Health monitoring method for reusable rocket turbine pump after basic maintenance

Abstract

The invention provides a health monitoring method for a reusable rocket turbine pump after basic maintenance, and relates to the technical field of spaceflight rockets, and the method comprises the steps: building a turbine pump reliability model under the condition of repeated use, and carrying out the health monitoring of the reusable rocket turbine pump based on the turbine pump reliability model; analyzing the normal use stage distribution of the reusable rocket turbine pump in the preventive maintenance period by using the ground accumulated working time-fault number data of the turbine pump; judging a starting point when the reusable rocket is cavitated in the basic repair state by using the normal use stage distribution; a VMD method is adopted to decompose vibration signals in the cavitation process of the turbine pump, and modal components are obtained; and a correlation coefficient method is utilized to position the fault characteristic component with high fitness. According to the method, self-adaptive decision-making can be carried out on a tiny but fatal fault of cavitation of the turbine pump which is used for multiple times according to the self-adaptive decision-making method.

Description

Technical Field

[0001] This invention relates to the field of aerospace rocket technology, and in particular to a method for health monitoring of reusable rocket turbopumps after basic maintenance. Background Technology

[0002] Space transportation systems are the most important foundation and prerequisite for all space activities. With the rapid development of the space transportation field in recent years, reusability is an important direction for its development, especially in the field of space rockets, which are receiving increasing attention from major spacefaring nations due to their advantages of high reliability, low cost, and strong mission adaptability.

[0003] High-thrust, multi-engine parallel liquid rockets, due to their higher thrust-to-weight ratio, present a significant problem: during the switching between the second-stage main engine and the vernier engine, the sudden closure of the main engine fuel valves can cause a water hammer effect, leading to a substantial pressure drop within the turbopump. When the pressure drops below the liquid's saturated vapor pressure, nuclei within the liquid begin to grow, resulting in cavitation cavities within the turbopump. These cavitation cavities disrupt the continuity of the liquid; when vapor bubbles flow towards the high-pressure region, they can implode and collapse, causing turbopump stall or cavitation. The movement, deformation, and collapse of cavitation cavities also cause intense oscillations and impacts within the turbopump. For high-thrust reusable rockets, given their harsh and variable load environments, insufficient component structural strength and fatigue resistance, repeated vibrations, impacts, and cyclic stresses can ultimately lead to fatigue failure of the equipment.

[0004] Reusable engines are characterized by highly integrated systems, harsh and variable operating environments, and insufficient component fatigue resistance, making them more prone to new types of failures. This is particularly true for turbopump systems, where structural strength and fatigue issues are more pronounced during cavitation due to the harsh load environment. Cavitation leads to excessive vibration and noise in the pump body; accumulated cavitation damage exacerbates material surface degradation and reduces pump hydraulic performance. To fully leverage the advantages of reusable rockets, cavitation detection through condition monitoring and fault diagnosis is the best solution to ensure the continuous reliability of reusable rocket turbopumps. As a rotating rotor, the impeller's vibration signals can be extracted and analyzed to obtain a wealth of information reflecting the pump's condition. Therefore, monitoring and diagnosis of turbopump cavitation under water hammer effects is largely based on the turbopump's own high and low pressure and vibration parameters. Time-frequency domain analysis of vibration signals can effectively obtain information on the time and frequency domain changes of the rotor. However, due to the low signal-to-noise ratio of vibration signals collected during rocket launch, identifying the characteristic frequencies of cavitation in the vibration signals under low signal-to-noise ratio conditions becomes a challenging problem. Therefore, timely and accurate detection of cavitation faults in turbopumps, reducing or avoiding their occurrence, improving the reliability of turbopumps for repeated use, minimizing losses caused by faults, and protecting the safety of personnel, equipment, and property make it particularly important to conduct research on intelligent diagnosis of turbopump cavitation faults.

