Stochastic Look-ahead Dispatch Using Newton Method

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional look-ahead dispatch methods for power systems struggle with accurately predicting wind power output and handling the stochastic nature of renewable energy, leading to high computational complexity and conservative optimization that increases costs.

Innovation Solution

A stochastic look-ahead dispatch method based on the Newton method that uses a Gaussian mixture model to fit renewable energy output and transforms chance constraints into deterministic linear constraints, allowing for efficient unit commitment and cost reduction.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If robust look-ahead dispatch is used to handle stochastic renewable energy, then system reliability is improved, but dispatch cost increases due to conservative optimization

Engineering Contradiction:
Improvesystem reliabilityVSAvoiddispatch cost
Core Design Contradiction:
ReliabilityVSLoss of energy

Solution Approach 1:

The patent transforms the stochastic optimization problem into a deterministic equivalent by changing the parameter representation from random variables to their quantile values. The chance constraints are converted into deterministic constraints using the relationship Pr{g(x,ξ)≤0}≥1-α, where the random parameters are replaced by their (1-α)-quantiles. This parameter transformation allows the system to achieve reliability guarantees without the excessive conservatism of robust optimization.

Inventive Principle:
Principle #35Parameter changes

2Loss of energy

If stochastic look-ahead dispatch with chance constraints is used, then dispatch cost is reduced, but computational complexity increases due to random variables in constraints

Engineering Contradiction:
Improvedispatch costVSAvoidcomputational complexity
Core Design Contradiction:
Loss of energyVSDevice complexity

Solution Approach 1:

The patent extracts the randomness from the constraints by separating the deterministic and stochastic components. The chance constraints Pr{g(x,ξ)≤0}≥1-α are transformed into deterministic equivalent constraints by taking out the probabilistic element and replacing it with quantile-based deterministic relationships. This extraction eliminates random variables from the constraints, significantly reducing computational complexity while maintaining the cost-saving benefits of stochastic optimization.

Inventive Principle:
Principle #2Taking out (Extraction)

3Device complexity

If conventional prediction methods are used for wind power output, then model simplicity is maintained, but prediction accuracy deteriorates

Engineering Contradiction:
Improvemodel simplicityVSAvoidprediction accuracy
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

The patent employs a composite prediction model that combines multiple probability distribution functions (Gaussian, Weibull, Rayleigh distributions) to model wind power output. This composite approach integrates the strengths of different distribution models, allowing accurate fitting of wind power characteristics while maintaining reasonable model complexity. The composite model provides superior prediction accuracy compared to single-distribution models.

Inventive Principle:
Principle #40Composite materials

Data Source

PatentUS11043818B2Random rolling scheduling method for power system based on Newton method
Publication Date: 2021.06.22 STATE GRID CORPORATION OF CHINA

AI summary

The disclosure provides a stochastic look-ahead dispatch method for power system based on Newton method, belonging to power system dispatch technologies. The disclosure analyzes historical data of wind power output, and uses statistical or fitting software to perform Gaussian mixture model fitting. A dispatch model with chance constraints is established for system parameters. Newton method is to solve quantiles of random variables obeying Gaussian mixture model, so that chance constraints are transformed into deterministic linear constraints, thus transforming original problem to convex optimization problem with linear constraints. Finally, the model is solved to obtain look-ahead dispatch. The disclosure employs Newton method to transform chance constraints containing risk level and random variables into deterministic linear constraints, which effectively improves model solution efficiency, and provides reasonable dispatch for decision makers. The disclosure is employed to the dispatch of the power system including large-scale renewable energy grid-connected.