Paper analyzed the wind speed signal with wavelet packet decomposition method from low-frequency and high-frequency, selected the optimal wavelet tree through the principle of minimum entropy. Short-term wind speed prediction model is built with support vector machine regression. This algorithm has advanced and better accuracy by comparing the results.

This paper presents a crack detection method for fibre reinforced composite beams based on the continuous wavelet transform. Finite element method simulations and experiments were carried out for flawed cantilever composite beams subjected to impact load at their free ends. The Gabor and Morlet wavelets were chosen for wave signal processing owing to their good time-frequency characteristics. Both wavelets perform well in the crack detection of composite beams. The results show that the crack location can be determined successfully, and the crack depth can be estimated from the crack-reflection ratio. Moreover, the Gabor wavelet can locate the cracks more accurately by virtue of its higher time resolution.

To eliminate limitation of the ID3 algorithm,an optimized algorithm for two-level information gain of attribute-value pairs is presented to establish the forecasting model of daily--characteristic-load decision tree.The algorithm has improved primitive ID3 algorithm in many aspects.It can prevent expansion biasing the attribute which has multi values.By this improved algorithm,the relationship of attributes can be considered well.Through setting threshold value sensitization of noise can be reduced.Daily characteristic load forecasting can be implemented by this model which associates day-forecasted information such as weather,week and so on.The analytic method of histogram is adopted to discretize the data of the load rate-of-change and the data of weather combined hierarchical clustering and discretization based on entropy;after the data is pre-processed,the forecasting model of load decision tree is established by the optimized algorithm for two-level information gain of attribute-value pairs and the characteristic load can be forecasted by giving the information of date-forecasted weather and week.The forecasting results meet even exceed the requirements of utility and demonstrate high-accuracy of the proposed model.If use the 24 or 96 load and its corresponding influent factors to train,24 or 96 forecasting models will be formed.Then 24 or 96 load can be forecasted by these models.