Aero-engine sensor fault-tolerant method and fault-tolerant system thereof

Abstract

The application relates to an aero-engine sensor fault tolerance method and a fault tolerance system thereof. The fault tolerance method comprises the following steps: S1, establishing a first type sensor signal reconstruction system and a second type sensor signal reconstruction system; S2, establishing a credibility evaluation mechanism for the first type sensor signal reconstruction system and the second type sensor signal reconstruction system, wherein the credibility evaluation mechanism selects an output value of the first type sensor signal reconstruction system, an output value of the second type sensor signal reconstruction system or a weight value of the output values of the two according to the output value of the first type sensor signal reconstruction system, the output value of the second type sensor signal reconstruction system and a residual error of the output value of the second type sensor signal reconstruction system. The scheme provided by the application can improve the reconstruction precision of the sensor signal of the aero-engine control system under the premise of meeting the basic robustness and safety of the control system.

Description

Technical Field

[0001] This invention relates to the field of fault tolerance technology for aero-engines, and particularly to a fault tolerance method and system for aero-engine sensors. Background Technology

[0002] Aero-engines are complex and highly nonlinear control objects. Given the specific operating conditions of aero-engines, the control system must possess extremely high reliability; this requirement is sometimes more urgent than simply improving system performance. System failures can lead to severe losses. Aero-engine sensors operate in high-temperature, high-vibration environments and are among the most unreliable control components in the system. Therefore, improving the fault-tolerant technology of airborne sensors for aero-engines is a crucial means to enhance the reliability of the engine control system and ensure flight safety.

[0003] Currently, airborne sensors primarily utilize two types of sensor redundancy technologies: hardware redundancy and analytical redundancy. Hardware redundancy uses multiple identical sensors to measure the same engine parameter, then uses a voter to detect faults. Analytical redundancy uses estimated values ​​of the measured parameters provided by engine models and estimation techniques as redundancy information; the deviation between the estimated values ​​and the sensor measurements is used to detect and isolate faults. Hardware redundancy acts as a backup, isolating faults. More hardware redundancy leads to higher control system reliability. However, hardware redundancy increases the weight, size, and development cost of the control system, impacting engine performance and fuel economy. Therefore, analytical redundancy is also a crucial fault handling method. Current engines widely employ Full Authority Digital Electronic Control (FADEC) systems. For critical sensor input signals in FADEC systems, analytical redundancy needs to be constructed. The purpose is to use the reconstructed sensor signals as a crucial basis for real-time sensor fault diagnosis. When all sensor hardware redundancy fails, the reconstructed sensor signals will still serve as backup inputs.

[0004] Regarding the research on constructing analytical redundancy of engine sensor signals, the method commonly used in engines both domestically and internationally is based on the synthesis of signals from other sensors. Engine service experience shows that this method is simple and reliable, and can reduce the impact of engine control system failures. In recent years, sensor reconstruction methods based on airborne engine models or data-driven approaches have been more popular, including Kalman estimation (virtual sensor technology), neural networks, support vector machines, and other adaptive or intelligent methods. Compared with sensor signal synthesis methods, these methods can more accurately diagnose and reconstruct faulty sensors in transient or performance deviation situations, and therefore have attracted great attention.

[0005] The linear Kalman filter algorithm, expressed in a recursive form, has been thoroughly studied by scholars. However, nonlinear phenomena are very common in practical applications. For example, nonlinear problems caused by external disturbances to the physical model, the presence of nonlinearity, and ill-conditioned variance matrices render the traditional linear Kalman filter algorithm unsuitable. There is an urgent need to improve the Kalman filter algorithm to obtain a Kalman filter technique applicable to nonlinear systems. Assuming all transformations are quasi-linear, the continuous nonlinear equations are first linearized and discretized. A first-order Taylor expansion is used to approximate the nonlinear model, resulting in the Extended Kalman Filter (EKF) algorithm corresponding to the nonlinear system. EKF expands the nonlinear state equations and observation equations into Taylor series based on the previous estimate and takes a first-order approximation to obtain a linearized model, thus following the standard KF recursive framework. When the system is weakly nonlinear, the EKF filtering accuracy is high, but when the system is strongly nonlinear, the EKF filtering accuracy is greatly reduced and may even lead to filter divergence. Because engines are highly nonlinear systems, the use and robustness of Kalman estimators have limitations. In terms of control systems, the reliability and safety requirements of FADEC control systems dictate that all control algorithms must have sufficient robustness. Therefore, the use of these new methods in control system signal reconstruction needs to be carefully considered.

[0006] In terms of sensor resolution redundancy design, there are two main methods: one is based on the synthesis of signals from other sensors, and the other is (sensor reconstruction method) based on adaptive or intelligent methods such as adaptive Kalman estimation, neural networks, and support vector machines. Among them, the Kalman filtering estimation method is currently the most promising technology for use in aero-engine control.

