Failure probability estimation of overhead transmission lines considering the spatial and temporal variation in severe weather
- 1.3k Downloads
Estimation for weather-related failure probability of overhead transmission lines is essential in the reliability assessment of a power system. This paper analyzes the outage and weather data of 110 kV overhead transmission lines in the Guangxi Zhuang Autonomous Region of China during 2011–2014. The result reveals obvious uneven distributions of outage events for time and space due to the spatial and temporal variation of severe weather. Based on the results, an estimation method is proposed in this paper. Split and aggregation is used to smooth the outage and weather data. The poisson model is adopted in our method to investigate the statistic characteristics of transmission line outage events. Regression analysis is applied to obtain the correlation between the weather intensity and history failure rate. Furthermore the method proposed is validated against the empirical outage data.
KeywordsSevere weather Transmission line Poisson model Failure probability Regression analysis
Overhead transmission lines are exposed to nature and are vulnerable to severe weather, such as lighting, storms and hurricanes and so on [1, 2, 3, 4]. Operational security of overhead transmission lines has a close relationship with the external meteorological conditions. Weather factors need to be taken into consideration for a reliability assessment of overhead transmission lines, which is important for economic- security decision making [5, 6, 7]. Failure probability of overhead transmission lines is essential in applying an appropriate reliability assessment method.
The Markov model  is widely used to estimate the failure rate of overhead transmission lines. Considering the intensity of the weather factors, the Markov model is being developed as a two-state, three-state, and multi-state weather model and so on [9, 10, 11]. Chong Wang presents a Markov-based deterioration model for transmission components . Ali Arab et al. use partially observable Markov decision processes to develop a generic model for integrated condition-based maintenance of the electric power systems infrastructure . Yong Liu and Chanan Singh introduced a DC-OPF based Markov cut-set method to evaluate composite power system reliability considering weather effects . These models are primarily suitable for relatively long-term applications. There are both temporal and spatial variations of weather factors [15, 16]. Estimation of the weather influence on transmission lines is required to capture the stochastic behavior of the weather .
The Monte Carlo method is applied to simulate the stochastic behavior after the modeling of complex system operational conditions . Billinton and Wenyuan  used the two-state weather model and the Monte Carlo sampling technique to incorporate the variable weather conditions in a composite analysis. Mathaios Panteli and Pierluigi Mancarella introduced a novel sequential Monte-Carlo-based time-series simulation model to assess power system resilience . The Monte Carlo simulation depends on random experiments and its convergence may require acceleration by using other techniques .
In order to establish a simple relationship between failure probability and weather factors, statistical analysis and the artificial intelligence method are also adopted . For example, the noisy OR-gate model, regression model and fuzzy model are proposed in many literatures [22, 23, 24, 25]. In these methods, the average failure rate model is still used after data splitting and aggregation. Weather can affect all the transmission lines in a certain area. The outage event in this area has a random nature and statistical characteristic, which is ignored in the average failure rate model. The poisson model is proposed in  to consider the statistical characteristics of transmission line outage events. But the model is not fully described and not validated against empirical outage data.
In order to determine the distribution of outage events on time and space, outage and weather data of an 110 kV overhead transmission line from the Guangxi Zhuang Autonomous Region of China during 2011–2014 are analyzed in this paper. A failure probability estimation method for overhead transmission lines is proposed in this paper considering lightning and wind. Statistical characteristics of an outage event are investigated based on the poisson model. The relationship between the poisson model parameter and history failure rate is presented. Furthermore the model is validated against the empirical outage data.
2 Outage events of transmission lines in Guangxi Zhuang Autonomous Region
3 Estimation for transmission line failure probability
3.1 Temporal and spatial split of weather and outage data
According to the weather and outage data of transmission lines, appropriate weather factors are determined in the method and denoted as w1, w2, …, wp respectively. Due to the temporal and spatial variations of weather factors, temporal and spatial splits are needed. Distribution of weather factors is uneven over a vast area, such as a province. A transmission line might be in multiple areas with a different meteorological environment. A spatial split is the division of weather areas. Assuming that the meteorological environment in each weather area is consistent, the divided weather areas are denoted as WA1, WA2, …,WAq respectively. According to the data from the meteorological department, the weather area is divided the same as the administrative area. There are several lines in one weather area, which are in the same meteorological environment. All transmission lines with the same voltage level are taken as the basic samples for the estimation of one weather area. Temporal treatment of weather data contains both split and aggregation. First weather and outage data are split by a step of one day and the weather data is classified into several intensities. Based on the intensity classification, the weather and outage data with the same intensity are aggregated. Both temporal and spatial aggregation result in data smoothing and provides more meaningful patterns and it converts the time distribution of the weather factors to the intensity distribution. There are several weather factors in one area and the combination of different weather factors is defined as a weather block. As the weather factor w1 and w2 in a weather area is at level k and m respectively, the weather block is defined as WBkm. The outage data is divided according to the classification of the weather block.
