A primary frequency modulation method based on the health status of the fan

Abstract

The present invention discloses a primary frequency modulation method based on the health status of a fan. A fault assessment is performed based on the sensor data obtained from each part of the fan system, including sequential fault detection and fault degree value determination; for health status monitoring, the influence of each factor on the fan is quantified as a weight factor based on the fault degree value combined with other factors through the hierarchical analysis method, and the health status of the fan is characterized by the weight factor; all fans are frequency modulated based on the weight factor and fault degree value of each fan. The present invention combines a new neural network algorithm, and for the first time proposes to combine the health status level of each fan to participate in the primary frequency modulation of the system, and reasonably distributes the output of each fan, thereby slowing down the deterioration of the fan state, extending the service life of the fan, and reducing the maintenance cost of the fan.

Description

Technical Field

[0001] The invention relates to a primary frequency modulation method for a power grid, and specifically designs a primary frequency modulation method based on the health status of a wind turbine. Background Art

[0002] In the past few decades, the wind power industry has achieved tremendous development both in China and around the world. With the increasing penetration of wind energy, the reliability of the power grid affected by wind energy is a challenge. Therefore, wind farms must provide corresponding active power to support the frequency in the power grid. In order to meet the requirements of primary frequency regulation, the operation requirements of modern wind farms are more like traditional power plants. In this case, it is very necessary to maintain the active power support of wind turbines. Wind turbines are often installed under harsh conditions, especially offshore wind farms, which brings challenges to their operation and maintenance. Wind turbines are also a complex power generation system. The health of each component will affect the working efficiency of the entire system, affect the power output of the wind turbine, and then affect its primary frequency regulation strategy. This plays a very important role in participating in primary frequency regulation after the wind farm is connected to the grid.

[0003] In order to achieve optimal control of wind farm power support, people have proposed relevant control strategies: due to different wind speeds, the droop control coefficient is adjusted according to the available power margin; or different weight coefficients are assigned according to the wind speed range to control wind turbines at different wind speeds to provide different power support. However, few works consider the impact of wind turbine health on wind farm power dispatching control.

[0004] Under the general control strategy, each wind turbine in the wind farm is assumed to have the same health status. Therefore, the usual primary frequency regulation active power control algorithm is based on the average power allocation method or the fixed ratio allocation method to participate in the system primary frequency regulation. For unhealthy wind turbines, the power target allocated is the same as that of healthy wind turbines. Such support pressure may aggravate the deterioration of unhealthy wind turbines, accelerate the shortening of their service life, and even cause failures. Summary of the invention

[0005] Based on the problem that there are few works considering the impact of wind turbine health status on wind farm power dispatching and control, the present invention is to design a primary frequency modulation method based on wind turbine health status. The method can perform real-time fault diagnosis and status evaluation on the wind turbine through the collected wind turbine information, and designs the primary frequency modulation strategy of the wind turbine according to the evaluation results, which can effectively improve the quality of wind turbine grid-connected power while slowing down the deterioration of the wind turbine and reducing maintenance costs.

[0006] The method of the present invention takes into account the health status level of the fan and can effectively alleviate the problem of fan deterioration.

[0007] The technical solution of the present invention is:

[0008] S1. Performing fault assessment based on the sensor data of each part of the fan system, including sequentially performing fault detection and determining the fault severity value;

[0009] The fault detection is to use the long short-term memory network (LSTM) algorithm to process the historical sensor data obtained in real time to obtain the predicted value of the future sensor data, and compare it with the actual measured value of the future sensor data to obtain the difference, and judge whether a fault occurs according to the size of the difference;

[0010] The fault degree value is determined by taking the difference value obtained in real time during fault detection as the observation value, and inputting the average value of the observation value over a period of time into the Bonferroni interval method to obtain the fault degree value for accurately quantifying the fault degree.

[0011] The fault degree value represents the fault degree of the fan. If the difference is zero, the fault degree value is taken as zero.

[0012] The sensor data of each part of the fan system include the fan drive shaft temperature, the drive shaft bearing temperature, blade wear degree, the gear box temperature between the fan and the generator, and the generator coil temperature, but are not limited thereto.

[0013] S2. For health status monitoring, based on the fault degree value obtained in S1, combined with factors such as wind turbine life, maintenance costs and maintenance time of each part of the wind turbine, temperature, meteorological conditions, and sea conditions of the wind turbine environment, the influence of each factor on the wind turbine is quantified into weight factors through the hierarchical analysis method, and the weight factors are used to represent the health status of the wind turbine;

[0014] S3. Perform frequency adjustment on all fans according to the weight factor and fault severity value of each fan.

[0015] The S3 is specifically:

[0016] First, the total amount of active power provided by the wind turbine is obtained through primary frequency modulation, and based on the weight factors of the wind turbines, the total amount of active power is distributed to the size of the active power output of each wind turbine according to the proportion of the corresponding weight factors;

[0017] Secondly, according to the different fault locations of the wind turbine, different wind turbine dynamic power control methods and constraints are set to control the wind turbine;

[0018] In response to the fault state of blade wear, power control is performed by controlling the pitch angle, and a dynamic constraint boundary is set for the wind turbine speed. The constraint range is adjusted in real time according to the health status of the wind turbine within a specified time interval; if the blade wear fault occurs, the pitch angle is controlled to become larger, the speed becomes smaller, and the power becomes smaller.

