Redundant detection method and system for rotor position of a variable pitch servo motor

Abstract

The embodiment of the application provides a kind of redundant detection method and system of variable pitch servo motor rotor position, it is through Kalman filtering-state prediction, utilize motor model to predict a theoretical rotor position in real time.Then, the actual current measurement of first and second position sensor is compared with this model prediction value respectively, and is generated by innovation calculation and statistical characteristic analysis for each sensor a quantification fault indicator that can reflect its data reliability in real time, namely normalized innovation square.When the readings of two physical sensors conflict, the system no longer blindly stops, but carries out fault diagnosis according to the fault indicators of each sensor, so as to accurately determine the faulty sensor.This provides a clear and reliable decision basis for the system, realizes the online rapid isolation of fault, and thus guarantees the continuous operation of the control system and the availability of the equipment without sacrificing safety.

Description

Technical Field

[0001] This invention relates to the field of new energy wind power generation, specifically to a method and system for detecting rotor position redundancy of a pitch servo motor. Background Technology

[0002] In fields like wind power generation, where safety and reliability are paramount, the pitch system is one of the core actuators ensuring the safe and stable operation of wind turbine generators. This system optimizes energy capture efficiency and controls aerodynamic loads by precisely adjusting the blade pitch angle. Especially in high wind speeds or emergency situations, the pitch system must be able to quickly and accurately feather the blades to a safe angle to prevent overspeed runaway, i.e., a turbine overrun. The servo motor, as the drive unit of the pitch system, provides the rotor position signal, which is fundamental to achieving precise angle closed-loop control. Therefore, a malfunction in the sensor used to measure the rotor position could lead to the control system acquiring incorrect blade angle information, resulting in flawed control decisions and potentially catastrophic consequences.

[0003] Currently, the mainstream technical solution for achieving redundant detection is to use dual-sensor hardware redundancy. This solution involves installing two physical position sensors in parallel, and the controller compares their measurements in real time. When the deviation exceeds a preset threshold, the system determines a fault and triggers a shutdown protection mechanism. However, while this simple deviation comparison strategy ensures basic safety, it exposes inherent flaws in practical applications. The core problem is that when the readings of the two sensors conflict, the control system can only know that they are inconsistent, but faces a thorny dilemma of choosing one over the other. Because the system lacks an absolutely reliable third-party arbitrator as a truth reference, the controller cannot accurately and quickly determine which sensor has failed based solely on two contradictory signal sources. This lack of decision-making basis forces existing systems to adopt the most conservative approach—immediate shutdown and alarm, awaiting manual intervention. While this strategy is safe, frequent unnecessary shutdowns severely impact the availability of wind turbines, resulting in significant power generation losses and high maintenance costs.

[0004] Therefore, an optimized rotor position redundancy detection scheme for pitch servo motors is desired. Summary of the Invention

[0005] The present invention aims to solve at least one of the technical problems existing in the prior art, and provides a method and system for detecting rotor position redundancy of a pitch servo motor.

[0006] In a first aspect, embodiments of the present invention provide a method for detecting rotor position redundancy of a pitch servo motor, comprising: Acquire the current measured value of the first position acquired by the first position sensor and the current measured value of the second position acquired by the second position sensor; Kalman filtering-state prediction is performed on the state estimate, control input, and covariance matrix of the previous time step to obtain the prior state estimate and prior covariance matrix of the current time step. Based on the prior state estimation at the current moment, innovation calculation and statistical feature analysis are performed on the current measurement values ​​of the first position and the second position to obtain the innovation of the first position sensor, the innovation of the second position sensor, the normalized innovation square of the first position sensor and the normalized innovation square of the second position sensor. Fault diagnosis is performed based on the normalized squared innovation of the first position sensor and the normalized squared innovation of the second position sensor to obtain fault indicators.

[0007] Secondly, embodiments of the present invention provide a rotor position redundancy detection system for a pitch servo motor, comprising: The data acquisition module is used to acquire the current measurement value of the first position collected by the first position sensor and the current measurement value of the second position collected by the second position sensor; The Kalman filter-state prediction module is used to perform Kalman filter-state prediction on the state estimate, control input, and covariance matrix of the previous time step to obtain the prior state estimate and prior covariance matrix of the current time step. The innovation calculation and statistical feature analysis module is used to perform innovation calculation and statistical feature analysis on the current measurement value of the first position and the current measurement value of the second position based on the prior state estimation at the current time, so as to obtain the innovation of the first position sensor, the innovation of the second position sensor, the normalized innovation square of the first position sensor and the normalized innovation square of the second position sensor. The fault diagnosis module is used to perform fault diagnosis based on the normalized square of the first position sensor and the normalized square of the second position sensor to obtain a fault indicator.

