Neural Network Wind Turbine Power Estimation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current methods for estimating achievable power production in wind turbines operated with a reduced power set point are not precise, leading to revenue discrepancies for operators and frequency imbalances in electricity networks.

Innovation Solution

A method using a neural network that determines and processes parameters like actual power production, blade pitch angle, and rotor speed to estimate achievable power production, allowing for precise power estimation with reduced computational requirements, enabling installation in simple processing devices within wind turbines.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Stability of the object's composition

If a reduced power set point is used to maintain frequency balance in the electricity network, then power balance and frequency stability are improved, but revenue accuracy deteriorates due to imprecise estimation of achievable power production

Engineering Contradiction:
Improvefrequency stabilityVSAvoidpower production estimation accuracy
Core Design Contradiction:
Stability of the object's compositionVSMeasurement precision

Solution Approach 1:

The patent replaces traditional mechanical measurement systems (wind speed sensors and power curves) with a neural network-based estimation system. The neural network processes multiple input parameters (actual power production, blade pitch angle, rotor speed) to accurately estimate achievable power production, resolving the contradiction between maintaining frequency stability through reduced power set points and preserving revenue accuracy through precise power estimation.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

2Measurement precision

If complex calculation methods are used to estimate achievable power production, then estimation precision is improved, but computational requirements and device complexity increase

Engineering Contradiction:
Improvepower production estimation accuracyVSAvoidprocessing device complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent employs a neural network that can be implemented on simple, low-cost processing devices with limited computational power. The neural network model is trained offline using comprehensive data, and during operation, it provides accurate power production estimates using only three easily measurable parameters, avoiding the need for complex real-time calculations while maintaining high estimation accuracy.

Inventive Principle:
Principle #27Cheap short-living objects (Disposable)

3Device complexity

If traditional wind speed measurement methods are used to estimate power production, then device simplicity is maintained, but estimation precision deteriorates

Engineering Contradiction:
Improvemeasurement system simplicityVSAvoidpower production estimation accuracy
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

The patent makes the existing control unit serve multiple functions: it not only controls the wind turbine operation but also estimates achievable power production. By utilizing data already collected by the control unit (actual power production, blade pitch angle, rotor speed) and adding neural network processing capability, the system achieves precise power estimation without adding separate complex measurement devices.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Data Source

PatentEP2192456B1Estimation an achievable power production of a wind turbine by means of a neural network
Publication Date: 2017.11.01 SIEMENS AG
  • EP2192456B1 patent drawingFigure 1
  • EP2192456B1 patent drawingFigure 2
  • EP2192456B1 patent drawingFigure 3~4

AI summary

It is described a method for estimating an achievable power production of a wind turbine (100), which is operated with a reduced power set point. The method comprises determining the values of at least two parameters being indicative for an operating condition of the wind turbine (100), inputting the values of the at least two parameters into a neural network (132, 232), and outputting an output value from the neural network (232). Thereby, the output value is an estimate of the achievable power production of the wind turbine (100). It is further described a control system (130), which is adapted to carry out the described power estimation method. Furthermore, it is described a wind turbine (100) and a computer program for controlling the described power estimation method.