Khan_cmu_0041E_10263.pdf (3.01 MB)
Novel Control Algorithms for Hierarchical Control of Power Systems
With the recent large scale integration of Distributed Generation (DG), the power system has changed drastically. Due to the intermittent nature of DG, real-time
control of power generating units has become much more challenging. Consequently, voltage abnormalities and power loss problems are more frequent now in power systems. Load frequency stability is another serious concern in this
changing paradigm of power systems. Also, large scale integration of DGs would require massive amounts of real-time data to be communicated from local systems
to the controller to achieve effective regulation. However, this ever-growing data sizes may incur delays in transmission that in turn, may slow down the control algorithms.
This thesis addresses the issue of communication congestion in transmission of real-time data from local power systems to the centralized controller. It addresses
this communication congestion issue in the context of three control problems in the power system: Load frequency control, secondary voltage control and optimal
reactive power control. Load Frequency Control (LFC), is employed to allow an area to first meet its own load demands, then to assist in returning the steadystate
frequency of the system with a response time of a few seconds. The fast LFC may be affected with the slow transmission of information from the system.
To deal with this communication bottleneck, a Singular Value Decomposition (SVD) based LFC algorithm is proposed where SVD is used to significantly reduce
the size of transmitted information. A second communication efficient control solution is proposed to address secondary voltage control of multi-area power
systems which utilizes compressive sensing (CS) techniques to reduce the data size to deal with the limited bandwidth problem of the communication channel.
It is also equipped with a technique based on Mathematical Morphology Singular Entropy (MSE) to identify faults/ abnormal disturbances locally in the system to
avoid bad data/corrupt data being sent to the central controller. To further filter the sensed measurements from local sensors, it is passed through Mathematical
Morphological Filters (MMF). Finally, a tertiary control algorithm for optimal reactive power generation control is proposed to minimize power loss, and voltage
deviation on the tertiary level control. A consensus based gradient distributed approach is proposed to deal with potential communication delays.
control of power generating units has become much more challenging. Consequently, voltage abnormalities and power loss problems are more frequent now in power systems. Load frequency stability is another serious concern in this
changing paradigm of power systems. Also, large scale integration of DGs would require massive amounts of real-time data to be communicated from local systems
to the controller to achieve effective regulation. However, this ever-growing data sizes may incur delays in transmission that in turn, may slow down the control algorithms.
This thesis addresses the issue of communication congestion in transmission of real-time data from local power systems to the centralized controller. It addresses
this communication congestion issue in the context of three control problems in the power system: Load frequency control, secondary voltage control and optimal
reactive power control. Load Frequency Control (LFC), is employed to allow an area to first meet its own load demands, then to assist in returning the steadystate
frequency of the system with a response time of a few seconds. The fast LFC may be affected with the slow transmission of information from the system.
To deal with this communication bottleneck, a Singular Value Decomposition (SVD) based LFC algorithm is proposed where SVD is used to significantly reduce
the size of transmitted information. A second communication efficient control solution is proposed to address secondary voltage control of multi-area power
systems which utilizes compressive sensing (CS) techniques to reduce the data size to deal with the limited bandwidth problem of the communication channel.
It is also equipped with a technique based on Mathematical Morphology Singular Entropy (MSE) to identify faults/ abnormal disturbances locally in the system to
avoid bad data/corrupt data being sent to the central controller. To further filter the sensed measurements from local sensors, it is passed through Mathematical
Morphological Filters (MMF). Finally, a tertiary control algorithm for optimal reactive power generation control is proposed to minimize power loss, and voltage
deviation on the tertiary level control. A consensus based gradient distributed approach is proposed to deal with potential communication delays.
History
Date
2018-07-02Degree Type
- Dissertation
Department
- Electrical and Computer Engineering
Degree Name
- Doctor of Philosophy (PhD)