[0007] Sensor signal synthesis methods mainly focus on the reconstruction of sensor signals under steady-state and quasi-steady-state conditions. However, under transient processes or when the engine is subject to individual differences, performance degradation, or failure of gas path components, the accuracy of the reconstructed signal may deviate from the original design effect. Nevertheless, this method is simple and efficient, and extensive operational experience with aero engines has shown that it is robust, reliable, and safe.

[0008] Kalman estimation is currently one of the most promising techniques for aero-engine control. It is typically used to reconstruct sensor signals. This method can account for the effects of individual engine differences, performance degradation, and gas path component failures through state variable augmentation, resulting in more accurate reconstructed signals. However, there are some limitations to applying this method in practical engineering control. The Kalman estimator uses more measurement parameters than signal synthesis methods. While this ensures the accuracy of the Kalman estimator, it also reduces reliability. Failure of any measurement sensor can introduce errors into the reconstructed signal. Furthermore, because engines are highly nonlinear systems, the robustness of the extended Kalman estimator is limited. The reliability and safety requirements of the FADEC control system dictate that the control algorithm must possess sufficient robustness. Therefore, a compromise is needed to control the use of this method in control system signal reconstruction. Summary of the Invention

[0009] The technical problem to be solved by the present invention is to provide a fault-tolerant method and system for aero-engine sensors, which can improve the reconstruction accuracy of sensor signals in aero-engine control systems while meeting the basic robustness and safety requirements of the control system.

[0010] To address the aforementioned technical problems, this invention provides a fault-tolerant method for aero-engine sensors, the fault-tolerant method comprising:

[0011] Step S1: Establish the first type of sensor signal reconstruction system and the second type of sensor signal reconstruction system;

[0012] Step S2: Establish a reliability evaluation mechanism for the first type of sensor signal reconstruction system and the second type of sensor signal reconstruction system. The reliability evaluation mechanism selects to output the output value of the first type of sensor signal reconstruction system, the output value of the second type of sensor signal reconstruction system, and the residual of the output value of the second type of sensor signal reconstruction system, or selects the weight value of the output values ​​of the two.

[0013] According to one embodiment of the present invention, the first type of sensor signal reconstruction system employs a signal synthesis method, and the second type of sensor signal reconstruction system employs a Kalman estimation method.

[0014] According to one embodiment of the present invention, the first type of sensor signal reconstruction system has a first priority, and the second type of sensor signal reconstruction system has a second priority. Based on the reliability assessment mechanism, the second priority is an alternative fault-tolerant method relative to the first priority.

[0015] According to one embodiment of the present invention, the signal synthesis method is based on sensor signal data, engine nonlinear model, engine component-level model, or channel sensor values.

[0016] According to an embodiment of the present invention, the evaluation steps of the credibility assessment mechanism include:

[0017] Step S21: Calculate the residual of the output value of the second type of sensor signal reconstruction system;

[0018] Step S22: Obtain a first reconstructed signal and a second reconstructed signal, wherein the first reconstructed signal is a sensor signal of the first type of sensor signal reconstruction system and the second reconstructed signal is a sensor signal of the second type of sensor signal reconstruction system;

[0019] Step S23: Obtain the absolute value of the difference between the first reconstructed signal and the second reconstructed signal;

[0020] Step S24: Divide the absolute value of the difference by the residual to obtain the deviation value, and sort the deviation values ​​by size.

[0021] Step S25: Based on the sorting results, select the range of deviation values ​​in segments, and select the output value of the first type of sensor signal reconstruction system, the output value of the second type of sensor signal reconstruction system, or the weight value of the output values ​​of both based on the range.

[0022] According to an embodiment of the present invention, in step S21, the residual S of the output value of the second type of sensor signal reconstruction system is:

[0023]

[0024] Where y i These are sensor measurements from the engine. The value is calculated from the state-space model of the Kalman estimator, where n is the number of sensor measurements.

[0025] According to one embodiment of the present invention, in step S25, at least three values ​​T1, T2, and T3 are selected within the range of the deviation value, wherein 0 <T1<T2<T3;

[0026] When the deviation value is less than T1 or not less than T3, select the weight value of the output value of the first type of sensor signal reconstruction system and the second type of sensor signal reconstruction system.

[0027] When the deviation value is not less than T1 and less than T2, select the weight value of the output value of the first type of sensor signal reconstruction system and the second type of sensor signal reconstruction system.

[0028] When the deviation value is not less than T2 and less than T3, select the weight value of the output value of the first type of sensor signal reconstruction system and the second type of sensor signal reconstruction system.

[0029] This invention also provides a fault-tolerant system for aero-engine sensors, used to execute the aforementioned fault-tolerant method for aero-engine sensors. The fault-tolerant system includes identical numerical control (CNC) channels A and B, which exchange data via a data bus (CCDL). The CNC channels A and B each include:

[0030] A signal detection and verification module is used to determine whether the sensor signal is valid.

[0031] There are N signal reconstruction modules, each of which includes a first type of sensor signal reconstruction module, a second type of sensor signal reconstruction module, and a computing module. The computing module is connected to the first type of sensor signal reconstruction module and the second type of sensor signal reconstruction module, respectively.

[0032] The control law module controls the output analog and digital signals according to the reliability evaluation mechanism to control the state of the engine.

