The Analysis Of Transmission Power Influence Factors Based On High Dimensional Data Nonparametric Additive Model

With the cross of computer science, power technology, social management and economic and financial development and other disciplines extends ,, data information Starting from power system operation control of the production level is starting to coordinate control, multi-level joint management-level development. In the electric power system operation in order to analyze and control the number of operating rules and system state power system, it is necessary to collect relevant data for research. With the improvement of measurement tools and monitoring systems, coupled with the complexity of the grid system, so often to collect data not only huge amount of data, and in particular the high dimension. For these high-dimensional data, and how to find the real impact of important variables and how to show an explicit function expression between variables, and ultimately achieve the purpose of mining useful information and provide regulatory strategy for the control of such person behind the data. The current regulation of some aspects of the power system or in the subjective judgment of experience to carry out such adjustment is not strong way not only reliability, timeliness and low inadequate and prone to accidents. If we can find effective ways to solve these problems, it will actually run on the actual operation of the power system play a guiding role, so that the actual operation of the power system becomes more convenient, more efficient and more reliable, which not only electricity systems bring huge economic benefits, even for some of the traditional theory of operation of the power system will learn. The following section of this paper in order to limit the transmission power of the dependent variable (TTC), for example, can be added by non-parametric modeling approach, in a sectional limit may affect the transmission power of high-dimensional variable dimensionality reduction at the same time, and list the dependent variable transmission section limits specific function expression of power and ultimately self-selected variables.The dependent variable section limits the transmission power studied in this paper is a very important factor in the reality of the power system, power system in the ordinary course of running to ensure that the section limits the transmission power within a certain range, otherwise there is a danger hazard even direct accident. When the power supply system, power is transmitted over time period in constant change, and sometimes the transmission power fluctuation is more severe, such as:the peak or trough is likely to result in the transmission of power instability and even accidents so timely and effective regulation of the transmission power is stable and very necessary. Current on which part of the transmission power needs to be adjusted to adjust from this node nearest conditioning equipment, but in fact when the adjustment means a node, not only affect the transmission power of the nearest node to other nearby transmission power will not need to adjust when there is a certain influence, there may be from the nearest node transmit power regulating device at this node transmission power is not the biggest, adjust the principle of proximity may appear chaotic situation directly, especially for more complex circuitry situation, with this adjustment method is more prone to blind adjustment, the weight it may cause an accident. In summary it can be seen that the current regulation mode the biggest problem is not clear at each transmission power which mainly affect the voltage, load and line active and specific variables affect the way by.Through the introduction of the several methods, principal component analysis is an unsupervised learning method, stepwise regression and Lasso method is described by linear relationship between independent variables and the dependent variable, and the problem that this article is to study the first supervised learning, is not a simple linear relationship between the second variable, but non-linear relationship, so to solve the problem of this article, the above method is still some drawbacks.For lack of purpose of this study as well as the above-described conventional method, this paper presents a nonparametric additive model, the method not only make up the above conventional methods, but also a good solution to the practical problems to be studied in this paper. First, non-parametric methods herein can be added to the model using non-parametric independent variables respectively screening methods and Group-Lasso two-step variable selection method, not only relatively easy choice truly influential independent variable, and with samples or dimensions increase, increasing the amount of computation of this method is much less than the above-mentioned data mining; in the first step followed by non-parametric independent variables to make the screening process for each variable B-spline basis expansion, fully consider the independent variable and dependent nonlinear relationship between variables, and nonparametric methods, it is also highly adaptable and flexible nature; and finally the establishment of non-parametric model can be applied in the second step, in a further variable selection at the same time, due to the good grasp the functional relationship between variables, so the effect is also very good fit. By 99 samples,53 examples of variable analysis results, the proposed method to select the number of variables 5 and 12 and Lasso than stepwise regression methods were less, but the effect is obvious fit than stepwise regression method and Lasso good.Nonparametric additive model proposed solution to the problem a higher dimension, but also clarified the limits of the transmission power dependent variable section which mainly affect the voltage, load and line active and specific variables affect the way by. Based on section factors limit the transmission power of the power system analysis, and ultimately find a specific function expressions affect the dependent variable, and hope that these conclusions can provide a more intuitive operation of the auxiliary control policy for the dispatcher, and then to improve the power grid scheduling behavior the efficiency and reliability of certain significance.