Search

The study of the uncertainty of wind power is very important for power system planning and operation decision. This paper presents a method of wind power probability interval prediction based on Bootstrap quantile regression method. Due to the large difference in wind power data, the method first divides the power at equal intervals. And then use the Bootstrap method to extract the data in different power intervals to form the pseudo samples. Then use the quantile method to obtain the upper and lower limits of the confidence interval of each power interval. Finally, the upper and lower limits of the confidence intervals of multiple power intervals are combined by using the quantile regression method to obtain the probabilistic prediction interval of wind power. Let the interval coverage and average bandwidth as evaluation index. Let the interval coverage rate and average bandwidth as evaluation index. Through the Bootstrap quantile regression method, the interval coverage rate of wind power probability interval prediction is higher, the average bandwidth is narrower, the precision is higher and the effect is better.

The knowledge representation based on fuzzy Petri nets included library, transition, input, output, input confidence, output confidence and so on. Since the fuzzy knowledge had been preferably expressed and processed by model of fuzzy Petri nets based on fuzzy production rules, a fuzzy Petri nets model of hydraulic variable pitch system for wind turbine generator was established. In this paper, the fuzzy Petri net was represented in the form of matrix and a parallel reasoning algorithm was applied in fault diagnosis. The algorithm had analyzed many factors, such as weights, thresholds of transitions, reliability of rules. Beside this, according to the algorithm, the trigger condition of transitions became more strictly. It avoided unnecessary repetitions of transitions trigger effectively, and reduced the complexity of the algorithm. At last, the algorithm was compiled on the platform of Matlab with its powerful capabilities of matrix operations and data analysis. The result had shown that it had the ability of parallel reasoning for complex system and that it is highefficiency and is easy to implement.

Intermittency and uncertainty pose great challenges to the large-scale integration of wind power, so research on the probabilistic interval forecasting of wind power is becoming more and more important for power system planning and operation. In this paper, a Naive Bayesian wind power prediction interval model, combining rough set (RS) theory and particle swarm optimization (PSO), is proposed to further improve wind power prediction performance. First, in the designed prediction interval model, the input variables are identified based on attribute significance using rough set theory. Next, the Naive Bayesian Classifier (NBC) is established to obtain the prediction power class. Finally, the upper and lower output weights of NBC are optimized segmentally by PSO, and are used to calculate the upper and lower bounds of the optimal prediction intervals. The superiority of the proposed approach is demonstrated by comparison with a Naive Bayesian model with fixed output weight, and a rough set-Naive Bayesian model with fixed output weight. It is shown that the proposed rough set-Naive Bayesian-particle swarm optimization method has higher coverage of the probabilistic prediction intervals and a narrower average bandwidth under different confidence levels.

In order to reduce the fluctuation rate of wind power output and improve the reliability when connected to grid, this paper puts forwards a control method of smoothing wind power output. This method is used to smooth wind power fluctuation with battery energy storage system(BESS) based on Empirical Mode Decomposition (EMD). After wind power is decomposed by EMD, the low-frequency parts are used as wind power grid-connected value and the high-frequency parts are stored by BESS; The orders of EMD filter can be adaptively adjusted according to the state of charge (SOC). Compared with another method of constant order with EMD filter, simulation results show that the method proposed in this paper can avoid the over-charging and over-discharging of battery energy storage system so as to keep SOC stable.

Reliability is an important issue in large-scale photovoltaic (PV) systems as their operations rely on business plans developed over periods of time of at least twenty years which often assume fault-free functioning. This paper presents a Bayesian network (BN) reliability model combined with Fault tree (FT) for assessing the reliability of PV systems. A fault tree was built based on electrical architecture diagram of a generic PV system firstly, and then it was mapped into a Bayesian network. The conditional probability table is derived by employing the bucket elimination according to the fault probability of each component for reliability assessment of the system. The paper shows that the basic inference techniques of BN may be used to obtain classical reliability parameters (i.e. reliability of the top event or any sub-system) and identify the weak components of PV system. Moreover, the framework based on the BN formalism is intuitive and easy to be applied into the reliability assessment of PV system, and it provides a basis for more advanced and useful analyses in fault diagnosis of systems.

An on-line fuzzy neural network control (FNNC) with decoupling unit is applied for ball mill system of blast furnace. The structure of FNNC is composed of premise and consequence part. Premise parameters reduce fuzzy rules to solve the combination explosion of multi-inputs. Consequence parameters achieve a good transient response. An adaptive step size equation created by a Lyapunov function guarantees convergence of algorithm. Neural network decoupler based on timedomain response curve is used to counter the interaction among channels. Simulation results prove effectiveness of proposed control system.

By using neural networks and multivariable decoupling control theory, the idea of model reference inverse dynamic control was applied in the boiler-turbine system. Four neural networks are used as an approximating model of the unknown nonlinear dynamic behavior of the boiler-turbine while another neural networks are decoupling and controller. Neural network controllers are trained by using dynamic backpropagation algorithm. Simulation shows the developed design method for the boiler-turbine control system can simultaneously control load megawatt and throttle pressure and has good adaptive ability

Photovoltaic (PV) power forecasting is one of the effective ways to decrease the influence caused by large-scale PV power station connecting to the grid. Currently, there are various forecasting methods. However, single forecasting method tends to perform unsatisfactorily. In this paper, a combination method in PV power forecasting based on the correlation coefficient was proposed. Firstly, the persistence method, support vector machine (SVM) method and prediction method based on similar data were used to forecast the PV power separately. Then the weight of single prediction method was determined by the correlation coefficient of prediction and actual values. Single prediction method with larger correlation coefficient has greater weight. Finally, a combination model based on the correlation coefficient was established. By means of simulation and analysis, the combination prediction method based on the correlation coefficient was proved to be effective and the prediction accuracy was improved.

The wind speed prediction method based on improved neural network was proposed in this paper by analyzing the relationships between wind direction and wind speed changes. A new network model was created by customizing the network object properties. Wind speed and wind direction were regarded as the network input at the same time using a single network layer to deal with trends in wind direction so as to realize wind speed prediction. Three different kinds of improved network models were provided in this paper, the validity of which was verified by the actual data in wind farm. Simulation results show that: It can significantly improve the network prediction accuracy by continuously improving the structure of neural networks.

Timely and accurate assessment of wind turbine operating condition can save the maintenance cost effectively, because equipment repair and replacement for large wind turbine are expensive. Therefore, an efficient operating condition assessment strategy for wind turbine is proposed based on fuzzy synthesis method. First, the traditional fuzzy assessment process is introduced briefly a foundation for comparison. Second, a wind turbine assessment model is established and assessed in detail, with an improved fuzzy prediction strategy and de-fuzzy strategy. Finally, a comparison between the improved strategy and the traditional strategy is studied based on wind turbine operation data, and the results validate that the strategy proposed in this paper can alarm operation fault more promptly and efficiently, which is better for wind turbine operation safety and maintenance.