[0033] According to one embodiment of the present invention, the CNC A channel and the CNC B channel further include a signal status information module. After receiving the analog signal from the signal detection and confirmation module, the signal status information module converts it into a digital signal and sends it to the first type of sensor signal reconstruction module and the second type of sensor signal reconstruction module.

[0034] According to one embodiment of the present invention, the CNC A channel and the CNC B channel further include signal reconstruction switches, which connect the calculation module and the control law module, for selectively transmitting the results of the calculation module to the control law module.

[0035] This invention provides a fault-tolerant method and system for aero-engine sensors. It combines the advantages of signal synthesis and Kalman estimation methods using a fusion strategy to form a sensor fault-tolerant method and system that meets the existing FADEC control system configuration. Under the premise of satisfying the basic robustness and safety of the control system, it can improve the reconstruction accuracy of sensor signals in aero-engine control systems.

[0036] It should be understood that the above general description and the following detailed description of the present invention are exemplary and illustrative, and are intended to provide further explanation of the present invention. Attached Figure Description

[0037] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings, wherein:

[0038] Figure 1 A flowchart of the dual-channel signal processing of a prior art FADEC system is shown.

[0039] Figure 2 This invention illustrates a fault-tolerant method for aero-engine sensors according to an embodiment of the present invention;

[0040] Figure 3 The evaluation method for the credibility assessment mechanism is shown;

[0041] Figure 4 The diagram shows the signal synthesis structure of the Ps3 sensor;

[0042] Figure 5 The diagram shows the structure of the Kalman estimator for Ps3 signal reconstruction.

[0043] Figure 6 A comparison of the engine, Kalman estimator, and signal synthesis Ps3 is shown. Figure 1 ;

[0044] Figure 7 A comparison of the engine, Kalman estimator, and signal synthesis Ps3 is shown. Figure 2 ;

[0045] Figure 8 A schematic diagram of the output selection strategy for signal synthesis and Kalman estimator is shown;

[0046] Figure 9 A schematic diagram illustrating the credibility evaluation of signal synthesis and Kalman estimator is shown.

[0047] Figure 10 A schematic diagram of the fault-tolerant system for aircraft engine sensors according to an embodiment of the present invention is shown. Detailed Implementation

[0048] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0049] Numerous specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the invention is not limited to the specific embodiments disclosed below.

[0050] As indicated in this application and claims, unless the context clearly indicates otherwise, the words "a," "an," "an," and / or "the" are not specifically singular and may include plural forms. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of explicitly identified steps and elements, which do not constitute an exclusive list, and the method or apparatus may also include other steps or elements.

[0051] In detailing the embodiments of this application, for ease of explanation, the cross-sectional views illustrating the device structure will be partially enlarged and not to scale. Furthermore, the schematic diagrams are merely examples and should not limit the scope of protection of this application. In actual fabrication, the three-dimensional spatial dimensions of length, width, and depth should be included.

[0052] For ease of description, spatial relation terms such as “below,” “below,” “lower than,” “below,” “above,” “upper,” etc., may be used herein to describe the relationship of an element or feature shown in the accompanying drawings to other elements or features. It will be understood that these spatial relation terms are intended to include orientations of the device in use or operation other than those depicted in the accompanying drawings. For example, if the device in the accompanying drawings is flipped, the orientation of an element described as “below,” “below,” or “below” to other elements or features will change to “above” said other elements or features. Thus, the exemplary terms “below” and “below” can encompass both upward and downward directions. The device may also have other orientations (rotated 90 degrees or in other orientations), and therefore the spatial relation descriptors used herein should be interpreted accordingly. Furthermore, it will be understood that when a layer is referred to as being “between” two layers, it can be the only layer between the two layers, or there may be one or more layers in between.

[0053] In the context of this application, the structure described above the second feature may include embodiments in which the first and second features are formed in direct contact, or embodiments in which additional features are formed between the first and second features, such that the first and second features may not be in direct contact.

[0054] First, let me introduce the basic fault-tolerant framework for sensors in aero-engine control systems. Figure 1A flowchart of the dual-channel signal processing of a prior art FADEC system is shown. Taking a typical dual-channel full authority engine digital electronic control system as an example, the electronic controller consists of two functionally identical dual-redundant CNC channels, namely Channel A 101 (main channel) and Channel B 102 (backup channel). Either channel can complete all engine control functions, and the two functionally equivalent channels adopt a similar redundancy design. The two channels of the controller are hot backups of each other, and data is exchanged between the two channels through the channel data bus CCDL 103. During operation, channels A and B convert the engine status signals and the sensor signals of the CNC system into corresponding digital signals through their respective processing circuits. The signals are detected and confirmed in the signal detection module 104. Common detection and confirmation methods include extreme value detection, slope detection, and BIT self-test. The signal detection module confirms the validity of the signal, and then it proceeds to the parameter selection module 105. The parameter selection module 105 receives not only the value of its own channel but also the signal value of the backup channel via CCDL data. The selection logic in the parameter selection module 105 determines which signal will be used by the subsequent control law calculation module 106. Typical fault-tolerance strategies for dual-channel signals include: if both signals are valid, the average value is used; if the cross-channel signal fails, the signal of the own channel is used; if the signal of the own channel fails, the cross-channel signal is used; when the two channel signals are inconsistent, the better signal is preferred or the value from the model value module 107 is used; when both channel signals are invalid, the model value or the default value is used. The control signal determined by the selection logic serves as the input to the control law calculation module 106. Finally, the main control channel outputs corresponding analog and digital signals to the corresponding actuators (output control module 108) to control the engine's state.

