阅读说明：本技术 一种风电机组测控系统故障检测与故障容错方法 (Fault detection and fault tolerance method for wind turbine measurement and control system ) 是由 柏林 刘小峰 罗宏林 于 2020-12-25 设计创作，主要内容包括：本发明涉及一种用于风电机组测控系统的故障检测与故障容错方法,适用于风电机组测控系统早期故障检测和故障容错,有利于提高风力发电机组的可用性和可靠性。本发明以风电测控系统存在的非线性、测试误差及扰动随机性、模型误差等问题为切入点,对其进行线性时变参数模型描述,采用集值观测的方法进行故障检测与隔离,使用虚拟传感器和执行器进行容错处理。本发明有助于增强风电机组测控系统可靠性,提高其可用性。(The invention relates to a fault detection and fault tolerance method for a wind turbine measurement and control system, which is suitable for early fault detection and fault tolerance of the wind turbine measurement and control system and is beneficial to improving the availability and reliability of a wind turbine. The method takes the problems of nonlinearity, test error, disturbance randomness, model error and the like of the wind power measurement and control system as entry points, carries out linear time-varying parameter model description on the wind power measurement and control system, adopts a value-integrated observation method to carry out fault detection and isolation, and uses a virtual sensor and an actuator to carry out fault tolerance treatment. The invention is beneficial to enhancing the reliability of the wind turbine measurement and control system and improving the availability of the wind turbine measurement and control system.)

1. A wind turbine measurement and control system fault detection method based on collection value observation is adopted. Designing a normal operation state set value observer of the wind turbine generator, wherein the input and output states of the observer are used as input and output variables of the observer, an approximate feasible set of state estimation in a sampling period is obtained through time updating and measurement updating, if the approximate feasible set is detected as an empty set, the system is judged to have a fault, and fault detection information is output at the same time; otherwise, the system is judged to be fault-free.

2. A wind turbine measurement and control system fault isolation method based on collection value observation is adopted. The method comprises the steps that various fault isolation set value observer units with strict differentiability of the wind turbine generator are designed, when a decision unit judges that the wind turbine generator has faults by using a normal operation state set value observer, the fault set value observer is initialized, the fault set value observer sequentially outputs an approximate feasible set of state estimation in a sampling period, and when the approximate feasible set of state estimation of a certain fault set value observer is not empty, the faults can be judged to occur, so that effective isolation of the faults is achieved.

3. A wind turbine measurement and control system fault tolerance method based on a virtualizer is adopted. The fault tolerance idea is implemented by adopting the virtual sensor actuator, and the virtual sensor actuator is utilized to process fault information instead of the controller to process the fault information. And fault evaluation quantitative information provided by the fault detection unit is combined with the prior functional redundancy and fault analysis knowledge base, and fault signals are processed in modes of replacement, correction or compensation and the like according to a mode matching principle and then provided to the controller in a mode of model matching input and output.

Technical Field

The invention relates to a method for fault detection and fault tolerance of a wind generating set measurement and control system, which is suitable for early fault detection and fault tolerance of a wind power measurement and control system and is beneficial to improving the availability and reliability of the wind generating set.

Background

The method has the advantages that fault detection and fault tolerance are carried out on the wind power measurement and control system, faults of a wind motor are reduced from the source, and major accidents of the wind power system and expansion of early faults are avoided; the early failure of the wind power system is solved by adopting fault-tolerant control, and the availability of the wind turbine generator is greatly improved. A kalman filter is generally used to detect whether a system is normal, but there are great difficulties in detecting a residual signal and a detection threshold that are not sensitive to uncertainty and test noise but are very sensitive to fault information. The most common fault-tolerant method is to utilize hardware redundancy to carry out fault tolerance, but the fault-tolerant method of hardware redundancy has the defects of high price and the like. Therefore, the invention provides a novel method for judging whether the system is normal or not based on the fault detection process of the collective estimation and correcting the fault state by implanting reconfigurable virtual sensors and actuators for fault tolerance.

Disclosure of Invention

Aiming at the characteristics of nonlinearity, test error, disturbance randomness, model parameter uncertainty and the like of the wind power measurement and control system, the early-stage fault of the wind power measurement and control system is reasonably dealt with by adopting a fault detection and virtualizer fault tolerance method of collective estimation, and the availability of the wind power generator set is improved.

In order to achieve the purpose, the invention adopts the following technical scheme:

(1) firstly, fault detection is carried out by adopting a method based on set value observation. And establishing a wind motor measurement and control system model under the consideration of factors such as noise, disturbance and parameter uncertainty, and delimiting the noise, the disturbance and the parameter uncertainty. Designing a normal operation state set value observer of the wind turbine generator, wherein the input and output states of the observer are used as input and output variables of the observer, an approximate feasible set of state estimation in a sampling period is obtained through time updating and measurement updating, if the approximate feasible set is detected as an empty set, the system is judged to have a fault, and fault detection information is output at the same time; otherwise, the system is judged to be fault-free.