[0005] While many countries have conducted research on overall engine fault diagnosis and health management, reusability, unlike expendable launch vehicle engines, is reflected in every minute system. Therefore, research on fatigue durability and reliability is particularly important when turbopump cavitation occurs, necessitating early cavitation fault diagnosis and health monitoring capabilities under complex and variable environments. Current research largely focuses on monitoring single-operation conditions and steady-state processes, lacking intelligent diagnosis under reusable conditions. This results in low coverage and sensitivity to turbopump cavitation—a minor but potentially fatal fault—and a lack of adaptive decision-making capabilities under fault conditions. Summary of the Invention

[0006] To overcome the shortcomings of the prior art, the purpose of this invention is to provide a method for monitoring the health of a reusable rocket turbopump after basic maintenance.

[0007] To achieve the above objectives, the present invention provides the following solution:

[0008] A method for monitoring the health of a reusable rocket turbopump after basic maintenance includes:

[0009] A reliability model of a turbopump under multiple use conditions is established, and based on the turbopump reliability model, the distribution of the normal use phase of the reusable rocket turbopump during the preventive maintenance cycle is analyzed using the ground cumulative working time-failure data of the turbopump.

[0010] The starting point of cavitation when a reusable rocket in a basic repair state is determined by using the distribution of the normal use phases.

[0011] The vibration signal during the cavitation process of a turbopump was decomposed using the VMD method to obtain modal components;

[0012] The correlation coefficient method is used to locate fault feature components with high fitness.

[0013] Preferably, it further includes:

[0014] Optimize the parameters in the VMD method using intelligent optimization algorithms.

[0015] Preferably, the intelligent optimization algorithm is the Raccoon Optimization Algorithm.

[0016] Preferably, the preventive maintenance cycle is the time cycle for inspection after a preset cumulative working time.

[0017] According to specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0018] This invention provides a method for health monitoring of reusable rocket turbopumps after basic maintenance, comprising: constructing a turbopump reliability model under multiple use conditions; and based on the turbopump reliability model, analyzing the distribution of normal use phases of the reusable rocket turbopump during the preventive maintenance cycle using the turbopump's cumulative ground operating time minus the number of failures data; using the distribution of normal use phases to determine the starting point of cavitation in the reusable rocket under basic repair conditions; using the VMD method to decompose the vibration signal during the turbopump cavitation process to obtain modal components; and using the correlation coefficient method to locate high-fitness fault characteristic components. This invention differs from most current technologies that only apply to monitoring single-operation conditions and steady-state processes. For intelligent diagnosis under reuse conditions, the system's determination of component health cannot solely rely on the zero-fault state at the time of manufacture; instead, it needs to make adaptive decisions based on previous usage data regarding the minor but potentially fatal fault of cavitation in repeatedly used turbopumps. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 A flowchart of the method provided in an embodiment of the present invention;

[0021] Figure 2 This is a schematic diagram of the technical route provided in the embodiments of the present invention;

[0022] Figure 3 This is a schematic diagram of the preventive maintenance cycle for a reusable system provided in an embodiment of the present invention;

[0023] Figure 4 This is a schematic diagram of the failure rate distribution after basic maintenance provided in an embodiment of the present invention;

[0024] Figure 5 A thermal diagram showing the cavitation stage distribution provided in an embodiment of the present invention. Detailed Implementation

[0026] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0027] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0028] Figure 1 The method flowchart provided in the embodiments of the present invention is as follows: Figure 1 As shown, this invention provides a method for health monitoring of a reusable rocket turbopump after basic maintenance, comprising:

[0029] Step 100: Build a turbopump reliability model under multiple use conditions, and based on the turbopump reliability model, analyze the distribution of the normal use phase of the reusable rocket turbopump during the preventive maintenance cycle using the turbopump's ground cumulative working time-failure data.

[0030] Step 200: Determine the starting point of cavitation when the reusable rocket experiences cavitation in its basic repair state using the distribution of the normal use phases.

[0031] Step 300: The vibration signal during the cavitation process of the turbopump is decomposed using the VMD method to obtain modal components;

[0032] Step 400: Use the correlation coefficient method to locate fault feature components with high fitness.