[0055] Figure 1 The parameter selection module 105 and model value module 107 determine how faults are tolerated. This depends on the hardware configuration, the impact of signal failure, and the signal reconstruction capability. Different signals require different fault tolerance strategies, with signal reconstruction being a crucial part of this module. Ideally, a highly reliable virtual sensor signal consistent with the real engine sensor should be reconstructed. Sensor reconstruction methods based on signal synthesis are widely used in aero-engines. These methods primarily rely on the engine's airflow coupling characteristics, constructing analytical redundancy through fitting relationships with other sensor signal data. This approach is simple, reliable, and highly practical in engineering.

[0056] Figure 2 This invention illustrates a fault-tolerant method for aero-engine sensors according to an embodiment of the present invention; Figure 3The evaluation method for the reliability assessment mechanism is shown. As shown in the figure, the present invention provides a fault-tolerant method for aero-engine sensors, which includes:

[0057] Step S1: Establish the first type of sensor signal reconstruction system and the second type of sensor signal reconstruction system;

[0058] Step S2: Establish a reliability evaluation mechanism for the first type of sensor signal reconstruction system and the second type of sensor signal reconstruction system. The reliability evaluation mechanism selects to output the output value of the first type of sensor signal reconstruction system, the output value of the second type of sensor signal reconstruction system, and the residual of the output value of the second type of sensor signal reconstruction system, or selects the weight value of the output values ​​of the two.

[0059] Preferably, the first type of sensor signal reconstruction system employs a signal synthesis method, while the second type employs a Kalman estimation method. These two methods will be explained in detail later.

[0060] Preferably, the signal synthesis method is based on sensor signal data, engine nonlinear model, engine component-level model, or channel sensor values. Figure 4 A diagram of the Ps3 sensor signal synthesis structure is shown. As shown in the figure, to illustrate the features of this invention, the high-pressure compressor outlet static pressure sensor Ps3 is selected as an example to illustrate the signal synthesis method. Among them, the Ps3 sensor is the compressor outlet static pressure sensor, which mainly ensures the operability of the engine and the stability of the compressor. The maximum value of the high-pressure compressor outlet static pressure Ps3 is mainly limited by the combustion chamber casing strength and fuel pump pressure. The minimum value of the high-pressure compressor outlet static pressure Ps3 is mainly limited by the combustion chamber quench boundary and the aircraft bleed air pressure. This signal is a relatively important signal used in engine control. According to the following formulas (1)-(5), Ps3 is mainly calculated from the total temperature, total pressure, flow rate and cross-sectional area. For any point on the common working line of the high-pressure compressor, if the input conditions T25, P25 and N2R25, PS 3 / P 25 remain unchanged, the compressor efficiency, converted flow rate, pressure ratio and flow rate remain unchanged. The Ps3 obtained by the model is the same as the actual engine Ps3 value. To ensure computation time, a relationship table between the N2R25, P25 sensors and Ps3 can be established, and its signal synthesis structure diagram is shown below. Figure 4 As shown.

[0061] Eff=MAP HPC (N2R25,P 3 / P 25) Formula (1)

[0062] Wac=MAP HPC (N2R25,P 3 / P 25) Formula (2)

[0063] Wa=f1(Wac,P25,T25) Formula (3)

[0064] T3=f2(P 3 / P 25,T25,Eff) Formula (4)

[0065] Ps3=f3(P3,T3,A,Wa) Formula (5)

[0066] In the formula, MAP HPC This refers to the compressor characteristics: Eff is the compressor efficiency, Wac is the compressor equivalent flow rate, Wa is the compressor flow rate, P25 is the compressor inlet pressure, T25 is the compressor inlet temperature, T3 is the compressor outlet temperature, A represents the compressor cross-sectional area, and Ps3 is the compressor outlet static pressure. Similarly, signals from other engine sensors can be combined with aerodynamic-thermodynamic coupling methods of aero-engines to establish similar signal synthesis methods.

[0067] Using the Ps3 sensor as an example again, we will illustrate the sensor fault signal reconstruction method of the Kalman estimation method. It is easy to understand that everything described here applies to other sensor signals. From the perspective of signal synthesis, there is an assumption that the compressor characteristics must be constant. However, in reality, compressor characteristics can change due to manufacturing differences, performance degradation during service, and engine-related changes. In fact, the synthesized Ps3 signal deviates from the actual Ps3 signal from the sensor. Therefore, the constructed Kalman estimation method will include the influence of compressor characteristic changes, using compressor efficiency and flow rate health parameters to indicate the degree to which compressor characteristics deviate from the design value. First, a state-space model is designed, using the compressor efficiency health parameter KE3D25 and the flow rate health parameter KW25R as augmented state variables. The state-space model described in this paper is shown in Equation 6. This state-space model is not unique; input, state, and output variables can be added or subtracted according to actual conditions.

