In the conventional technique, in the evaluation of the severity index, clustering and loading suffer from more iteration leading to more computational delay. Hence this research article identifies, a novel progression for fast predicting the severity of the line and clustering by incorporating machine learning aspects. The polynomial load modelling or ZIP (constant impedances (Z), Constant Current (I) and Constant active power (P)) is developed in the IEEE-14 and Indian 118 bus systems considered for analysis of power system security. The process of finding the severity of the line using a Hybrid Line Stability Ranking Index (HLSRI) is used for assisting the concepts of machine learning with J48 algorithm, infers the superior affected lines by adopting the IEEE standards in concern to be compensated in maintaining the power system stability. The simulation is performed in the WEKA environment and deals with the supervisor learning in order based on severity to ensure the safety of power system. The Unified Power Flow Controller (UPFC), facts devices for the purpose of compensating the losses by maintaining the voltage characteristics. The finite element analysis findings are compared with the existing procedures and numerical equations for authentications.

Electricity is a necessity for the maximum elementary condition of the contemporary world. Hence, safeguarding the safety of the power system is equally important. The major goal of power system security is to deliver consistent power to the clients short of disruption, harm to the user utilisation, and financial process of the power system. But such a power system is also disposed to numerous problems, like the transmission line outage, the generator outage, the rapid increase in load demand, the loss of a transformer, etc., which are known as contingencies. Safety is a key motivator in the power system. In recent years, blackouts due to power system contingencies have increased. It causes industries to lose money and disrupts people’s daily lives. Thus, it’s important to accept a precise and active measure to stop the blackout’s contingency from spreading to other lines. Contingency analysis is used to quickly guess the system stability right after the outage or any abnormal conditions. The purpose of the contingency plan is to ascertain the change within the device’s functioning that can occur after the fault element is removed. Large violations of inline flow may end in single line outages, which may cause cascading effects of the outages and may also cause overloading on the other lines. If such an overload results from a line outage, there’s an immediate need for the regulator’s action to be taken. Therefore, contingency analysis is one of the most important tasks to be completed by power system planners and operation engineers. Several different methods of ranking potential outcomes, both static and dynamic, are used in the process of contingency screening [

Load models may be used to predict how loads will respond to changes in voltage or frequency. It’s important to choose a load model that is easy to understand and can appropriately represent a variety of load response situations while executing. In this study, the effects of steady-state and polynomial load models are examined. This model is less precise than the polynomial load model because it depicts a combination of three different sorts of features in a single load. It is possible to express resistive loads, induction motor loads, and variable-frequency loads in a polynomial load model by using the constant impedance, the constant current, and the constant power. As a result, the polynomial load model is more accurate since it accurately depicts load [

UPFC is the most adaptable and versatile FACTS device [

Contingency analysis is a well-known function in modern Energy Management Systems. The system’s contingency analysis could be a major movement in power grid planning and process. Generally, an outage of a single line or transformer may lead to overloads in other branches and also cause sudden system voltage rises or drops [

There are various types of contingencies such as generator outages, line outages, bus outages, and transformer outages. Line outages are analysed in this paper as they are the most frequently occurring contingency in the system.

Contingency occurs mainly due to sudden increase or decrease of loads. Load modeling plays a major role in analyzing various types of loads. Load modeling refers to the mathematical illustration of the connection between the power and the voltage in a load bus. It has a significant impact on power system studies.

Two types of load models are considered in this project for analyzing the contingency, which are named as Steady state load model and polynomial load model [

Steady state load model is also known as constant load model. The active and reactive power equations in this model are represented as shown in

The diagonal and off-diagonal elements of active power and reactive power are developed.

Polynomial load model is also known as ZIP load model. Z represents constant impedance, which represent constant current and P represents constant power [

At Bus-i:

At Bus-j:

The diagonal and off-diagonal elements of active power and reactive power at Bus-i and Bus-j are developed. These load models are incorporated with Newton-Raphson power flow solution method. Newton-Raphson method is used as it is faster, more reliable; results are accurate and quadratic type convergence. Jacobian matrix is formed using the developed diagonal and off-diagonal elements. The Jacobian matrix springs the linear connection between the small vagaries in voltage magnitude and phase angle with the small vagaries in real and reactive power as shown in

The variance between the programmed and the designed values known as power residual is given in

The new estimates for voltage angle and voltage magnitude are shown in

These two load models are applied to IEEE-118 bus system for analyzing the behavior of the loads. Line stability index gives support to find most severe lines.

To get the Hybrid Line Stability Ranking Index (HLSRI), we first look at the Line Stability Index and the Fast Voltage Stability Index (FVSI) Then, we showed that the FVSI is a calculation of the LSI, and we went on to figure out the HLSRI, which is faster and more accurate. LSI is given in

where,

The FVSI is derived from LSI when the voltage angle difference between sending and receiving end is assumed to be very small (i.e.,

Therefore, propose to combine _{C}, to determine β is 1 or 0. The Proposed index is formed by combining

The HLSRI is given as:

where,

The Large voltage angle differences may signal a blackout or system failure. Angle delta (‘