[0008] Compared with existing technologies, this invention proposes a method for detecting rotor position redundancy in a pitch servo motor. It constructs a virtual sensor based on a servo motor dynamics model as a neutral arbitration reference. This method uses Kalman filtering and state prediction to predict a theoretical rotor position in real time using the motor model. Then, the actual current measurements of the first and second position sensors are independently compared with this model prediction. Through innovation calculation and statistical feature analysis, a quantitative fault index (normalized innovation squared) is generated for each sensor, reflecting its data reliability in real time. When the readings of two physical sensors conflict, the system no longer blindly shuts down but performs fault diagnosis based on their respective fault indices, thus accurately identifying the faulty sensor. This provides the system with a clear and reliable decision-making basis, effectively solving the problem of traditional redundancy schemes being unable to make decisions due to the lack of an arbitration reference when faced with inconsistent signals. It achieves rapid online fault isolation, thereby ensuring the continuous operation of the control system and the availability of equipment without sacrificing safety. Attached Figure Description

[0009] The above and other objects, features, and advantages of the present invention will become more apparent from the more detailed description of the embodiments of the invention in conjunction with the accompanying drawings. The drawings are provided to further illustrate the embodiments of the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings, the same reference numerals generally represent the same parts or steps.

[0010] Figure 1 This is a flowchart of a rotor position redundancy detection method for a pitch servo motor according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the data flow in the rotor position redundancy detection method for a pitch servo motor according to an embodiment of the present invention. Figure 3 The flowchart illustrates the process of performing Kalman filtering-state prediction on the state estimate, control input, and covariance matrix of the previous moment in the rotor position redundancy detection method for a pitch servo motor according to an embodiment of the present invention to obtain the prior state estimate and prior covariance matrix of the current moment. Figure 4 The flowchart illustrates the prior state estimation based on the current moment in the rotor position redundancy detection method for a pitch servo motor according to an embodiment of the present invention, which involves performing innovation calculation and statistical feature analysis on the current measured values ​​of the first position and the second position to obtain the innovation of the first position sensor, the innovation of the second position sensor, the normalized innovation square of the first position sensor, and the normalized innovation square of the second position sensor. Figure 5This is a block diagram of a rotor position redundancy detection system for a pitch servo motor according to an embodiment of the present invention. Detailed Implementation

[0011] To enable those skilled in the art to better understand the technical solutions of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the described embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0012] Unless otherwise specifically stated, the technical or scientific terms used in the embodiments of this invention should be understood in their ordinary meaning as understood by one of ordinary skill in the art to which this invention pertains. The terms "comprising" or "including," as used in the embodiments of this invention, do not limit the shapes, numbers, steps, actions, operations, components, elements, and / or groups thereof mentioned, nor do they exclude the appearance or addition of one or more other different shapes, numbers, steps, actions, operations, components, elements, and / or groups thereof, or the inclusion of these.

[0013] Unless otherwise specifically stated, the relative arrangement, numerical expressions, and values ​​of the components and steps described in these embodiments do not limit the scope of the invention. It should also be understood that, for ease of description, the dimensions of the various parts shown in the drawings are not drawn to actual scale, and techniques, methods, and apparatus known to those skilled in the art may not be discussed in detail; however, where appropriate, the illustrated techniques, methods, and apparatus should be considered part of the specification. In all the examples shown and discussed herein, any other specific example may have different values. It should be noted that similar symbols and letters in the following figures denote similar items; therefore, once an item is defined in one figure, it need not be further discussed in subsequent figures.

[0014] In the description of the embodiments of the present invention, the terms "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In the embodiments of the present invention, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in a suitable manner in any one or more embodiments or examples. Furthermore, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in the embodiments of the present invention, as well as the features of different embodiments or examples.

[0015] Hereinafter, exemplary embodiments according to the present invention will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of the present invention, and not all embodiments of the present invention. It should be understood that the present invention is not limited to the exemplary embodiments described herein.