[0068]

[0069] With a state-space model, an augmented linear Kalman estimator can be designed using the Kalman estimation method, a traditional control theory approach that will not be detailed here. A sensor signal reconstruction method using the Kalman estimation method is as follows: Figure 5 As shown, this structure uses N1, N2, T25, T3, and P25 as the measurement parameters of the Kalman estimator 501. The Kalman estimation method is used to estimate N1, N2, and the health parameters KE3D25 and KW25R. These health parameters, along with the fuel quantity, are used as the output of the state-space model 502 to obtain the predicted Ps3 signal. Figure 5The diagram shows the structure of the Kalman estimation method for Ps3 signal reconstruction.

[0070] Compared to signal synthesis methods, the Kalman estimation method additionally incorporates the measurement signals from sensor N1 and sensor T3, as well as compressor efficiency and flow health parameters. This enhances the fault tolerance of the method under transient and Ps3 sensor failure conditions. However, due to the increased number of dependent sensors, the reliability of Ps3 signal fault tolerance is actually reduced. Assuming the reliability of the sensors is the same, the more sensors relied upon, the greater the probability of overall measurement system failure, which may lead to Ps3 signal errors. Taking the above example, assuming that sensor N1 fails simultaneously, the reconstruction results of the Kalman estimation method and the signal synthesis method under N1 sensor failure are as follows: Figure 4 As shown. Figure 6 The comparison of the engine, Kalman estimator, and signal synthesis Ps3 is shown. Figure 1 .from Figure 6 As can be seen, the Ps3 signal reconstructed by the signal synthesis method is consistent with the real engine Ps3, but the Ps3 signal reconstructed based on the Kalman estimation method will be affected because the error information of N1 "may" be brought into the reconstruction of Ps3. In this example, the failure of the N1 sensor caused a large deviation in the Ps3 signal reconstructed by the Kalman estimator.

[0071] The use of the word "may" in the above description has a specific meaning. Because in signal reconstruction using the Kalman estimation method, if there is a linear relationship between the measured sensor signals, then even if one sensor fails, the Kalman estimation method will add that effect to the state variables, thus reducing the impact on the reconstructed sensor signals. Using the example above, assuming that all N2 sensors fail simultaneously, the reconstruction results of the Kalman estimator and the signal synthesis method under the failure of N2 sensors are as follows: Figure 7 As shown. From Figure 7 As can be seen, the Ps3 signal reconstructed by the Kalman estimator is consistent with the actual engine Ps3. Conversely, the Ps3 signal reconstructed by the signal synthesis method shows a significant deviation. This indicates that the sensor signal added to the Kalman filter can increase the estimation accuracy of Ps3 in certain situations. It is precisely this characteristic of the Kalman filtering estimation method that can compensate for the shortcomings of the signal synthesis method in some cases. The basic idea of ​​this invention is to propose a method that integrates the aforementioned signal synthesis method and Kalman estimation method, taking advantage of their respective strengths to increase the accuracy of the reconstructed signal, thereby constructing a more accurate fault-tolerant system. A crucial aspect of this is how to integrate these two signals.

[0072] Preferably, the first type of sensor signal reconstruction system has first priority, and the second type of sensor signal reconstruction system has second priority. Based on the reliability assessment mechanism, the second priority is an alternative fault-tolerant method relative to the first priority, meaning that the output value of the first type of sensor signal reconstruction system is the preferred selection scheme when the signal deviation is the same.

[0073] Better, refer to Figure 3 The credibility assessment mechanism includes the following steps:

[0074] Step S21: Calculate the residual of the output value of the second type of sensor signal reconstruction system;

[0075] Step S22: Obtain the first reconstructed signal and the second reconstructed signal. The first reconstructed signal is the sensor signal of the first type of sensor signal reconstruction system, and the second reconstructed signal is the sensor signal of the second type of sensor signal reconstruction system.

[0076] Step S23: Obtain the absolute value of the difference between the first reconstructed signal and the second reconstructed signal;

[0077] Step S24: Divide the absolute value of the difference by the residual to obtain the deviation value, and sort the deviation values ​​by size.

[0078] Step S25: Based on the sorting results, select the range of deviation values ​​in segments, and select the output value of the first type of sensor signal reconstruction system, the output value of the second type of sensor signal reconstruction system, or the weight value of the output values ​​of both based on the range.

[0079] Preferably, in step S21, the residual S of the output value of the second type of sensor signal reconstruction system is:

[0080]

[0081] Where y i These are sensor measurements from the engine. The value is calculated from the state-space model of the Kalman estimator, where n is the number of sensor measurements.

