Day-ahead joint generation scheduling and bidding optimization for power plants

Patent metadata

Field	Value
Publication number	US-11055732-B2
Application number	US-201816128939-A
Country	US
Kind code	B2
Filing date	Sep 12, 2018
Priority date	Sep 12, 2018
Publication date	Jul 6, 2021
Grant date	Jul 6, 2021

Abstract

Systems and methods to control activation of a generation unit. Compute an offer amount, representative of a value at which power is available to be supplied by a generation unit for a current or each subsequent upcoming time period in a time window. Construct energy supply curves using a set of points for each time period in the time window, to obtain associated amounts of power to supply with corresponding prices per each time period. Use a gradual decreasing price staircase approach to iteratively determine multiple offer amount scenarios. Submit the multiple offer amount scenarios along with corresponding values representative of available quantities of power to an operator computer that controls a market-based resource allocation system. Receive back a submitted supply value for the current or upcoming time period. Compare the submitted supply value to the submitted offer amount, and activate or deactivate the generation unit based on the comparison.

First claim

What is claimed is:

1. A system to control activation of a generation unit, comprising: a memory and a hardware processor, the hardware processor is configured to compute an offer amount representative of a value at which power is available to be supplied by a generation unit for a current or the each subsequent upcoming time period in a time window, such that the computation of the offer amount is based on multiple factors for each time period, wherein some of the multiple factors include submitted supply value information from an energy futures market and a predicted market clearing value submission (MCVS) and an associated distribution probability, and wherein a total number of MCVS scenarios per each time period equals to a maximum allowed number of points for an energy supply curve given by an energy operator, such that the time window includes a set of time periods with equal length; construct the energy supply curve using a set of points for each time period in the time window, to obtain associated amounts of power to supply with corresponding prices per each time period, wherein the set of points per each time period are calculated from a determined corresponding scheduling decision variable associated with the generation unit for per each time period in the time window; use a gradual decreasing price staircase approach to iteratively determine multiple offer amount scenarios during the current or the each subsequent upcoming time period; submit the multiple offer amount scenarios along with corresponding values representative of quantities of power that are available to be supplied by the generation unit during the current or the each subsequent upcoming time period to an energy operator computer that controls a market-based resource allocation system; receive from the energy operator computer a submitted supply value for the current or upcoming time period, the submitted supply value being based at least in part on the submitted multiple offer amount scenarios; compare the submitted supply value to the submitted offer amount scenarios; and activate or deactivate the generation unit based on the comparison.

2. The system of claim 1 , wherein a mean MCVS is estimated using a trained machine learning model using trading-related explanatory data over a set of historical time windows relevant to the time window, and non-trading related explanatory data for the time window, and standard deviations are estimated based on the estimated mean MCVS and a statistics of mismatches between actual market clearing prices (MCPs) and the market clearing value submissions (MCVSs) estimated using the trained machine learning model over the set of historical time windows, such that the historical time windows relevant to the time window include a previous time window, and a previous week same time window, and wherein the machine learning model is trained using historical market clearing prices (MCPs) for a set of historical time windows for training, and trading-related explanatory data and non-trading-related explanatory data for a set of time windows relevant to the set of historical time windows for training.

3. The system of claim 1 , wherein historical trading related explanatory data includes regional market clearing prices (MCPs), system market clearing power quantities (MCPQs), system total selling power quantities, and system total buying power quantities, and a non-trading related explanatory data include regional environmental data and inter-regional bi-directional power transfers and capacities.

4. The system of claim 1 , wherein the multiple factors for computing the offer amount for each time period include a set of MCVS scenarios that are evenly distributed within three times standard deviations of the predicted MCVS, such that an associated distribution probability for the MCVS scenario is determined based on the deviation of the MCVS scenario with predicted mean values of MCVS along with predicted standard deviations of MCVS for the time period, and normalized to make the associated distribution probability sum as 1 over all MCVS scenarios.

5. The system of claim 1 , wherein a determination of the set of points used for constructing the energy supply curves for each tim e period in the time window is performed by formulating and solving a double-objective optimization problem to maximize an expected profit gain over all price scenarios for each time period in the time window for market clearing prices (MCPs), and the expected profit gain over most risky price scenarios for each time period in the time window for the MCPs, while satisfying a non-decreasing offer quantity and price requirements among points per each time period, upward and downward ramping capacity constraints for required distance price scenario transition among consecutive time periods, and generator capacities, ramping rates, on/off times and durations constraints per each time period for each price scenario, wherein the price scenarios are arranged in ascending order on price.

6. The system of claim 5 , wherein most risky price scenarios are determined as a set of price scenarios whose accumulated probabilities equals or are greater than a given confidence level, wherein the accumulated probabilities for a scenario is determined by accumulating the acceptance probability of price scenarios from lowest scenario to the said scenario.

7. The system of claim 5 , wherein variations of scheduling for on/off statuses of the generation unit over time periods between price scenarios are limited by a scheduling capability of the unit determined by ramping capacities of the unit, wherein a same schedule for the on/off statuses over time periods is used for all scenarios for the generator with scenario invariant scheduling capability, wherein a different schedule for the on/off statuses over time periods may be used for each scenario for a generator with scenario-based scheduling capability, wherein a common schedule for the on/off statuses over time periods is used for each of a scenario group for the generator with scenario group-based scheduling capability.

8. The system of claim 5 , wherein the expected profit gain is determined as a sum of a product of offer power quantity and offer power price for each price scenario weighted by a normalized acceptance probability of the said price scenario, wherein the normalized acceptance probability of a price scenario is determined by dividing an acceptance probability of the price scenario by a sum of the acceptance probabilities of all scenarios chosen according to the expected profit gain calculation.

9. The system of claim 8 , wherein the acceptance probability of the price scenario is determined based on an accumulation of the distribution probability of all price scenarios whose prices equal or are greater than the price of the said scenario.

10. The system of claim 5 , wherein the optimization problem is converted into a single-objective optimization problem by maximizing an overall objective satisfying degree of two objectives, wherein the overall objective satisfying degree is determined as a minimal of an objective satisfying degrees for each objective, wherein the objective satisfying degree of each objective is determined based on a desired maximum value, and a tolerable minimum value for the objective.

11. The system of claim 10 , wherein the gradual decreasing price staircase approach is performed by using an outer loop to iterate price scenario in a descending order from a highest price to a lowest price chosen from a set of predicted MCVSs, and an inner loop to iterate an offering step level in descending order from an highest level defined as a minimum of the level corresponding to a current price scenari

Assignees

• Mitsubishi Electric Res Laboratories Inc

Inventors

• Sun Hongbo

Classifications

• G06N3/0499

  Feedforward networks · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06Q50/06

  Energy or water supply · CPC title

• G06N7/08

  using chaos models or non-linear system models · CPC title

• G06Q10/063

  Operations research, analysis or management · CPC title