[0016] Existing rotor position redundancy detection schemes for pitch servo motors typically employ dual hardware sensors for deviation comparison. However, when the readings of the two sensors are inconsistent, the system lacks a reliable arbitration basis to determine the fault source, often resorting to a conservative shutdown strategy, which severely impacts equipment availability. Therefore, this application proposes a method for detecting rotor position redundancy in pitch servo motors. Specifically, this method first acquires the current measurement values ​​collected by the first and second position sensors. Simultaneously, it performs Kalman filtering-state prediction on the previous state estimate (including servo motor position and speed), the previous control input, and the covariance matrix. That is, it predicts the prior state estimate and prior covariance matrix at the current moment using a simplified dynamic model of the servo motor. Subsequently, based on this prior state estimate, it performs innovation calculation and statistical feature analysis on the current measurement values ​​of the two sensors. This analysis process includes: calculating the deviation of the two measurement values ​​from the prior state estimate to obtain the respective sensor innovations; quantifying the uncertainty of the innovations to obtain the innovation covariance; and finally calculating the normalized squared innovations of the first and second position sensors, which serve as a fault indicator for quantifying the health status. Finally, fault diagnosis is performed based on these two normalized squared innovation values. When a sensor's index is significantly abnormal, it can be identified as a fault source, and a corresponding fault flag is generated. In this way, this scheme provides a clear decision-making basis for the system when facing sensor signal conflicts, solves the technical problem of the traditional two-choice approach, and ensures the continuity of control.

[0017] Figure 1 This is a flowchart of a rotor position redundancy detection method for a pitch servo motor according to an embodiment of the present invention. Figure 2 This is a schematic diagram of the data flow in the rotor position redundancy detection method for a pitch servo motor according to an embodiment of the present invention. Figure 1 and Figure 2As shown, the method and system for detecting rotor position redundancy of a pitch servo motor according to an embodiment of the present invention include the following steps: S100, acquiring the current measurement value of the first position collected by the first position sensor and the current measurement value of the second position collected by the second position sensor; S200, performing Kalman filtering-state prediction on the state estimate of the previous moment, the control input of the previous moment, and the covariance matrix of the previous moment to obtain the prior state estimate of the current moment and the prior covariance matrix of the current moment; S300, based on the prior state estimate of the current moment, performing innovation calculation and statistical feature analysis on the current measurement value of the first position and the current measurement value of the second position to obtain the innovation of the first position sensor, the innovation of the second position sensor, the normalized innovation square of the first position sensor, and the normalized innovation square of the second position sensor; S400, performing fault diagnosis based on the normalized innovation square of the first position sensor and the normalized innovation square of the second position sensor to obtain a fault flag.

[0018] Specifically, in step S100, the current measurement value of the first position acquired by the first position sensor and the current measurement value of the second position acquired by the second position sensor are obtained. It should be understood that, because the closed-loop control of the pitch servo motor has high requirements for the real-time performance and accuracy of rotor position feedback, and the failure of any single sensor may affect the safety and stability of the control system, a redundant monitoring mechanism is needed. Therefore, in the technical solution of this application, the current measurement value of the first position acquired by the first position sensor and the current measurement value of the second position acquired by the second position sensor are obtained to provide two parallel and independent physical measurement data sources for subsequent fault diagnosis and data fusion. This establishes the necessary data foundation for real-time determination of sensor operating status and output of reliable position estimates, thereby improving the fault tolerance of the entire control system.

[0019] More specifically, in a particular example of this application, the process of acquiring the current measured values ​​of two positions is a hardware event strictly synchronized with the execution cycle of control loops such as the current loop and speed loop in the servo controller. At the beginning of each digital control cycle, a synchronization trigger pulse is generated. This pulse is sent to a dedicated peripheral interface connected to the two position sensors. For example, for the quadrature encoder serving as the first position sensor, the trigger pulse latches the hardware register value of the Quadrature Encoder Pulse (QEP) counting module; for the resolver serving as the second position sensor, the trigger pulse activates a multi-channel synchronous analog-to-digital converter (ADC) to sample the SIN and COS analog signals of the resolver. The latched count value or the sampled and converted digital signal is then processed by the hardware or low-level driver within microseconds, converting it into a high-precision angle value in a uniform-scale format, such as a 32-bit floating-point number. These two angle values ​​are timestamped to ensure that they reflect the rotor position at the same sampling time k, ultimately forming the current measurement value of the first position and the current measurement value of the second position, and updating them to a specific variable address in memory for immediate use by subsequent algorithms such as state prediction and information calculation.