[0082] Preferably, in step S25, at least three values ​​T1, T2, and T3 are selected within the range of deviation values, where 0 <T1<T2<T3;

[0083] When the deviation value is less than T1 or not less than T3, select the output value of the second type of sensor signal reconstruction system;

[0084] When the deviation value is not less than T1 and less than T2, select the output value of the first type of sensor signal reconstruction system;

[0085] When the deviation value is not less than T2 and less than T3, select the weight value of the output value of the first type of sensor signal reconstruction system and the second type of sensor signal reconstruction system.

[0086] The following details the process of fusing the signal synthesis method with the Kalman estimation method. This invention uses the output of the Kalman calculation method and the residual of the measurement sensor output. In this embodiment, the output residual S is defined as shown in Formula 7, where y i The values ​​are measured by five sensors on the engine, i.e., n equals 5. Values ​​calculated for the state-space model of the Kalman estimator:

[0087]

[0088] The results of two Ps3 reconstruction methods under combined sensor failure and engine degradation conditions are compared below. The output residuals of the Kalman estimator are obtained, and the deviations between the Ps3 signal reconstructed by the Kalman estimation method and the actual engine Ps3 signal are obtained (column 5 of Table 1), as well as the deviations between the Ps3 signal reconstructed by the signal synthesis method and the actual engine Ps3 signal (column 6 of Table 1). The information shown in Table 1 can be obtained.

[0089] Table 1. Statistical information of Ps3 signal reconstructed by the two methods.

[0090]

[0091] Table 1 can be transformed according to the following principles to obtain Table 2.

[0092] (1) Since model values ​​are required, the Ps3 signal of the control system channel is invalid and can only rely on two reconstructed signals. Therefore, subtract the 5th and 6th columns in Table 1 to obtain the difference between the Ps3 signal reconstructed by the Kalman estimator and the Ps3 signal synthesized by the signal synthesis, denoted as SVM_Syn.

[0093] (2) Obtain the absolute value of the difference between the Ps3 signal reconstructed by the Kalman estimation method and the Ps3 signal reconstructed by the signal synthesis method, denoted as |SVM_Syn|.

[0094] (3) Divide |SVM_Syn| by the residual S to obtain the deviation value of the two methods under unit residual, denoted as |SVM_Syn| / S. This calculation has two advantages: first, it can eliminate the influence of fault amplitude; second, the introduction of residual can measure the reliability of the Kalman estimator and the signal reconstruction deviation.

[0095] (4) Sort according to the size of |SVM_Syn| / S (column 6 of Table 2).

[0096] Table 2 Comparison of Ps3 signals reconstructed by the two methods

[0097]

[0098]

[0099] The data in Table 2 can be summarized into three parts. Rows 1-3 and 18-21 indicate that the Ps3 signal reconstructed using the Kalman estimation method is superior to or equivalent to the Ps3 signal reconstructed using the signal synthesis method, with the corresponding |SVM_Syn| / S range defined as 0-0.2 and >5. This data can be modified accordingly. Rows 4-12 indicate that the Ps3 signal reconstructed using the signal synthesis method is superior to the Ps3 signal reconstructed using the Kalman estimation method, with the corresponding |SVM_Syn| / S range defined as 0.2-1.0. Rows 13-17 indicate that both the Kalman estimation method and the signal synthesis method will have significant errors in reconstructing the Ps3 signal. However, considering that the deviations of the two methods are in opposite directions, a weighted average method will be used to obtain Ps3, which can reduce the deviation of the reconstructed Ps3 signal. The corresponding |SVM_Syn| / S range is defined as 1.0-5. Based on the above results, we can summarize... Figure 8 The credibility assessment mechanism, or fusion strategy, of the signal synthesis method and Kalman estimation method shown is as follows: [The fusion structure is described in the original text.] Figure 9 As shown. From Figure 8 It can be seen from this:

[0100] As is easily understood, at least three values, T1, T2, and T3, are selected within the range of deviation values. In practice, four or more values ​​can also be selected. Based on the segmented range formed by these selected values, the weight values ​​for the output values ​​of the first-type sensor signal reconstruction system and the second-type sensor signal reconstruction system are chosen.

[0101] When |SVM_Syn| / S is less than T1, it indicates that the index is small, and the difference between the signal synthesis method and the Kalman filtering method is small. Therefore, a more accurate Kalman estimator method, i.e., a fault-tolerant method, can be used to select the output value of the second type of sensor signal reconstruction system (SVM). This can be understood as follows: here, the weight of the output value of the first type of sensor signal reconstruction system is 0, while the weight of the output value of the second type of sensor signal reconstruction system is 1.0.

[0102] When |SVM_Syn| / S is between T1 and T2, it indicates that the difference between the signal synthesis method and the Kalman filtering method is within a residual range. At this point, it is impossible to determine which of the two signals is more accurate, and the Kalman estimator is more likely to err. Therefore, the more robust signal synthesis method is chosen, i.e., the fault-tolerant method selects the output value (Syn) of the first type of sensor signal reconstruction system with a higher priority. This can be understood as the weight of the output value of the first type of sensor signal reconstruction system being 1.0, while the weight of the output value of the second type of sensor signal reconstruction system is 0.

