Search

CN-122022582-A - Comprehensive evaluation weight predictive adjustment method and system for offshore wind farm direct current collector system based on digital twin and rolling time window

CN122022582ACN 122022582 ACN122022582 ACN 122022582ACN-122022582-A

Abstract

The invention belongs to the technical field of intelligent operation and maintenance of offshore wind power and digital twin, and discloses a comprehensive evaluation weight predictive adjustment method and system of a DC current collection system of an offshore wind power plant based on a digital twin and rolling time window, wherein the method comprises the steps of realizing advanced rolling prediction of system performance indexes at T moments in the future by constructing a digital twin model comprising electric, mechanical and thermodynamic multi-physical field coupling and combining environment prediction data; and (3) establishing a rolling optimization problem with a weight smoothness constraint aiming at the maximum accumulated discount comprehensive score, solving an optimal future weight sequence, and applying the first prediction weight to system evaluation in real time to form a 'prediction-optimization-execution-correction' closed loop. According to the method, weight adjustment is improved from passive response to active guidance, and evaluation accuracy and decision prospective of the system under complex working conditions are remarkably enhanced.

Inventors

  • CHE YANBO
  • WANG LEI
  • LI PEIYI

Assignees

  • 天津大学

Dates

Publication Date
20260512
Application Date
20260131

Claims (8)

  1. 1. The method for predictively adjusting the comprehensive evaluation weight of the direct current collecting system of the offshore wind farm based on the digital twin and rolling time window is characterized by comprising the following steps: Constructing a digital twin model synchronous with a physical direct current collecting system, and combining environment prediction data to realize advanced rolling prediction of performance indexes of the direct current collecting system of the offshore wind farm at T moments in the future; based on the result of advanced rolling prediction, establishing a rolling optimization problem with weight smoothness constraint aiming at the maximum accumulation of discount comprehensive scores, and solving an optimal weight sequence; and applying the optimized first future moment weight to real-time comprehensive evaluation of the physical system, and updating the digital twin model based on actual operation data.
  2. 2. The method of claim 1, wherein the digital twin model is a multi-physical field coupling model comprising an electrical property sub-model, a mechanical stress sub-model, and a thermodynamic property sub-model.
  3. 3. The method of claim 1, wherein the objective function of the rolling optimization problem is a cumulative and maximum value of a discount composite score over a future time window expressed as: ; Wherein, the As a discount factor, the number of times the discount is calculated, Is the dot product of the weight vector and the performance score vector, i.e., the composite score, T is the number of steps of the rolling prediction time window, k=1,.., As the performance vector at time t + k, Is a time step weight vector.
  4. 4. A method according to claim 3, wherein the rolling optimization problem comprises a weight smoothness constraint: ; Wherein, the And presetting a maximum weight change threshold value.
  5. 5. The system for adjusting the comprehensive evaluation weight predictability of the offshore wind farm direct current collector system based on the digital twin and rolling time window is used for realizing the method of any one of claims 1-4, and is characterized by comprising a digital twin engine module, a prediction database module, a rolling optimization solver module, a weight prediction execution module and a model self-adaptation update module; The digital twin engine module is used for running a digital twin model and carrying out multi-step advanced state prediction; The prediction database module is used for storing a predicted state sequence and environment data; the rolling optimization solver module is used for solving the rolling optimization problem and outputting an optimal weight sequence; The weight prediction execution module is used for implementing optimal weight and managing weight switching; The model self-adaptive updating module is used for correcting digital twin model parameters based on measured data.
  6. 6. The system of claim 5, wherein the digital twin model is a multi-physical field coupling model comprising an electrical property sub-model, a mechanical stress sub-model, and a thermodynamic property sub-model.
  7. 7. The system of claim 5, wherein the objective function of the rolling optimization problem is a cumulative and maximum value of a discount composite score over a future time window expressed as: ; Wherein, the As a discount factor, the number of times the discount is calculated, Is the dot product of the weight vector and the performance score vector, i.e., the composite score, T is the number of steps of the rolling prediction time window, k=1,.., As the performance vector at time t + k, Is a time step weight vector.
  8. 8. The system of claim 7, wherein the rolling optimization problem comprises a weight smoothness constraint: ; Wherein, the And presetting a maximum weight change threshold value.