[0020] Specifically, in step S200, Kalman filtering-state prediction is performed on the state estimate, control input, and covariance matrix of the previous time step to obtain the prior state estimate and prior covariance matrix of the current time step. In particular, the state estimate of the previous time step includes the servo motor position and servo motor speed. It should be understood that simply comparing the measurements from two physical sensors cannot provide an objective benchmark for determining the truth when they conflict, and the motion state of the servo motor follows predictable physical laws. Therefore, in the technical solution of this application, Kalman filtering-state prediction is further performed on the state estimate, control input, and covariance matrix of the previous time step to obtain the prior state estimate and prior covariance matrix of the current time step, thereby generating an independent state prediction value based on the motor dynamics model that does not depend on any single physical sensor. This provides a neutral reference standard for subsequent information calculations. This standard serves as a virtual arbitration basis and is the foundation for achieving independent health assessments of the two physical sensors and resolving data conflict issues.

[0021] Figure 3 This is a flowchart illustrating the Kalman filtering-state prediction process used in the rotor position redundancy detection method for a pitch servo motor according to an embodiment of the present invention to obtain the prior state estimate and the prior covariance matrix at the current moment, based on the previous moment's state estimate, control input, and covariance matrix. Figure 3 As shown, step S200 includes: S210, inputting the state estimate of the previous moment and the control input of the previous moment into the simplified dynamic model of the servo motor to obtain the prior state estimate of the current moment; S220, updating the covariance matrix of the previous moment based on the process noise covariance matrix to obtain the prior covariance matrix of the current moment.

[0022] In step S210, the state estimate from the previous moment and the control input from the previous moment are input into the simplified dynamic model of the servo motor to obtain the prior state estimate for the current moment. This is expressed by the following formula:

[0023] in, Here is the state transition matrix. For the state estimation of the previous moment, To control the input matrix, This is the control input from the previous moment.

[0024] It is understandable that to generate an arbitration reference independent of physical sensors, a method is needed to deduce the motor's state based on its own operating laws. This method must simultaneously consider the motor's inertial motion tendency and the control force applied to it. Therefore, in the technical solution of this application, the state estimate and control input of the previous moment are further input into the simplified dynamic model of the servo motor to obtain the prior state estimate of the current moment. Specifically, the state estimate of the previous moment is processed by the state transition matrix F to calculate the state component of the motor naturally evolving from the previous moment's state to the current moment due to its own inertia without external control. At the same time, the control input applied to the motor in the previous moment, such as torque current, is processed by the control input matrix B to calculate the contribution component of the control input to the change in motor state. Finally, these two components are added to calculate the prior state estimate of the current moment k, which includes the predicted values ​​of the current rotor position and speed. In this way, a specific motor state vector based purely on model prediction can be generated, which serves as a benchmark for measuring the accuracy of physical sensor measurements in subsequent steps.

[0025] In step S220, the covariance matrix of the previous time step is updated based on the process noise covariance matrix to obtain the prior covariance matrix of the current time step. This is expressed by the following formula:

[0026] in, Here is the state transition matrix. Let be the covariance matrix of the previous time step. Let be the process noise covariance matrix.

[0027] It is understandable that, since the simplified dynamic model of a servo motor cannot completely and accurately describe the real physical process, its predicted state estimate inevitably carries a certain degree of uncertainty, and this uncertainty evolves and accumulates over time. Therefore, in the technical solution of this application, the covariance matrix of the previous time step is further updated based on the process noise covariance matrix to obtain the prior covariance matrix of the current time step. Specifically, this update process involves using the state transition matrix F and its transpose... The covariance matrix of the previous time step A transformation is performed to calculate how the uncertainty from the previous time step propagates and evolves to the current time step according to system dynamics. Then, a process noise covariance matrix Q is superimposed on this propagated uncertainty. This matrix Q is used to quantify the new uncertainties introduced in the current control cycle by factors such as model simplification, unmodeled dynamics, or external disturbances, thereby calculating the prior covariance matrix at the current time k. This provides a quantitative assessment of the reliability of state prediction, which is a necessary prerequisite for subsequent information covariance calculations and statistical judgment of sensor faults.

[0028] Specifically, in step S300, based on the prior state estimate at the current moment, innovation calculation and statistical feature analysis are performed on the current measured values ​​of the first position and the second position to obtain the innovation of the first position sensor, the innovation of the second position sensor, the normalized squared innovation of the first position sensor, and the normalized squared innovation of the second position sensor. It should be understood that since the actual measured values ​​of the physical sensors and the independent predicted values ​​of the dynamic model have been obtained, a standardized method is needed to quantify the significance of the deviation between the two to distinguish between normal random fluctuations and real sensor faults. Therefore, in the technical solution of this application, innovation calculation and statistical feature analysis are further performed on the current measured values ​​of the first position and the second position based on the prior state estimate at the current moment to obtain the innovation of the first position sensor, the innovation of the second position sensor, the normalized squared innovation of the first position sensor, and the normalized squared innovation of the second position sensor. This converts the original deviation between each sensor measured value and the model predicted value into a dimensionless anomaly metric with clear statistical significance. In this way, two parallel quantitative judgment criteria that can be directly compared for the final fault diagnosis step can be provided, thus providing a decisive input signal for accurately identifying specific faulty sensors and resolving redundancy conflicts.