[0103] When |SVM_Syn| / S is between T2 and T3, it indicates that the difference between the signal synthesis method and the Kalman estimation method is within a few residual ranges. At this point, it's still impossible to determine which signal is more accurate, but it's clear that there's a significant deviation between the two signals. There's a high probability that both the Kalman estimator and the signal synthesis method are malfunctioning, mainly because a common input signal they use has a problem. In this case, the average of the output values ​​of the first and second type of sensor signal reconstruction systems (SVM + Syn) should be chosen. Alternatively, choosing the signal synthesis method with a higher priority at this stage is also an option, or choosing the output value (Syn) of the first type of sensor signal reconstruction system is also feasible. This can be understood as the weight of the output value of the first type of sensor signal reconstruction system being 0.5, and the weight of the output value of the second type of sensor signal reconstruction system being 0.5. Besides choosing the average of the two, a non-average approach can also be used, such as choosing 0.6 and 0.4, as long as their weights are between 0 and 1.0.

[0104] When |SVM_Syn| / S is greater than T3, it indicates that S is very small, and the Kalman estimator method can track the sensor output well. In this case, a more accurate Kalman estimator method can be used, namely, the fault-tolerant method to select the output value of the second type of sensor signal reconstruction system (SVM). Here, the weight of the output value of the first type of sensor signal reconstruction system is 0, while the weight of the output value of the second type of sensor signal reconstruction system is 1.0.

[0105] Figure 10 A schematic diagram of a fault-tolerant system for an aero-engine sensor according to an embodiment of the present invention is shown. As shown, the present invention also provides an aero-engine sensor fault-tolerant system 1000. This fault-tolerant system is used to execute the aforementioned aero-engine sensor fault-tolerant method. The fault-tolerant system includes identical CNC A channel and CNC B channel, which exchange data via a data bus CCDL. Figure 10 This explanation will focus on one of the CNC A channel and the CNC B channel. For example, the CNC A channel includes:

[0106] The signal detection and confirmation module 1001 is used to determine whether the sensor signal is valid.

[0107] There are N signal reconstruction modules 1002. Each signal reconstruction module 1002 includes a first type of sensor signal reconstruction module 1003, a second type of sensor signal reconstruction module 1004, and a calculation module 1005. The calculation module 1005 is connected to the first type of sensor signal reconstruction module 1003 and the second type of sensor signal reconstruction module 1004, respectively.

[0108] The control law module 1006 controls the output analog and digital signals based on a reliability assessment mechanism to control the engine's state.

[0109] Preferably, the CNC A channel and the CNC B channel also include a signal status information module 1007. After receiving the analog signal from the signal detection and confirmation module 1001, the signal status information module 1007 converts it into a digital signal and sends it to the first type of sensor signal reconstruction module 1003 and the second type of sensor signal reconstruction module 1004.

[0110] Preferably, the CNC A channel and the CNC B channel also include a signal reconstruction switch 1008, which connects the calculation module 1005 and the control law module 1006, and is used to selectively transmit the results of the calculation module 1005 to the control law module 1006.

[0111] This invention provides a fault-tolerant method for aero-engine sensors, which includes a first type of sensor signal reconstruction system and a second type of sensor signal reconstruction system, each with different priorities. The first type of sensor signal reconstruction system employs a more robust fault-tolerant method (such as signal synthesis, engine nonlinear models, engine component-level models, or other modeling methods, even channel sensor values), and has the highest priority. The second type of sensor signal reconstruction system uses Kalman estimation, which offers superior reconstruction performance and has the second highest priority. A reliability assessment mechanism is established for both systems. The reconstructed signal output is determined according to this mechanism, and the output can be selected from the output values ​​of the first and second types of sensor signal reconstruction systems, or a weighted value of their respective output values. The output value serves as a voter for the dual-channel aero-engine sensor fault-tolerant system and as a backup model control signal. The reliability assessment mechanism is based on the output values ​​of the first and second types of sensor signal reconstruction systems, as well as their residuals.

[0112] The beneficial effects of the fault-tolerant method and fault-tolerant system for aero-engine sensors provided by this invention are as follows:

[0113] (1) While maintaining the original control fault tolerance robustness, the fault tolerance performance has been improved. The sensor fault reconstruction system of the signal synthesis method has been shown to be robust and safe through the operation and service experience of existing aero engines. As long as the fault tolerance signal is within the range, the robustness of the engine control fault tolerance can be guaranteed. The sensor fault reconstruction system established by introducing the constrained Kalman estimation method ensures that a better sensor fault tolerance performance is selected within a reasonable confidence range. It solves the influence of factors such as individual engine differences, performance degradation, and gas path component failures on the signal synthesis method, and has better transient fault tolerance capability. It achieves the goal of maintaining robustness and safety while improving reconstruction performance.

[0114] (2) By introducing a reliability assessment mechanism, the fault-tolerant control of the engine always prioritizes the signal synthesis method as the first fault-tolerant approach, without compromising the unique safety requirements of the aero-engine control system or significantly affecting the original control logic and structure. Before this invention, only the sensor fault reconstruction system output based on the signal synthesis method was used for dual-channel signal voting or as a backup control signal. In this invention, only a sensor fault reconstruction system based on the Kalman estimation method is added and fused with the output of the signal synthesis reconstruction system. The fused signal is then used for dual-channel signal voting or as a backup control signal, without affecting the subsequent voting and control logic.