3.2 Probability distribution of a transmission line failure based on poisson model
Historical outage data of transmission lines can be used to calculate the parameter of the poisson model. According to historical weather and outage data, the relationship between the historical failure rate and weather factors can be established. Based on the poisson model, the failure probability of the transmission line can be predicted using the weather forecast data.
3.3 Regression analysis
The least square method is used to obtain the regression coefficient matrix. Furthermore, the significance test and applicability test are adopted to test the regression equation. Then the regression equation can be used to estimate.
3.4 Failure probability of transmission lines
Determine main weather factors by analyzing the historical outage and weather data. Weather and outage data is classified according to the division of the weather area.
Classify the weather intensity and establish the weather block.
Count the outage number in different weather blocks and calculate the historical failure rate of the transmission line.
Establish the regression model of the historical failure rate and weather intensity and select the suitable regression model via the hypothesis test.
Obtain the poisson model parameter based on the historical outage and weather data.
Calculate the failure probability of the transmission line with the weather forecast data.
4.1 Temporal and spatial split of historical data
According to the weather and outage data of the 110 kV transmission line in the Guangxi Zhuang Autonomous Region of China during 2012–2014, two parameters denoted as w1 and w2 are chosen as the lightning density and maximum wind speed per day. The lightning density data is obtained from LLS  and the wind speed data is the instantaneous value obtained on the ground from the meteorological service of Guangxi. Taking Fangchenggang city with the most outages for example, the weather area of the whole city is named as WA0. Four counties in its jurisdiction, Gangkou, Fangcheng, Shangsi and Dongxin, are named as WA1, WA2, WA3 and WA4 respectively. Gangkou and Dongxin are next to the sea, and Shangsi is farthest from the seaside.
4.2 Historical failure rate of the transmission line
According to the third step of the calculation process, the outage numbers at different weather blocks are counted. The historical failure rate of the transmission line is calculated. The data during January 2012 to July 2014 is taken to establish the estimation model and the data during August 2014 to December 2014 is taken to verify the model.
4.3 Estimation and verification of model
According to the above formulas and (5), the poisson model parameter is calculated. Using (10), the estimation model for failure probability of each weather area is finally obtained. Putting the forecast weather data into the estimation model, the failure probability of the transmission line can be predicted.
There is only one peak value more than 0.2 on the two curves of the predicted failure probability, which is during the typhoon “seagull.” In the curve of line X1, most of the predicted failure probability is less than 0.05 and the proportion is about 97.4%. There was just one outage on September 3, 2014 with the predicted failure probability less than 0.1. The outage reason recorded is lightning. But the weather data shows very low lightning activity density on this day. There are two days during typhoon “seagull” with one outage and the predicted failure probability is more than 0.2. In the curve of line X2, 96.1% of the predicted failure probability is less than 0.05. There is no outage with the predicted failure probability less than 0.1. There are two days during typhoon “seagull” with two outages and the predicted failure probability is more than 0.3. In general, the calculation model can reliably predict the outage probability of transmission lines. 0.2 can be set as a pre-warning value of outage probability for the power grid company.
Weather-related outage accounts for 78.09% according to the outage data analysis of the 110 kV transmission line in the Guangxi Zhuang Autonomous Region of China during 2012–2014. The distribution of weather-related outages for time and space is consistent with the temporal and spatial variation of weather factors.
An estimation method for failure probability of transmission lines is proposed in this paper. It splits and aggregates the historical weather and outage data with an interval of one day for each weather area. The weather data is converted to weather intensity and aggregated as a weather block. The statistical characteristics of the outage event in each weather block are investigated based on the poisson model. The relationship between the poisson model parameter and weather data is analyzed via regression analysis.
The example shows that the model can reliably predict the failure probability of transmission lines. It can effectively reflect the influence of a typhoon. According to the weather forecast data, the power grid company could predict the failure probability for each weather area using this method. 0.2 can be set as a pre-warning value of the outage probability. Then the power grid company can send operational staff to the area with the highest probability, especially more than 0.2. When there is an outage of transmission lines caused by severe weather, the repair time could be reduced.
This work is supported by National Natural Science Foundation of China (No. 51307125).
- Chen J, Zhao C, Gu S et al (2016) Present status and development trend of lightning detection and protection technology of power grid in China. High Volt Eng 42(11):3361–3375Google Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.