[0029] Figure 4This document describes a flowchart illustrating the prior state estimation based on the current moment in the rotor position redundancy detection method for a pitch servo motor according to an embodiment of the present invention. It describes the process of performing innovation calculation and statistical feature analysis on the current measured values ​​of the first and second positions to obtain the innovation of the first position sensor, the innovation of the second position sensor, the normalized squared innovation of the first position sensor, and the normalized squared innovation of the second position sensor. (See attached flowchart.) Figure 4 As shown, step S300 includes: S310, calculating the deviations of the current measurement value of the first position and the current measurement value of the second position relative to the prior state estimate at the current time to obtain the innovation of the first position sensor and the innovation of the second position sensor; S320, based on the prior covariance matrix at the current time, quantifying the innovation uncertainty of the measurement noise covariance of the first position sensor and the measurement noise covariance of the second position sensor to obtain the innovation covariance of the first position sensor and the innovation covariance of the second position sensor; S330, calculating a normalized fault index based on the innovation covariance of the first position sensor, the innovation covariance of the second position sensor, the innovation of the first position sensor and the innovation of the second position sensor to obtain the normalized innovation square of the first position sensor and the normalized innovation square of the second position sensor.

[0030] In step S310, the deviations of the current measured values ​​of the first position and the second position relative to the prior state estimate at the current time are calculated to obtain the first position sensor information and the second position sensor information. This is expressed by the following formula:

[0031]

[0032]

[0033] in, For the measurement matrix, For the estimation of the prior state at the current moment, and These are the current measurement values ​​for the first position and the second position, respectively.

[0034] It is understandable that, since the state prediction result of Kalman filtering is a multi-dimensional state vector containing position, velocity, and other dimensions, while the physical sensor provides a single position measurement value, the two cannot be directly compared. It is necessary to first transform the predicted state vector to the same dimensional space as the measured value. Therefore, in the technical solution of this application, the deviations of the current measurement value of the first position and the current measurement value of the second position relative to the prior state estimate at the current time are further calculated to obtain the first position sensor information and the second position sensor information. This calculation first extracts the predicted position component from the prior state estimate at the current time through the measurement matrix H. Then, the predicted position component is subtracted from the current measurement value of the first position to obtain the first position sensor information. Similarly, the predicted position component is subtracted from the current measurement value of the second position to obtain the second position sensor information. This allows for the independent quantification of the original difference between the measured value and the model prediction value for each sensor. In this way, two deviation signals with clear physical meaning can be generated. These two signals are the direct input and basis for subsequent normalization processing and statistical feature analysis.

[0035] In step S320, based on the prior covariance matrix at the current moment, the innovation uncertainty of the first position sensor measurement noise covariance and the second position sensor measurement noise covariance is quantified to obtain the innovation covariance of the first position sensor and the innovation covariance of the second position sensor. It should be understood that since the innovation obtained in the previous calculation step is a raw deviation with physical units, its absolute magnitude cannot be directly used for fault judgment. Whether a deviation value is abnormal depends on the expected fluctuation range of that deviation, which is jointly determined by the uncertainty of the model prediction and the uncertainty of the sensor's own measurement noise. Therefore, in the technical solution of this application, the innovation uncertainty of the first position sensor measurement noise covariance and the second position sensor measurement noise covariance is further quantified based on the prior covariance matrix at the current moment to obtain the innovation covariance of the first position sensor and the innovation covariance of the second position sensor, thereby calculating a variance value that quantifies the total uncertainty of the innovation signal of each sensor. This provides a key scaling factor for subsequent normalization processing, which is the basis for constructing a fault judgment index with consistent statistical properties.

[0036] More specifically, in a concrete example of this application, the quantification of innovation uncertainty involves a series of numerical calculations performed in the digital controller. First, the prior covariance matrix, measurement matrix, and pre-calibrated measurement noise variances of the first and second position sensors are retrieved from memory. Next, matrix operations are performed, multiplying the measurement matrix, the prior covariance matrix, and the transpose of the measurement matrix. This operation projects the uncertainty of the state prediction from the state space to the measurement space, yielding a scalar value representing the uncertainty of the model's predicted position itself. Subsequently, this scalar value is added to the measurement noise variances of the first and second position sensors, respectively, resulting in two independent scalar results: the innovation covariance of the first and second position sensors. These two calculation results are stored in memory for direct use in the normalized fault index calculation step.