[0033] Specifically, this invention proposes an algorithm model for the availability and fault diagnosis of reusable rocket turbopumps based on a preventive maintenance cycle T. First, by building a turbopump reliability model under multiple use conditions, ground data is used to analyze the change in single-run time after turbopump use and maintenance with the number of reuses, verifying the distribution of the normal use phase of the reusable system, and using this as the basis for determining the cavitation initiation point during a single run. Second, for each run, the characteristic distribution of each cavitation stage is explored, and vibration signals are extracted using VMD decomposition. To better extract the cavitation characteristics of a single run, an intelligent optimization algorithm is used to optimize the VMD parameters, enabling them to adaptively match the penalty coefficient α and the number of decomposed modes k according to the reuse characteristics. Finally, based on the decomposed modal components, the correlation coefficient method is used to locate fault feature components with high fitness.

[0034] Based on the reliability model of the turbopump, the normal operating time distribution during cavitation failure is fitted to locate the cavitation initiation point during a single use. The cavitation failure mechanism and stage characteristics of the turbopump are analyzed, and a dataset is created using the cumulative operating time and failure data from ground data. Simultaneously, the feature extraction algorithm is optimized using an optimization algorithm, and relevant code is written in Python and experimentally verified. Detailed technical roadmap is as follows: Figure 2 As shown.

[0035] The technical approach of this invention is mainly divided into the following parts:

[0036] (1) Availability analysis of turbopump after cavitation under basic maintenance

[0037] By using ground-based cumulative working time minus failure data, we analyze the distribution of normal use phases of reusable rocket turbopumps during preventive maintenance cycles.

[0038] (2) The problem of cavitation initiation point location in the operating state after basic repair

[0039] By using the fitted distribution of normal use phases, we can help determine the starting point of cavitation when a reusable rocket experiences cavitation in a basic repair state.

[0040] (3) Problem of cavitation state division and diagnostic algorithm optimization for each stage of turbopump

[0041] The cavitation characteristics distribution of turbopumps at different stages was studied. To address the feature extraction problem under strong noise environment, the VMD parameters were optimized using an optimization algorithm to accurately decompose high-fit modal components in the vibration signal and avoid the phenomenon of mode aliasing.

[0042] Furthermore, the research content of this embodiment mainly consists of the following parts:

[0043] (1) Analyze the distribution of cavitation stages and related mechanisms of reusable rocket turbopumps, and discuss the mechanism of cavitation and the characteristics of fault signals.

[0044] (2) The definition of reusable systems differs from that of single-use systems. Reusable systems follow an operational pattern of "use – maintenance – reuse – re-maintenance – scrapping"; for example... Figure 3 As shown, the preventive maintenance cycle T is the time period during which the system must be inspected regardless of whether a fault occurs, after the system has accumulated a working time of T.

[0045] (3) As shown in Table 1 and Figure 4 As shown, the availability of key components is defined, and ground data is used to explore the distribution of normal use-cumulative normal use time of key components.

[0046] Figure 4 In the middle, ω * ω and T represent the preventive maintenance threshold and the failure threshold, respectively. i This indicates the time of the i-th maintenance. N represents the upper limit of the number of maintenance activities. This indicates that each maintenance performed after a failure will restore the lifespan, and a mapping moment with the same failure probability can be found during the first use. N This indicates that after N maintenance cycles, the device can no longer maintain normal operation and has reached the maximum number of maintenance cycles, at which point it must be replaced.

[0047] Table 1. Cumulative working time and number of malfunctions during ground testing.

[0048]

[0049]

[0050] (4) Distribution of cavitation stage characteristics as follows Figure 5 As shown.

[0051] Figure 5 In this context, RC stands for rotating cavitation, and CS stands for cavitation surge. Figure 5 The acceleration spectrum of the pump body is shown. N is the pump's rotational frequency, 338 Hz. Since the turbine pump's rotational speed remained constant during the experiment, its frequency remained constant with the cavitation coefficient. A rotating cavitation (RC) component, slightly faster than the rotor's rotational frequency, was observed at σ = 0.051–0.088. The cavitation surge (CS) frequency was slower than the rotational frequency (approximately half that of the rotational frequency), manifested at σ = 0.060–0.076.

