Modelling the uncertainty evaluation loss by mountain fire under power system for computer control based on ontology theory

Abstract

This paper proposes a method for assessing the loss uncertainty of power system under complex conditions based on ontology. First, the mountain fire model and uncertainty analysis are carried out to estimate the influence of the mountain fire on the target area. We design the distributed rules of the power system elements of the network. It uses of weighting factors to get different important degree of elements in the power system. The theory of ontology is introduced to evaluate the loss uncertainty of grid system. Then the most sensitive factor for the loss will be determined. The simulation verify the performance of loss uncertainty of the power system, which demonstrate that it has improved by 34.5% accuracy and better effectiveness.

Keywords

Uncertainty power system ontology mountain forest fire

1. Introduction

Power systems are complex systems that interconnect multiple types of engineering structures and power facilities. Due to geographical isolation, the power system elements in the space to show isolated coverage [1]. Power system elements are part of geospatial, depended on the existence of geographical space. So it will be affected by other factors in the geospatial, and there are serious uncertainties, as Fig. 1 shown.

Mountain forest fire is the geographical space time, space random sudden events. It will damage the elements of power system attached the geographical space, which is difficult to estimate the loss. However, such losses are local, temporary and self-healing for strong power grids [2]. The development and change of the mountain fire will directly affect the environmental factors, such as temperature and humidity in the geographical space.

Figure 1.

Mountain forest fire effect on the power of complex space to uncertainties

Therefore, the researches of mountain fire on the power grid system loss, are important means to uncertain environment for effective planning and layout. In the study of the influence of the development of the mountain fire, it is an important means to analyze the situation of the fire [4]. There was a negative correlation between the number of hot spots and the amount of precipitation, and the proportion of the fire in the field was proportional. And seasonal factors have become an important factor in the occurrence of mountain fires and the normal operation of the grid. AC transmission lines in the mountain fire conditions are easy to breakdown. As the mountain fire caused by electrical equipment failure pathways and mechanisms are different, the mountain fire information and geographical environment and meteorological information combined [5]. It is possible to effectively use the information of the law and the high knowledge intensity to assist in predicting the content with high uncertainty. In the advanced sensing equipment for the effective monitoring of mountain fires, method based on the BP neural network transmission line mountain fire hazard point recognition, determined the transmission line mountain fire disaster factors.

Research on Wireless Sensing and RBF Recognition of Mountain Fire [6], proposed result of a sudden drop in the level of insulation caused by flame temperature, flame conductivity, and ashes and smoke. The use of intelligent wireless sensor network can effectively overcome the serious problems of monitoring areas are not concentrated. In the power system for the development of early warning measures, the use of extreme learning machine to predict the method of mountain fire [7], use statistical fire data to carry out training and learning, transmission line mountain fire emergency plan and its system. Above all, there are some problems the loss uncertainty of power system under complex conditions, such as lack of coverage and Elements redundancy.

Therefore, this paper proposes a method for assessing the loss uncertainty of power system under complex conditions based on ontology. First, the mountain fire model and uncertainty analysis are carried out to estimate the influence of the mountain fire on the target area. We design the distributed rules of the power system elements of the network. It uses of weighting factors to get different important degree of elements in the power system. The theory of ontology is introduced to evaluate the loss uncertainty of grid system. Then the most sensitive factor for the loss will be determined. The simulation verify the performance of loss uncertainty of the power system.

2. Mountain fire model and uncertainty analysis

2.1 Probability modeling of fire hazard in power grid

Mountain fires generally need to be controlled by trees and other flammable materials to form a long enough disaster. Therefore, this paper will refer to the forest fire risk level to build a mountain fire probability model. The division of the forest fire rating takes into account various factors related to the occurrence of mountain fires, including seasonal, climatic conditions, weather changes, geographical latitudes, terrain, fire sources and other factors. Using the Braun Davis Fire Barrier Method, it constructs the IRHAJ (Initial Risk Hidden Area Judgment). Its fire risk can be expressed,

$\displaystyle U(V,T,F,M)=f(V)+f(T)+f(F)+f(M)$ (1)

In which, $U$ is fire risk indicators, $V$ is wind speed, $T$ is temperature, $F$ is combustible moisture, $M$ is precipitation interval days. In the above factors, Geographical latitude, terrain does not change by $T$ . Fire condition changes little with $T$ . The above factors can be transformed into a dependent variable, with geographic coordinates as an independent variable. We have $V=V(x,y)$ , $T=T(x,y)$ , $F=F(x,y)$ and $M=M(x,y)$ . The fire hazard is normalized, we can get a single risk $D$ . Here is a comprehensive factor,