[0037] In step S330, a normalized fault index is calculated based on the covariance of the first position sensor's innovation, the covariance of the second position sensor's innovation, the innovation of the first position sensor, and the innovation of the second position sensor to obtain the normalized squared innovation of the first position sensor and the normalized squared innovation of the second position sensor. This is expressed by the following formula:

[0038]

[0039] in, Normalized squared information of the first position sensor, For the normalized squared information of the second position sensor, The inverse of the information covariance of the first position sensor. The inverse of the information covariance of the second position sensor. For the transpose of the information from the first position sensor, This is a transpose of the information from the second position sensor.

[0040] It is understandable that, since the original innovation and innovation covariance obtained from the previous steps both have physical units and their numerical ranges are not fixed, it is not conducive to setting a unified fault judgment threshold with clear statistical confidence. Therefore, it is necessary to combine the two to form a standardized dimensionless index. Thus, in the technical solution of this application, a normalized fault index is further calculated based on the innovation covariance of the first position sensor, the innovation covariance of the second position sensor, the innovation of the first position sensor, and the innovation of the second position sensor to obtain the normalized squared innovation of the first position sensor and the normalized squared innovation of the second position sensor. The calculation process is as follows: for the first position sensor, its innovation value is squared and then divided by its innovation covariance value; for the second position sensor, the same calculation is performed, that is, its innovation value is squared and then divided by its innovation covariance value. This normalizes the original deviation of each sensor through a measure of its own total uncertainty, thereby eliminating the influence of dimensions. In this way, two final fault characteristic indicators can be generated that follow a known statistical distribution when the sensor is working normally. When a sensor fails, its corresponding indicator value will deviate significantly from the normal range, thus providing a clear and reliable decision basis for fault diagnosis that can be directly judged by threshold.

[0041] Specifically, in step S400, fault diagnosis is performed based on the normalized squared innovation of the first position sensor and the normalized squared innovation of the second position sensor to obtain a fault flag. It should be understood that since the previous steps have transformed the complex sensor signal characteristics into two independent health status indicators with clear statistical significance, the final decision logic needs to convert these continuously changing indicator values ​​into a discrete and stable state signal for direct use by the control system to trigger corresponding fault-tolerant actions. Therefore, in the technical solution of this application, fault diagnosis is further performed based on the normalized squared innovation of the first and second position sensors to obtain a fault flag, thereby making a final and continuous judgment on the health status of the two sensors. This generates a digital flag that clearly indicates the system's health status or identifies a specific faulty sensor. This flag is the direct execution basis for resolving sensor redundancy conflicts and achieving fault tolerance in the control system.

[0042] More specifically, in a concrete example of this application, fault diagnosis is implemented as a logical judgment and status update process executed in each control cycle. First, the calculated normalized square of the first position sensor and the normalized square of the second position sensor are compared with a preset fault judgment threshold based on the chi-square distribution information interval. To increase the robustness of the diagnosis and avoid misjudgments caused by transient interference, a fault confirmation counter is set for each sensor. If the normalized square of a sensor exceeds the threshold in the current cycle, its corresponding counter increments; conversely, if it does not exceed the threshold, the counter decrements or is cleared. A sensor is formally determined to be faulty only when the accumulated fault confirmation counter of one sensor reaches a preset duration limit, such as 10 consecutive control cycles. At this time, the fault flag variable is set to a specific status code; for example, setting it to 1 represents a fault in the first position sensor, and setting it to 2 represents a fault in the second position sensor. Once the fault flag is set, it is locked and continuously output to the pitch servo controller and the upper-level monitoring system until a reset command is received.

[0043] In summary, the rotor position redundancy detection method for a pitch servo motor according to an embodiment of the present invention is explained. It uses Kalman filtering and state prediction to predict a theoretical rotor position in real time using a motor model. Then, the actual current measurement values ​​of the first and second position sensors are independently compared with this model prediction value. Through innovation calculation and statistical feature analysis, a quantitative fault index, namely the normalized innovation square, is generated for each sensor, reflecting the reliability of its data in real time. When the readings of two physical sensors conflict, the system no longer blindly shuts down but performs fault diagnosis based on their respective fault indices, thereby accurately identifying the faulty sensor. This provides the system with a clear and reliable decision-making basis, effectively solving the problem of traditional redundancy schemes being unable to make a decision due to the lack of arbitration reference when facing inconsistent signals. It achieves rapid online fault isolation, thus ensuring the continuous operation of the control system and the availability of equipment without sacrificing safety.

