Optimal sensor deployment and data analytics for power distribution network visibility and control

点击翻译按钮获取中文摘要

Retrofitting Network Measurement devices such as CT/VT/PMU to gain network observability can be costly and require system access (outages) which can be difficult to plan and typically/currently require years of planning. Measuring devices are typically used for protection, metering and control purpose and currently or often not used for real time network observability. Characteristics of domestic and industrial load is evolving; network infrastructure is also evolving due to emerging DER installations in recent years. Network observability must be established by new tools that should not rely much on historical load forecast and network model rather on the data from smart sensor. It is in Distribution Network Operator's (DNO) interest to have Active network management (ANM) in place for DSO readiness. New and existing measurement and control devices play a vital role in active network management such as knowing voltage stability and thermal limits of all feeders. It is impractical to install measurement devices at end of every feeder due to high capital cost and system access requirement - strategic placement of measurement devices can be explored. This will form the backbone of voltage and var control (VVC) involving slow and fast voltage control devices. New tools are needed first for the evaluation of various quantities such as voltage, active/reactive power and flow which can then establish a network operating situation map in the primary control which needs state estimators that must cope with the changes that is taking place in the demand side.The objectives are:1.Costeffectivesensorplacementinthenetwork 2. Data compression algorithm to reduce the volume of data transmission to control centre 3. Voltage and var control which will address slow discrete voltage control with fast smooth voltage control from power converter through convexification of mixed integer non-linear programming. The optimal sensor allocation requires robust algorithm covering all evolving future scenarios in distribution network flow. The data analytics will be based on predictive modelling which is new and will not rely much on network topology processing. The convexification of VVC should respect the DG power capability characteristic while handling both continuous (DG reactive power control) and discrete (OLTC, and switchable shunt capacitor and reactor banks) decision variables to provide optimum voltage and power flow control support in active distribution network.