$\displaystyle K=\sqrt[2]{{f(V)}^{2}+{f(T)}^{2}+{f(F)}^{2}+{f(M)}^{2}}$ (2) $\displaystyle D=\frac{1}{K}\left(\frac{{\Delta}^{2}f(V)}{\Delta x\Delta y}+% \frac{{\Delta}^{2}f(T)}{\Delta x\Delta y}+\frac{{\Delta}^{2}f(F)}{\Delta x% \Delta y}+\frac{{\Delta}^{2}f(M)}{\Delta x\Delta y}\right)$ (3)

At this time, the probability of occurrence of mountain fire under different fire rating conditions is

$\displaystyle p(f_{D})=\frac{N(f_{D})}{N(D)}$ (4)

And, $N(f_{D})$ is the number of occurrences of mountain fire under the condition of single fire risk. $N(D)$ is the number of times of a single fire hazard rating

2.2 Method and mechanism in power grid malfunction by mountain fire

Power system is a relatively stable structure, and under the influence of different single fire risk meteorological level. Occurrence of mountain fires and grid faults will be a conditional probability event [8]. One of the transition factors will be the air gap is broken and the insulator damage. The cause of grid failure is due to flame temperature, air breakdown factor and solid particles and other factors.

2.2.1 Characteristics of mountain fire spread and temperature variation

The speed of mountain fires is an important factor that affects the characteristics of air gap around the grid and the solid particles. The velocity of fire is mainly related to the pattern of combustible configuration, initial spreading speed, slope and its relationship with wind direction, wind speed and so on. It can be expressed as

$\displaystyle R=R_{0}K_{s}K_{w}K_{\varphi}$ (5)

$R_{0}$ is initial spreading speed, which can be obtained through meteorological elements. $K_{s}$ is combustible configuration pattern correction factor. $K_{w}$ is wind speed correction factor, $K_{\varphi}$ is terrain correction factor. According to Byram and Wiegel’s empirical formula, the relative temperature can be further solved as follows,

$\displaystyle\Delta T=29.54{(\textit{QwR})}^{0.17}$ (6)

$Q$ is combustible heat, $w$ is the fuel weight per unit area.

2.2.2 Air breakdown factor

With the spread of mountains, and gradually change the surrounding environment of the power system, resulting in air insulation performance. Fire conditions under the air gap breakdown voltage $U$ is

$\displaystyle U=K_{t}K_{p}U^{0}$ (7)

$K_{p}$ is the smoke concentration correction factor. The relationship between the two can be fitted by the hyperbolic equation,

$\displaystyle K_{p}=\frac{1}{14R+1}$ (8)

$K_{t}$ is air correction factor, Which is the product of the air density correction coefficient and the air humidity correction coefficient.

3. Rule of power system element network

From the geographical distribution characteristics, when the mountain fires occur, the spatial characteristics of the power system elements can be equivalent to the original geospatial distribution superimposed on the characteristics of the mountain fire spread. As the power system in the various types of equipment in addition to transmission lines are concentrated in the power plant and substation, plant equipment and a variety of transmission lines connected to each other constitutes a power system network topology [9]. The topology model can be represented by a set of nodes and edges $G$ ,

$\displaystyle G=(V(s),E(s))$ (9)

Where $V$ is the set of nodes, $E$ is the edge set, and $s$ is the edge opening and closing state. This further defines the state of the grid failure due to mountain fires, as Fig. 2 shown.

Figure 2.

Distribution of power system elements.

When the node of the power system is destroyed by the mountain fire, there is $s=$ 0, then $V(s)=\infty$ , when the node device is not damaged, there are $s=$ 1, $V(s)=+\infty$ . When the transmission line was destroyed by the mountain fire, there are $s=$ 1.

The overall on-off state of the power system can be defined as a matrix $A$ , we have $A=\{a_{ij}\}$ , in which,

$\displaystyle a_{ij}=\left\{\begin{array}[]{cl}1&i=j\\ 0&i\neq j,i\notin j\\ s&i\neq j,i\in j\end{array}\right.$ (10)

The connectivity between the midpoint and the point in the topology can be passed by edge. The influence of mountain fire on the loss of power system is mainly through finding out the connectivity of any two points in the topology. It is possible to transmit the closure array as an effective decision content of the power system topology,

$\displaystyle A_{c}=\int^{n-1}_{i=0}{A^{l}}=I\oplus A\oplus A^{2}\oplus\dots% \oplus A^{n-1}$ (11)

For a node topology, as long as the calculation of the correlation matrix $A^{n-1}$ power array can be obtained within the topology of the connectivity of the nodes to achieve the topology of the network analysis.