[0044] Furthermore, a rotor position redundancy detection system for a pitch servo motor is also provided.

[0045] Figure 5 This is a block diagram of a rotor position redundancy detection system for a pitch servo motor according to an embodiment of the present invention. Figure 5As shown, the rotor position redundancy detection system 100 for a pitch servo motor according to an embodiment of the present invention includes: a data acquisition module 110, used to acquire the current measurement value of the first position collected by the first position sensor and the current measurement value of the second position collected by the second position sensor; a Kalman filter-state prediction module 120, used to perform Kalman filter-state prediction on the state estimate of the previous moment, the control input of the previous moment, and the covariance matrix of the previous moment to obtain the prior state estimate of the current moment and the prior covariance matrix of the current moment; an innovation calculation and statistical feature analysis module 130, used to perform innovation calculation and statistical feature analysis on the current measurement value of the first position and the current measurement value of the second position based on the prior state estimate of the current moment to obtain the innovation of the first position sensor, the innovation of the second position sensor, the normalized innovation square of the first position sensor, and the normalized innovation square of the second position sensor; and a fault diagnosis module 140, used to perform fault diagnosis based on the normalized innovation square of the first position sensor and the normalized innovation square of the second position sensor to obtain a fault flag.

[0046] As described above, the pitch servo motor rotor position redundancy detection system 100 according to embodiments of the present invention can be deployed in the pitch control system of a wind turbine or a dedicated condition monitoring unit, and interact with the first position sensor, the second position sensor, and the servo driver in real time. In one possible implementation, the pitch servo motor rotor position redundancy detection system 100 according to embodiments of the present invention can be integrated into the pitch control system of a wind turbine as an independent software module or hardware module. For example, the core models and parameters used for condition prediction and fault diagnosis in this system, including the simplified dynamic model parameters of the servo motor, such as the state transition matrix F, the control input matrix B, the process noise covariance matrix Q, and the measurement noise covariance, can be calibrated and optimized offline on the back-end server of the wind farm control center using motor specifications, system identification experiments, and historical operating data, and the optimized model parameter package can be sent to the front-end control unit. Of course, the complete process of performing real-time online detection in this system, including acquiring the current measurement values ​​of the two sensors, performing Kalman filtering state prediction, performing innovation calculation and statistical feature analysis, and finally performing fault diagnosis to generate fault signs, can also be embedded in dedicated edge computing hardware, such as the digital signal processing (DSP) or field-programmable gate array (FPGA) module inside the pitch controller, to accelerate the iterative process of matrix operations and filtering algorithms in the real-time data stream and ensure low-latency generation and output of the final fault signs.

[0047] It is understood that the above embodiments are merely exemplary implementations used to illustrate the principles of the present invention, and the present invention is not limited thereto. For those skilled in the art, various modifications and improvements can be made without departing from the spirit and essence of the present invention, and these modifications and improvements are also considered to be within the scope of protection of the present invention.

Claims

1. A method for detecting rotor position redundancy in a pitch servo motor, characterized in that, include: Acquire the current measured value of the first position acquired by the first position sensor and the current measured value of the second position acquired by the second position sensor; Kalman filtering-state prediction is performed on the state estimate, control input, and covariance matrix of the previous time step to obtain the prior state estimate and prior covariance matrix of the current time step. Based on the prior state estimation at the current moment, innovation calculation and statistical feature analysis are performed on the current measurement values ​​of the first position and the second position to obtain the innovation of the first position sensor, the innovation of the second position sensor, the normalized innovation square of the first position sensor and the normalized innovation square of the second position sensor. Fault diagnosis is performed based on the normalized squared innovation of the first position sensor and the normalized squared innovation of the second position sensor to obtain fault indicators.

2. The method for detecting rotor position redundancy of a pitch servo motor according to claim 1, characterized in that, The state estimate of the previous moment includes the servo motor position and servo motor speed.

3. The method for detecting rotor position redundancy of a pitch servo motor according to claim 2, characterized in that, Kalman filtering-state prediction is performed on the state estimate, control input, and covariance matrix from the previous time step to obtain the prior state estimate and prior covariance matrix for the current time step, including: The state estimate and control input from the previous moment are input into the simplified dynamic model of the servo motor to obtain the prior state estimate for the current moment. The prior covariance matrix at the current time is obtained by updating the covariance matrix at the previous time step based on the process noise covariance matrix.