[0019] For the fault state of transmission shaft deformation or bearing wear, power control is performed by controlling torque, and a dynamic constraint boundary is set for the torque. The constraint range is adjusted in real time according to the health status of the wind turbine within a specified time interval. If the transmission shaft is deformed or the bearing is worn, the control torque is reduced and the power becomes smaller.

[0020] like Figure 6 As shown, in S3, the control is specifically carried out in the following manner: the fan outputs the speed of the generator and the torque of the transmission shaft in real time through the sensor, and the speed error is obtained by subtracting the speed of the generator from the rated speed of the control input, and then inputting it into the pitch angle PI controller to process the output of the fan pitch angle, and applying the fan pitch angle to the fan for pitch angle control; at the same time, the speed of the generator is input into the power-torque controller to process the ideal torque, and the torque of the transmission shaft is input into the transmission system damper to process the speed loss, and the torque of the generator is obtained by subtracting the speed loss from the ideal torque, and the torque of the generator is applied to the fan for torque control of the generator. In this way, dual coordinated control of speed and torque is achieved.

[0021] The present invention combines a new neural network algorithm and for the first time proposes to combine the health status level of each fan to participate in the primary frequency modulation of the system, so as to reasonably distribute the output of each fan, slow down the deterioration of the fan state, extend the service life of the fan, and reduce the maintenance cost of the fan.

[0022] The beneficial effects of the present invention are:

[0023] After the grid frequency fluctuates, if the grid frequency deviation or frequency change rate exceeds the frequency regulation dead zone, the wind turbine participates in the primary frequency regulation, avoiding unnecessary operation of the wind turbine when the grid frequency fluctuates slightly; and considering the health status level of each wind turbine, the size of the active power output increment of each wind turbine is reasonably allocated;

[0024] At the same time, the fault location of the fan is taken into consideration, and the control method of the faulty fan is reasonably adjusted. On the basis of meeting the primary frequency regulation and improving the quality of the wind turbine grid-connected power, the deterioration of the fan fault state is alleviated and the maintenance cost of the fan is reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 It is the overall flow chart of the present invention;

[0026] Figure 2 is a schematic diagram of the blower fault degree assessment of the present invention;

[0027] Figure 3 is a structural diagram of the fan health status monitoring of the present invention;

[0028] Figure 4 It is the power characteristic curve of the fan under load reduction operation;

[0029] Figure 5 This is a primary frequency modulation principle diagram of the present invention;

[0030] Figure 6 This is the schematic diagram of the fan load reduction operation. DETAILED DESCRIPTION

[0031] The technical solution of the present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0032] like Figure 1 As shown, the specific implementation of the present invention mainly includes three parts / steps: fault degree assessment, health status monitoring and primary frequency modulation strategy design.

[0033] S1, such as Figure 2 As shown, the wind turbine fault degree assessment includes fault detection and fault degree value determination.

[0034] Among them, fault detection is performed on each fan through the LSTM algorithm. The processing method is: for the status of the mth fan, the sensor data obtained at time t is input into the LSTM algorithm model for processing, and the predicted data at time t+1 is output to represent the predicted value of the fan, which is compared with the measured value at time t+1, and whether the fan is faulty is determined by the size of the difference.

[0035] The fault degree value is determined by the Bonferroni interval method, which quantifies the fault degree and improves the accuracy. The processing method is: the difference between the predicted value and the measured value obtained by the previous step of fault detection is processed, and the quantitative fault degree value of the fan is obtained according to the confidence level.

[0036] S2, such as Figure 3 As shown in the figure, the wind turbine health status monitoring adopts the hierarchical analysis method to model the algorithm. In the specific implementation, the fault degree value of the wind turbine obtained in the previous step is used as one of the inputs, and other inputs include but are not limited to the wind turbine life, maintenance costs and maintenance time of each part of the wind turbine, temperature, meteorological conditions and sea conditions.

[0037] Different factors have different effects on the health status of the wind turbine. An influence degree matrix of different factors is established and processed as weight factors using the analytic hierarchy process to characterize the health status level of the wind turbine.

[0038] S3, such as Figure 4 As shown, the primary frequency modulation of the fans is based on that all fans are operating in the load reduction operation mode, and different sizes of power margins are reserved according to actual needs.

[0039] In the specific implementation, in the case of a sudden drop in grid frequency, the virtual inertia response and the primary frequency regulation droop characteristics are combined to obtain the active power output increment required for the wind turbine to participate in the primary frequency regulation.

[0040] like Figure 5 As shown in Figure 1, for each wind turbine, the active power output is allocated according to the corresponding weight coefficient based on its health status level. Assume that there are N wind turbines, of which M are unhealthy wind turbines, and the corresponding health status level is S i (t) and 0≤S i ≤1, 0 represents a complete failure and 1 represents a completely healthy state. When the frequency deviates by Δf, the active power increment ΔP is obtained. in and ΔP d For unhealthy wind turbines, the optimized active power reference value for

[0041]

[0042] For healthy fans, the optimized active power reference value is for:

[0043]

[0044] At the same time, if Figure 6 As shown, according to different fault locations of the wind turbine, corresponding dynamic power control methods and constraints of the wind turbine are set.