4. Uncertainty evaluation of power system based on ontology of mountain fire

The stability rule of the power system affected by the mountain fire needs to be further determined. In the state of mountain fire, the power system will be affected by temperature, climate, terrain and other factors. It is difficult to cover a single factor in the overall development trend effectively. With different dimensions and depth, it needs to dig out the core elements, in order to deconstruct the most appropriate variables effectively.

Therefore, this paper introduces the ontology, which is mainly used to describe the nature of the object. It is the objective world elements to describe the conceptual model, to achieve the elements of the cluster and understanding and communication. It can be abstractly abstract and distinguish the complex and heterogeneous elements of the mountain fire on the loss of the power system. It intricate attributes and operations.

Set the mountain fire on the power system loss of the body structure is a six set. We have make $U$ , $V$ , $F$ , $T$ , $M$ and $G$ considering together.

Its composition structure becomes,

$\displaystyle O=\{\mathcal{C},\mathcal{R},\mathcal{H},\textit{Rel},\mathcal{A}% ,\mathbb{A}\}$ (12)

Where $\mathcal{C}$ is the concept of elements for the uncertainty assessment, $\mathcal{R}$ is the relational operator between the uncertainty factors, $\mathcal{H}$ is the conceptual hierarchy, clustering relationship between the uncertainty factors, and Rel is the non-clustering relationship between the uncertainty factors, $\mathcal{A}$ is the ontology, and $\mathbb{A}$ is the intrinsic attribute of the element.

The uncertainty of loss $Y$ can be expressed as $H(Y)=I(XY)+H(Y|X)$ , which is the random distribution of the influencing factors, and has the exact boundary of the loss degree. In addition, the factors that make up $X$ are multi-source and heterogeneous, and there are some factors of gray scale. Therefore, the uncertainty of the analysis of the degree of loss is mainly to determine the controllable factors and gray factors $X$ , and determine its impact on the degree of loss of uncertainty, reduce the dimension of the estimated analysis.

Assuming that there are $n$ kinds of loss degree influencing factors, each factor has $m$ loss degree evaluation value, then the $i$ voice quality influencing factors m indicators, there are

$\displaystyle H(Y|X)=\left\{\left[\begin{array}[]{ccc}x_{0,0}&\cdots&x_{m-1,0}% \\ \vdots&\ddots&\vdots\\ x_{0,n-1}&\cdots&x_{m-1,n-1}\end{array}\right]\right.$ (13)

$H(Y|X)$ establishes a model to evaluate the loss uncertainty, effectively analyzes the direct impact of the uncertainty factors on the uncertainty of the loss, but does not take into account the loss of the interaction between the factors Certainty, resulting in error assessment of the error. Therefore, based on the uncertainty entropy, this paper analyzes the gray-scale correlation of each factor, establishes the optimization model, and determines its reference value.

If the uncertainty of the factors $x_{n,m}$ changes in the trend (mountain fire spread, the size of the spread of the mountain fire, mountain fire spread, etc.) are basically the same, then the greater the degree of correlation between them. On the contrary, the degree of correlation is small. Through the gray correlation analysis, effectively reduce the interaction between the uncertain factors. Design correlation coefficient

$\displaystyle{\xi}_{i,j}=\frac{\min_{i}\min_{j}||x_{ij}-x_{j}||+\rho\min_{i}% \min_{j}||x_{ij}-x_{j}||}{|x_{ij}-x_{j}|+\rho\max_{i}\max_{j}||x_{ij}-x_{j}||}$ (14)

${\min}_{i}\min_{j}||x_{ij}-x_{j}||$ is the smallest difference between the factors, $\max_{i}\max_{j}||x_{ij}-x_{j}||$ is the biggest difference between the factors. $\rho$ is the resolution coefficient. The logarithmic form of the likelihood function of $\rho$ can be expressed as $l(\rho)=\textit{log}(E({\xi}_{i,j}))$ .

5. Simulation and testing

The geographical remote sensing data of a province in southern China and the distribution of power system factors in the province are taken as the input conditions of simulation. The parameters for the simulation settings are shown in the Table 1.