[0052] (5) Vibration signal feature extraction and identification: the initial, intermediate and final frequencies may be different each time, that is, the fault component contained in each signal is different from the previous one; therefore, km,αm needs to be able to accurately extract the IMF component that can reflect the current cavitation state in the signal decomposition. The VMD decomposition parameter optimization details are shown in Table 2.

[0053] Table 2. Detailed Rules for VMD Decomposition Parameter Optimization

[0054]

[0055]

[0056] Furthermore, the life prediction for a single component after preventative maintenance is detailed in this embodiment as follows:

[0057] Once cavitation erosion occurs in a liquid rocket engine, it is impossible to restore it to its original condition after maintenance. The situation is even more complex for reusable rocket turbopumps. This embodiment is based on a basic repair (i.e., incomplete repair) model, assuming that the failure rate after repair (the probability of a liquid rocket engine failing per unit time at any given moment) is the same as before maintenance.

[0058] To infer the operational reliability of the rocket after maintenance, according to relevant research, the risk rate function of the engine turbopump, a critical component, follows a two-parameter Weibull distribution, which can be expressed as:

[0059] h(t)=β / α(t / α) β-1 (1)

[0060] In the formula: α, β, and t represent the scale parameter, shape parameter, and running time, respectively. The hazard rate function h(t) can be calculated using the failure probability density function f(t) and the reliability function R(t), and can be expressed as:

[0061] h(t)=f(t) / R(t) (2)

[0062] Let F(t) be the failure distribution function. The relationship between the reliability function R(t) and the failure distribution function F(t) is R(t) = 1 - F(t). The reliability function R(t) can be expressed as:

[0063]

[0064] Basic maintenance involves two tasks: repair and replacement. These two processes span the entire lifecycle of a single component, from its first use to its final replacement. Assuming N maintenance tasks are performed throughout a component's lifecycle, there will be (N-1) repair tasks and a final replacement task. Repair tasks reduce the component's deterioration and extend its cumulative working time; while replacement tasks force the component's effective age to zero. However, due to the cumulative damage to adjacent components, combined operation can lead to an increased risk rate, thus accelerating the degradation of the new component.

[0065] Due to the role of repair, a service life reduction factor γ is introduced into the service life prediction of reusable rockets after preventive maintenance. i It represents the rollback reduction usage time in the i-th complete lifetime τ. i The percentage within. During preventative maintenance after a single use, the reliability of a component recovers to some extent, but cannot be fully restored to its initial state. Here, i represents the number of preventative maintenance operations during use. γ i =1 represents the equipment performance being restored to its initial service state after maintenance, γ i =0 means that the maintenance activities have no substantial impact on the reliability of the equipment itself, and the equipment performance can only return to the state before the failure.

[0066] Effective age can be considered as the actual age of a component, and can be expressed as:

[0067]

[0068] Where Y i and Y i ′ represent the effective age before and after the i-th preventive maintenance, respectively.

[0069] For a component replacement task, after the m-th component replacement or the m-th life cycle, the hazard rate of the component before replacement is:

[0070]

[0071] Among them, b i Add a factor to the system's risk rate after task replacement. i This refers to the environmental factors for the launch mission. Common operating environmental factors are shown in Table 3.

[0072] Table 3 Common Working Environment Coefficients

[0073]

[0074] Therefore, according to equation (5), the risk rate drops to zero and the gradient is slightly large, indicating that after replacement, the operating condition deteriorates due to the cumulative damage of adjacent components, and the deterioration rate increases.

[0075] According to reliability theory, the relationship between the failure rate λ(t) and the reliability function R(t) is as follows:

[0076]

[0077] The number of times the maintained system can be reused is:

[0078]

[0079] Where T is the number of times the system can be reused, R(t) is the system reliability, and γ i ρ is the service life reduction factor, ρ is the environmental coefficient, and λ(t) is the failure rate.