Description

Comprehensive evaluation weight predictive adjustment method and system for offshore wind farm direct current collector system based on digital twin and rolling time window Technical Field The invention belongs to the technical field of intelligent operation and maintenance of offshore wind power and digital twin, and particularly relates to a comprehensive evaluation weight predictive adjustment method and system of a direct current collecting system of an offshore wind power plant based on a digital twin and rolling time window. Background The direct current collecting system of the offshore wind farm is a key link for connecting the offshore wind turbine generator and the land power grid, and accurate evaluation of the running performance of the direct current collecting system is important for guaranteeing safe and economic running of the wind farm. Comprehensive evaluation generally relates to three dimensions of stability, reliability and energy efficiency, and reasonable distribution of weights of all dimensions directly influences scientificity and decision guiding value of an evaluation result. In the prior art, weight adjustment methods are mainly divided into two types, namely a fixed weight method based on expert experience and incapable of adapting to time-varying working conditions of a system, and a dynamic weight method based on real-time feedback, such as a reinforcement learning method, for adjusting weights through analysis of the current state and historical data of the system. The latter, while enabling dynamic adjustment of weights, has inherent drawbacks, decision hysteresis. That is, the weight adjustment is always based on the system state change which occurs, belongs to the 'perception-response' mode, and the response is slow when the sudden and severe working condition change is dealt with. For example, the wind speed will increase dramatically before a storm comes, and the traditional method must wait for the wind speed to actually increase and the voltage to start to fluctuate before increasing the stability weight. This hysteresis may lead to the system failing to obtain optimal assessment guidance during critical periods, missing optimal intervention opportunities. The digital twin technology provides a new approach for predictive maintenance and optimization by constructing virtual mapping of physical systems. However, the existing digital twin technology is mostly used for state monitoring and fault diagnosis, and the report of deep fusion of the existing digital twin technology and the rolling time window optimization is not yet seen, and the existing digital twin technology is used for solving the problem of predictive adjustment of comprehensive evaluation weights. Disclosure of Invention In order to solve the decision hysteresis defect of the existing weight adjustment method, the invention provides a comprehensive evaluation weight predictive adjustment method and system for a direct current collecting system of an offshore wind farm based on a digital twin and rolling time window, which realize the crossing from 'passive adaptation' to 'active guidance' and improve the comprehensive evaluation accuracy and decision validity of the system under complex working conditions. In order to achieve the above object, the present invention provides the following solutions: A comprehensive evaluation weight predictive adjustment method of a direct current collector system of an offshore wind farm based on a digital twin and rolling time window comprises the following steps: Constructing a digital twin model synchronous with a physical direct current collecting system, and combining environment prediction data to realize advanced rolling prediction of performance indexes of the direct current collecting system of the offshore wind farm at T moments in the future; based on the result of advanced rolling prediction, establishing a rolling optimization problem with weight smoothness constraint aiming at the maximum accumulation of discount comprehensive scores, and solving an optimal weight sequence; and applying the optimized first future moment weight to real-time comprehensive evaluation of the physical system, and updating the digital twin model based on actual operation data. Preferably, the digital twin model is a multi-physical field coupling model, including an electrical property sub-model, a mechanical stress sub-model and a thermodynamic property sub-model. Preferably, the objective function of the rolling optimization problem is the cumulative and maximum value of the discount composite score over a future time window expressed as: ; Wherein, the As a discount factor, the number of times the discount is calculated,Is the dot product of the weight vector and the performance score vector, i.e., the composite score, T is the number of steps of the rolling prediction time window, k=1,..,As the performance vector at time t + k,Is a time step weight vector. Preferably, the rol