Table 1
Simulation conditions for the impact of mountain fires on power systems

Factor	Parameter	Factor	Parameter
Combustible	Grass, trees	Calorific value	18.5 kJ/g
Initial spreading speed	1.58 km/h	average wind speed	10.8 km/h
Power element	6	Power connection section	16

It has the necessary lines, including 8-segment 500 kV and 8-segment 220 kV transmission lines. In this experiment, 20 data sampling points will be set up and Monte Carlo analysis of the data obtained from the sampling points will be made. In order to verify the accuracy and efficiency of the new algorithm in the evaluation of the loss of the power system, the main indicators include the correct rate of loss assessment and the time consumption [10].

5.1 Analysis of the accuracy of loss assessment

The correct rate of loss assessment refers to the ratio of the correct number of estimates to the estimated number of losses when the mountain fire spreads. In the construction of simulation conditions, set in a fire accident, the number of loss assessment of 100 times, assuming that the number of indicators can affect the 1 to 200, limited assessment time of 10 s, 30 s. Through the data acquisition, you can get the correct rate of loss assessment shown in Fig. 3.

Figure 3.

Analysis of the accuracy of loss assessment.

As shown in Fig. 3, with the increase of the observed observation index, the correctness of the loss uncertainty assessment is gradually decreased. This is due to the increase in the number of indicators, resulting in increased complexity of the assessment. With the increase of the evaluation time, the accuracy of the different loss uncertainty assessment algorithms has increased. When the number of observed indicators is 120 and the evaluation time is 30 s, it can be seen from the figure that the correct rate of the loss assessment of power system based on ontology is 66.6%, and based on the accuracy of the vector machine evaluation method to 32.1%, to improve the correct rate of about 34.5%.

5.2 Analysis of time efficiency of loss assessment

Loss Assessment Time efficiency refers to the time required for a single loss assessment. Here, the number of evaluation indicators is set to random, and the two evaluation methods will be carried out under the same conditions. Assuming the number of indicators can be affected by 1 to 200, through the data collection, you can get the loss assessment time efficiency as shown in Fig. 4.

Figure 4.

Time assessment of loss assessment time.

As shown in Fig. 4, the time efficiency of the assessment of the loss uncertainty of the power system is higher than that of the vector machine based on the number of observations. This is due to the introduction of the ontology method, effectively through the clustering compression evaluation of the influencing factors, so that the assessment process can be less affected factors, you can get the assessment results, in order to obtain faster assessment of performance.

6. Conclusion

Footnotes

Acknowledgments

Project supported by major scientific and technological project of State Grid: Study on the transmission line wildfire wide area monitoring system based on meteorological synchronous satellite (5216A015001M).

References

Morales

and Pineda

, On the inefficiency of the merit order in forward electricity markets with uncertain supply, European Journal of Operational Research261(2) (2017), 789–799.

Wang

Liao

and Liu

, Possibility-Based ELECTRE II Method with Uncertain Linguistic Fuzzy Variables, International Journal of Pattern Recognition and Artificial Intelligence31(9) (2017), 1759016.

Deng

et al., The uncertainty entropy of low-rate speech quality evaluation and the analyses of the gray correlation, IEICE Electronics Express12(3) (2015), 20141019.

Lacerda

and Seiler

, Stability of uncertain systems using Lyapunov functions with non-monotonic terms, Automatica82 (2017), 187–193.

Cavalagli

Gusella

and Severini

, The safety of masonry arches with uncertain geometry, Computers and Structures188 (2017), 17–31.

and Zhu

, Parametric optimal control for uncertain linear quadratic models, Applied Soft Computing56 (2017), 543–550.

Shahidehpour

and Fotuhi-Friuzabad

, Grid modernization for enhancing the resilience, reliability, economics, sustainability, and security of electricity grid in an uncertain environment, Scientia Iranica23(6) (2016), 2862–2873.

Khalili-Damghani

and Shahmir

, Uncertain network data envelopment analysis with undesirable outputs to evaluate the efficiency of electricity power production and distribution processes, Computers and Industrial Engineering88 (2015), 131–150.

Lin

and Deng

, Dimensional functional differential convergence for Cramer-Rao lower bound, Journal of Difference Equations and Applications23(1–2) (2017), 249–257.

10.

Bassamzadeh

and Ghanem

, Robust scheduling of smart appliances with uncertain electricity prices in a heterogeneous population, Energy and Buildings84 (2014), 537–547.