(2) On the basis, fault isolation is carried out by adopting a method based on set value observation. In order to effectively isolate faults when a wind power measurement and control system breaks down, various fault isolation set value observer units of a wind turbine generator are designed, each fault set value observer in each unit is responsible for detecting one fault and has strict differentiability, the differentiability between each fault set value observer is guaranteed theoretically through the rank relation of a state space equation matrix, and meanwhile, the further verification is carried out through experiments. For some early faults, when fault signals are drowned by noise and are difficult to detect or the detection time is too long, targeted excitation signals are designed by combining a fault analysis information base to carry out active fault diagnosis. On the premise of guaranteeing the stability and reasonable load of the system, the optimal amplitude and frequency of the excitation signal are determined through experiments, and the separability of the fault signal is enhanced. When the decision unit judges that the wind turbine generator has a fault by using the normal operation state set observer, the fault set observer is initialized, the fault set observer sequentially outputs an approximate feasible set of state estimation in a sampling period, and when the approximate feasible set of state estimation of a certain fault set observer is not empty, the fault can be judged to occur, so that the effective isolation of the fault is realized.

(3) And finally, adopting a fault tolerance method based on the virtual device. Continued use of the normal state controller in a fault condition will result in degraded performance and even damage to the wind turbine, but a non-fatal fault causing a shutdown pending repair will result in lost power generation. The fault tolerance idea is implemented by adopting the virtual sensor actuator, and the virtual sensor actuator is utilized to process fault information instead of the controller to process the fault information. And fault evaluation quantitative information provided by the fault detection unit is combined with the prior functional redundancy and fault analysis knowledge base, and fault signals are processed in modes of replacement, correction or compensation and the like according to a mode matching principle and then provided to the controller in a mode of model matching input and output.

The invention has the advantages that: the set value observation fault detection isolation method fusing linear time-varying parameters in an unknown bounded form avoids the problems that a residual error threshold value fault detection method needs to accurately set a threshold value and test errors and disturbance are difficult to parameterize, and improves the detection efficiency and accuracy by an active fault detection method. The quantitative information provided by fault detection is fully utilized, fault-tolerant processing is realized by the idea of a virtual device, and the economic practicability is realized while the fault-tolerant performance is ensured.

Drawings

The accompanying drawings, which are included to provide a further understanding of the embodiments of the invention and are incorporated in and constitute a part of this application, illustrate embodiment(s) of the invention and together with the description serve to explain the principles of the invention. In the drawings:

FIG. 1 is a flow diagram of a method for implementing fault detection based on a collective observer;

FIG. 2 is a schematic diagram of a fault detection isolation implementation method based on a set-value observer.

FIG. 3 is a schematic diagram of a fault tolerance implementation method of a virtual machine.

Detailed Description

The specific implementation of fault detection by using a method based on collective value observation is shown in fig. 1, and a model of a wind turbine measurement and control system under the factors of noise, disturbance, parameter uncertainty and the like is established to delimit the noise, the disturbance and the parameter uncertainty. Designing a normal operation state set value observer of the wind turbine generator, wherein the input and output states of the observer are used as input and output variables of the observer, an approximate feasible set of state estimation in a sampling period is obtained through time updating and measurement updating, if the approximate feasible set is detected as an empty set, the system is judged to have a fault, and fault detection information is output at the same time; otherwise, the system is judged to be fault-free.

On the basis, a specific flow of fault isolation by using a set value observation-based method is shown in fig. 2. When the decision unit judges that the wind turbine generator has a fault by using the normal operation state set observer, the fault set observer is initialized, the fault set observer sequentially outputs an approximate feasible set of state estimation in a sampling period, and when the approximate feasible set of state estimation of a certain fault set observer is not empty, the fault can be judged to occur, so that the effective isolation of the fault is realized. The structure adopts a recursion mode to carry out fault isolation, can realize the detection and isolation of multiple faults and provides quantized fault information.

Implementation of fault tolerance using virtualizers is shown in fig. 3. The wind turbine measurement and control system is controlled by a controller K(s), and when a fault occurs, the system G(s) is changed into G_f(s), accessed virtual executor G_a(s) satisfies the condition G_f(s)G_a(s) G(s), so that the virtual actuatorSensor is changed from H (s is changed into H)_f(s) virtual sensor G of Access_s(s) satisfies the condition G_s(s)H_f(s) ═ H(s), soA plurality of sensor actuators will be provided withDescribed in matrix form.The elements of the transfer function matrix of the virtual actuator,a virtual sensor transfer function matrix element. For a sensor signal stuck fault, a measured value is replaced by an estimated value based on a model and other sensors, a scaling factor fault is compensated by estimating a fault scaling factor, a signal offset fault is corrected by estimating a fault offset, and for a fault with changed dynamic characteristics, the fault dynamic parameter estimation is multiplied by a fault inverse transfer function to compensate.

When a certain variable-speed variable-pitch horizontal shaft double-fed fan fails, a fault detection and fault isolation method based on collective value observation is adopted to judge that a pitch angle sensor has a stuck fault, and air is mixed into hydraulic oil of a variable-pitch actuator to cause the change of the dynamic characteristics of a variable-pitch system to cause a performance degradation fault. The virtual device is used for realizing fault tolerance processing, power fluctuation and cabin acceleration fluctuation in a full-load state after a fault caused by air mixed in the variable pitch device exist are compared and analyzed, and the result shows that after the fault tolerance processing is adopted, the power and vibration fluctuation is reduced, and the fault tolerance scheme is good in effect.