[0115] (3) The reliability assessment mechanism defined in this invention is applicable to all first-class sensor signal reconstruction systems and second-class sensor signal reconstruction systems established by adaptive methods. Therefore, this system framework is suitable for the fault tolerance assessment of all first-class sensor signal reconstruction systems and second-class sensor signal reconstruction systems established by adaptive methods, promoting the use of new fault tolerance technologies in the relatively "conservative" field of aero-engines.

[0116] Although the present invention has been described with reference to specific embodiments, those skilled in the art should recognize that the above embodiments are merely illustrative of the invention, and various equivalent changes or substitutions can be made without departing from the spirit of the invention. Therefore, any changes or modifications to the above embodiments within the essential spirit of the invention will fall within the scope of the claims of this application.

Claims

1. A fault-tolerant method for aircraft engine sensors, the fault-tolerant method comprising: Step S1: Establish the first type of sensor signal reconstruction system and the second type of sensor signal reconstruction system; Step S2: Establish a reliability evaluation mechanism for the first type of sensor signal reconstruction system and the second type of sensor signal reconstruction system. The reliability evaluation mechanism selects to output the output value of the first type of sensor signal reconstruction system, the output value of the second type of sensor signal reconstruction system, and the residual of the output value of the second type of sensor signal reconstruction system, or selects the weight value of the output values ​​of the two. The first type of sensor signal reconstruction system employs a signal synthesis method, while the second type of sensor signal reconstruction system employs a Kalman estimation method. The first type of sensor signal reconstruction system has a first priority, and the second type of sensor signal reconstruction system has a second priority. Based on the reliability assessment mechanism, the second priority is an alternative fault-tolerant method relative to the first priority. The evaluation steps of the credibility assessment mechanism include: Step S21: Calculate the residual of the output value of the second type of sensor signal reconstruction system; Step S22: Obtain a first reconstructed signal and a second reconstructed signal, wherein the first reconstructed signal is a sensor signal of the first type of sensor signal reconstruction system and the second reconstructed signal is a sensor signal of the second type of sensor signal reconstruction system; Step S23: Obtain the absolute value of the difference between the first reconstructed signal and the second reconstructed signal; Step S24: Divide the absolute value of the difference by the residual to obtain the deviation value, and sort the deviation values ​​by size. Step S25: Based on the sorting results, select the range of deviation values ​​in segments, and select the output value of the first type of sensor signal reconstruction system, the output value of the second type of sensor signal reconstruction system, or the weight value of the output values ​​of both based on the range.

2. The fault-tolerant method as described in claim 1, characterized in that, The signal synthesis method is based on sensor signal data, engine nonlinear model, engine component-level model, or channel sensor values.

3. The fault-tolerant method as described in claim 1, characterized in that, In step S21, the residual S of the output value of the second type of sensor signal reconstruction system is: Where y i These are sensor measurements from the engine. The value is calculated from the state-space model of the Kalman estimator, where n is the number of sensor measurements.

4. The fault-tolerant method as described in claim 1, characterized in that, In step S25, at least three values ​​T1, T2, and T3 are selected within the range of the deviation value, where 0 <T1<T2<T3; When the deviation value is less than T1 or not less than T3, the output value of the second type of sensor signal reconstruction system is selected to be output. When the deviation value is not less than T1 and less than T2, the output value of the first type of sensor signal reconstruction system is selected to be output. When the deviation value is not less than T2 and less than T3, select the weight value of the output value of the first type of sensor signal reconstruction system and the second type of sensor signal reconstruction system.

5. A fault-tolerant system for aircraft engine sensors, characterized in that, For executing the fault-tolerant method for aero-engine sensors according to any one of claims 1 to 4, the fault-tolerant system includes identical numerical control (CNC) channels A and B, which exchange data via a data bus CCDL. The CNC channels A and B each include: A signal detection and verification module is used to determine whether the sensor signal is valid. There are N signal reconstruction modules, each of which includes a first type of sensor signal reconstruction module, a second type of sensor signal reconstruction module, and a computing module. The computing module is connected to the first type of sensor signal reconstruction module and the second type of sensor signal reconstruction module, respectively. The control law module controls the output analog and digital signals according to the reliability evaluation mechanism to control the state of the engine.

6. The fault-tolerant system as described in claim 5, characterized in that, The CNC A channel and CNC B channel also include a signal status information module. After receiving the analog signal from the signal detection and confirmation module, the signal status information module converts it into a digital signal and sends it to the first type of sensor signal reconstruction module and the second type of sensor signal reconstruction module.

7. The fault-tolerant system as described in claim 5, characterized in that, The CNC A channel and CNC B channel also include signal reconstruction switches, which connect the calculation module and the control law module, and are used to selectively transmit the results of the calculation module to the control law module.

Images