From the

The Waikato Environment for Knowledge Analysis (WEKA) was developed at the University of Waikato, New Zealand. The software provides tools for data processing, implementation of several machine learning algorithms, and visualisation tools. It is free, open-source software that provides a range of tools for analysing large amounts of data. Supervised learning refers to ML forecasting methods. Unsupervised learning shows a dataset’s case distribution. Supervised learning methods detect the relationship between input and target. Relationships are prototypical. Classification and regression models are common supervised methods. Machine learning offers various data analysis types. This paper uses the J48 algorithm to cluster data from proposed indices as shown in

The HLSRI is used in conjunction with the J48 classification model in order to quickly predict and cluster the list of critical lines from high to low value. Once the critical line is found, compensation is used to make sure that the system stays stable and safe [

A FACTS device plays a vital role in controlling power and enhancing the working volume of existing lines. The basic purpose of the parallel inverter is to supply real power required by the series inverter through the common DC link. The parallel inverter can also grip or produce controllable reactive power, as shown in

The active and reactive power equation of the Unified Power Flow Controller is developed with ZIP load model as shown in below.

The crosswise and off-diagonal elements are developed. By using this developed mathematical model, MATLAB program is developed and used for compensating the most severe lines.

PSO searches for the optimal objective location using random agents. Each agent’s position and velocity are updated over generations. Each agent’s new position combines its previous velocity and current position. At each stage of an agent’s situation, P_{best} is updated. Each agent knows the G_{best} group is the P_{best’s} best. After a fixed number of rounds, P_{best} & G_{best} agents reach global optimality. Updated agent velocity and position.

^{th} particle current velocity at k itr,

rand1 & rand2 = random numbers range from 0 to 1

^{th} particle until iteration k,

C_{1}, C_{2} = Accelerating constants

The order of magnitude and phase angles of the infused voltage of the UPFC’s VSC serve as optimal control parameters as shown in

Parameters | Values |
---|---|

Max number of iterations | 80 |

Swarm size | 50 |

Problem dimension | 2 |

Inertia weight factor | 1.0 |

c1 and c2 | 2 and 2 |

rand1 and rand2 | Rand (0,1) |

In this paper,

% Loading | FB-TB | P_{Transfer} _{capability} (MW) |
Total P_{loses} (MW) |
||||||
---|---|---|---|---|---|---|---|---|---|

Normal | Contingency | Compensation (UPFC) | PSO tuned UPFC | Normal | Contingency | Compensation (UPFC) | PSO tuned UPFC | ||

100 | 10–11 | 13.62 | 15.82 | 14.62 | 18.62 | 1.37 | 3.77 | 2.37 | 1.87 |

06–12 | 25.16 | 27.36 | 26.16 | 29.16 | 4.53 | 6.93 | 5.53 | 4.63 | |

13–14 | 15.43 | 17.63 | 16.43 | 19.43 | 1.14 | 3.54 | 2.14 | 1.34 | |

150 | 10–11 | 11.62 | 13.82 | 12.62 | 16.22 | 2.37 | 4.77 | 3.37 | 2.87 |

06–12 | 21.16 | 23.36 | 22.16 | 25.46 | 5.53 | 7.93 | 6.53 | 4.63 | |

13–14 | 13.43 | 15.63 | 14.43 | 17.53 | 2.14 | 4.54 | 3.14 | 2.46 |

Note: FB: From Bus TB: To Bus.

From the below

% Loading | FB-TB | P_{Transfer} _{capability} (MW) |
Total P_{loses} (MW) |
||||||
---|---|---|---|---|---|---|---|---|---|

Normal | Contingency | Compensation (UPFC) | PSO tuned UPFC | Normal | Contingency | Compensation (UPFC) | PSO tuned UPFC | ||

100 | 70–74 | 17.18 | 9.81 | 41.87 | 44.67 | 137.85 | 138.33 | 133.43 | 126.62 |

71–72 | 15.15 | 1.84 | 21.54 | 26.64 | 137.85 | 137.68 | 132.45 | 128.52 | |

24–72 | 3.020 | 17.53 | 14.29 | 18.48 | 137.85 | 137.71 | 132.92 | 128.68 | |

150 | 69–75 | 440.79 | 755.81 | 437.17 | 430.24 | 861.68 | 985.25 | 689.38 | 684.62 |

69–77 | 603.98 | 595.17 | 587.06 | 589.06 | 861.68 | 948.64 | 692.02 | 688.31 | |

70–74 | 5.17 | 23.81 | 48.60 | 50.601 | 861.68 | 861.83 | 688.26 | 688.43 |

The Indian 118 bus test system is in

The proposed index is applied to the IEEE 14 and Indian 118 test systems and mathematical models of steady state and polynomial load models (ZIP) are developed and analysed with the IEEE test system in view of the single line outage at a time. The Indian 118 test system is analysed for various percentages of ZIP load model for contingency with and without compensation using the PSO algorithm. A Hybrid Line Stability Ranking Index shows the most critical lines in the system for which compensation is provided by placing a unified power flow controller with proper tuning using PSO. Based on the outcomes, it is apparent that the stress level is reduced, and the system is kept stable and healthy by providing suitable compensation at a suitable location.

The authors would like to thank JNT University Anantapur, Ananthapuramu and Sree Vidyanikethan engineering college, Tirupati, Andhra Pradesh, India and also, we like to thank Anonymous reviewers for their so-called insights.

The authors received no specific funding for this study.

The authors declare that they have no conflicts of interest to report regarding the present study.