4. The method for detecting rotor position redundancy of a pitch servo motor according to claim 3, characterized in that, The simplified dynamic model of the servo motor is input with the state estimate and control input from the previous moment to obtain the prior state estimate for the current moment. This includes processing the state estimate and control input from the previous moment using the simplified dynamic model of the servo motor with the following formula to obtain the prior state estimate for the current moment: in, Here is the state transition matrix. For the state estimation of the previous moment, To control the input matrix, This is the control input from the previous moment.

5. The method for detecting rotor position redundancy of a pitch servo motor according to claim 3, characterized in that, Updating the covariance matrix of the previous time step based on the process noise covariance matrix to obtain the prior covariance matrix of the current time step includes: updating the covariance matrix of the previous time step based on the process noise covariance matrix using the following formula: in, Here is the state transition matrix. Let be the covariance matrix of the previous time step. Let be the process noise covariance matrix.

6. The method for detecting rotor position redundancy of a pitch servo motor according to claim 1, characterized in that, Based on the prior state estimate at the current moment, innovation calculation and statistical feature analysis are performed on the current measurement values ​​of the first position and the second position to obtain the innovation of the first position sensor, the innovation of the second position sensor, the normalized squared innovation of the first position sensor, and the normalized squared innovation of the second position sensor, including: The deviations of the current measurement value of the first position and the current measurement value of the second position relative to the prior state estimate at the current time are calculated respectively to obtain the information of the first position sensor and the information of the second position sensor; Based on the prior covariance matrix at the current moment, the innovation uncertainty of the first position sensor measurement noise covariance and the second position sensor measurement noise covariance is quantified to obtain the first position sensor innovation covariance and the second position sensor innovation covariance. Based on the covariance of the first position sensor information, the covariance of the second position sensor information, the first position sensor information, and the second position sensor information, a normalized fault index is calculated to obtain the normalized squared information of the first position sensor and the normalized squared information of the second position sensor.

7. The method for detecting rotor position redundancy of a pitch servo motor according to claim 6, characterized in that, Calculating the deviations of the current measurement values ​​of the first position and the second position relative to the prior state estimate at the current time to obtain the first position sensor information and the second position sensor information includes: calculating the deviations of the current measurement values ​​of the first position and the second position relative to the prior state estimate at the current time using the following formula: in, For the measurement matrix, For the estimation of the prior state at the current moment, and These are the current measurement values ​​for the first position and the second position, respectively.

8. The method for detecting rotor position redundancy of a pitch servo motor according to claim 6, characterized in that, The normalized fault index is calculated based on the covariance of the first position sensor's innovation, the covariance of the second position sensor's innovation, the innovation of the first position sensor, and the innovation of the second position sensor to obtain the normalized squared innovation of the first position sensor and the normalized squared innovation of the second position sensor. This includes calculating the normalized fault index using the following formula: in, Normalized squared information of the first position sensor, For the normalized squared information of the second position sensor, The inverse of the information covariance of the first position sensor. The inverse of the information covariance of the second position sensor. For the transpose of the information from the first position sensor, This is a transpose of the information from the second position sensor.

9. A rotor position redundancy detection system for a pitch servo motor, characterized in that, include: The data acquisition module is used to acquire the current measurement value of the first position collected by the first position sensor and the current measurement value of the second position collected by the second position sensor; The Kalman filter-state prediction module is used to perform Kalman filter-state prediction on the state estimate, control input, and covariance matrix of the previous time step to obtain the prior state estimate and prior covariance matrix of the current time step. The innovation calculation and statistical feature analysis module is used to perform innovation calculation and statistical feature analysis on the current measurement value of the first position and the current measurement value of the second position based on the prior state estimation at the current time, so as to obtain the innovation of the first position sensor, the innovation of the second position sensor, the normalized innovation square of the first position sensor and the normalized innovation square of the second position sensor. The fault diagnosis module is used to perform fault diagnosis based on the normalized square of the first position sensor and the normalized square of the second position sensor to obtain a fault indicator.

10. The rotor position redundancy detection system for a pitch servo motor according to claim 9, characterized in that, Kalman filter-state prediction module, including: The prior state estimation acquisition unit is used to input the state estimate of the previous moment and the control input of the previous moment into the simplified dynamic model of the servo motor to obtain the prior state estimate of the current moment. The covariance matrix update unit is used to update the covariance matrix of the previous time step based on the process noise covariance matrix to obtain the prior covariance matrix of the current time step.

Images