[0045] In the specific implementation, for the fault state of wind turbine blade wear, power control is performed by controlling the pitch angle, and a dynamic constraint boundary is set for the wind turbine speed. The constraint range is adjusted in real time according to the health status of the wind turbine within a specified time interval;

[0046] For fault conditions such as drive shaft deformation or bearing wear, power control is performed by controlling the torque, and a dynamic constraint boundary is set for the torque. The constraint range is adjusted in real time according to the health status of the wind turbine within a specified time interval.

[0047] To sum up, the present invention, on the basis of meeting the primary frequency regulation requirements, takes into account the health status level of each fan, reasonably allocates the size of the active power output increment of each fan, and at the same time takes into account the fault location of the fan, reasonably adjusts the control method of the faulty fan, alleviates the deterioration of the fan fault state, and reduces the maintenance cost of the fan.

Claims

1. A primary frequency modulation method based on the health status of a fan, characterized in that: S1. Performing fault assessment based on the sensor data of each part of the fan system, including sequentially performing fault detection and determining the fault severity value; In S1, the fault detection is to use the long short-term memory network algorithm to process the sensor data obtained in real time to obtain the predicted value of the future sensor data, and compare it with the actual measured value of the future sensor data to obtain the difference, and judge whether a fault occurs according to the difference; In S1, the fault degree value is determined by taking the difference value obtained in real time during fault detection as the observation value, and inputting the average value of the observation value into the Bonferroni interval method to obtain the fault degree value for accurately quantifying the fault degree; The fault degree value represents the fault degree of the fan. If the difference is zero, the fault degree value is taken as zero. The sensor data of each part of the fan system includes the fan drive shaft temperature, the drive shaft bearing temperature, the blade wear degree, the gear box temperature between the fan and the generator, and the generator coil temperature; S2. For health status monitoring, based on the fault degree value obtained in S1, combined with the wind turbine life, maintenance costs and maintenance time of each part of the wind turbine, meteorological conditions, and sea conditions, the influence of each factor on the wind turbine is quantified into weight factors through the hierarchical analysis method, and the weight factors are used to characterize the health status of the wind turbine; S3, performing frequency modulation on all fans according to the weight factor and fault degree value of each fan; The S3 is specifically: First, the total active power provided by all wind turbines is obtained through a frequency modulation, and the total active power is distributed to the size of the active power output of each wind turbine according to the proportion of the corresponding weight factor; After the grid frequency fluctuates, if the grid frequency deviation or frequency change rate exceeds the frequency regulation dead zone, the wind turbine participates in primary frequency regulation; For unhealthy wind turbines, the optimized active power reference value for: Where N is the number of fans, S i (t) represents the health status level corresponding to the unhealthy fan, 0≤S i (t)≤1, 0 represents complete failure and 1 represents complete health, Indicates the active power reference value of the unhealthy fan before optimization, ΔP in , ΔP d They respectively represent the active power increment obtained based on the virtual inertia response and the primary frequency modulation droop characteristics when the frequency deviates; For healthy fans, the optimized active power reference value is for: Where M represents the number of unhealthy fans. Indicates the active power reference value of the healthy fan before optimization; Secondly, according to the different fault locations of the wind turbine, different wind turbine dynamic power control methods and constraints are set to control the wind turbine; The primary frequency regulation of the wind turbine is based on the fact that all wind turbines are operating in the load reduction mode, and different power margins are reserved according to actual needs; specifically, for the case where the grid frequency suddenly drops, the active power output increment required for the wind turbine to participate in the primary frequency regulation is obtained by combining the virtual inertia response and the primary frequency regulation droop characteristics; In response to blade wear faults, power control is performed by controlling the pitch angle, and a dynamic constraint boundary is set for the wind turbine speed. The constraint range is adjusted in real time according to the health status of the wind turbine within a specified time interval; For fault conditions such as drive shaft deformation or bearing wear, power control is performed by controlling the torque, and a dynamic constraint boundary is set for the torque. The constraint range is adjusted in real time according to the health status of the wind turbine within a specified time interval.

2. A primary frequency modulation method based on the health status of a fan according to claim 1, characterized in that: In the S3, the control is specifically carried out in the following manner: the wind turbine outputs the speed of the generator and the torque of the transmission shaft in real time, and the speed error is obtained by subtracting the speed of the generator from the rated speed, which is then input into the pitch angle PI controller to process and output the pitch angle of the wind turbine, and the pitch angle of the wind turbine is applied to the wind turbine for pitch angle control; at the same time, the speed of the generator is input into the power-torque controller to obtain the ideal torque, the torque of the transmission shaft is input into the transmission system damper to obtain the speed loss, the torque of the generator is obtained by subtracting the speed loss from the ideal torque, and the torque of the generator is applied to the wind turbine for generator torque control.

Images