[0080] Furthermore, this embodiment employs the VMD method to decompose the vibration signal during the cavitation process of a turbopump. VMD is a commonly used non-recursive signal mode decomposition method that can decompose complex multimodal time series data into a series of characteristic mode components with different bandwidths. The VMD decomposition process can be correspondingly transformed into the construction and solution of a variational problem. First, it is assumed that the original signal S is decomposed into K mode components u with finite bandwidths. k (t), that is:

[0081]

[0082] Assuming each modal component u k (t) at its specific frequency ω k The region (t) is sufficiently sparse. The decomposition property of VMD is that the sum of the estimated bandwidths of each modal component is minimized, and the constraint that the sum of all modal components equals the original input signal must be satisfied simultaneously. This transforms into an optimization problem requiring bandwidth estimation. The optimization process is as follows:

[0083] (1) Obtain the analytic signal u using the Hilbert transform. k (t), and calculate its one-sided spectrum.

[0084] (2) will u k The center band of (t) is modulated to the corresponding baseband.

[0085]

[0086] (3) Finally, the square norm of the demodulation gradient is calculated, the bandwidth of each component is estimated, and the corresponding constrained variational model is expressed as:

[0087]

[0088] Among them, {u k}={u1,u2,u3,…,u1,} and {ωk}={ω1,ω2,ω3,…,ω1,} represents the decomposed modal components and their corresponding center frequencies. k represents the number of decomposed modal components (k=1,2,…,K), and δ(t) represents the Dirac distribution function. * represents the convolution operator. It is an operator, and st indicates that it is subject to.

[0089] By introducing the Lagrange multiplier λ and the second-order penalty factor α, the constrained variational problem is transformed into an unconstrained variational problem, thereby finding the optimal solution. α improves the accuracy of signal reconstruction in Gaussian noise environments, while λ ensures that the signal satisfies the strictness of the constraints during processing. Therefore, the extended Lagrange expression is as follows:

[0090]

[0091] Then, the alternating direction multiplier method of the multiplier is used to cyclically update each modal component and its corresponding center frequency, thereby transforming the search for the optimal solution of the constrained variational problem into the search for the saddle point of the unconstrained model. Each modal component can be obtained in the frequency domain. k ω k The update processes for λ and λ are as follows:

[0092]

[0093] in, Represents the Fourier transform of the k-th modal component. This represents the Fourier transform of the original signal S(t). Let ω represent the Fourier transform of λ(t). k Let represent the center frequency of the k-th modal component. And... yes After Wiener filtering, the remaining quantity is used by the algorithm to change the center frequency according to the power spectrum of each component.

[0094] Repeat the above process, and stop iterating when the following condition is met.

[0095]

[0096] Where ε represents the convergence accuracy.

[0097] As a metaheuristic optimization scheme, Dehghani et al., inspired by raccoons' attacks on iguanas and their behavior of escaping predators, proposed the Raccoon Optimization Algorithm. The search mechanism of the Raccoon Optimization Algorithm includes two phases: exploration and development, and it possesses strong global search and local optimization capabilities. The specific steps are as follows:

[0098] I. Initialize algorithm parameters

[0099] Here we represent each raccoon as a distinct decision variable optimization scheme, and the raccoon's position in its behavior is randomly represented as X. i Specifically defined as:

[0100] X i :x i,j =lb j +r·(ub j -lb j ),i=1,2,…M,j=1,2,…d (16)

[0101] Among them, ub j The upper bound matrix with varying parameters; lb j is the lower bound matrix with variable parameters; r is a randomly generated real number in the range [0,1].

[0102] Based on the characteristics of the original signal, a suitable fitness function is selected, and the fitness value of each individual is calculated as the result of the possible solution position in the decision variables.

[0103] II. Algorithm Exploration

[0104] To highlight the algorithm's global search capability, a mathematical model simulating raccoon hunting behavior is established. Raccoon groups are divided into two groups when hunting: one group climbs trees to ambush prey, while the other group waits below for the prey to fall. Therefore, formula (17) is used to represent the position of the raccoon when it emerges from the tree.

[0105]

[0106] in, Let I be the location information of the i-th raccoon in the (t+1)-th iteration; I is a random integer in [1,2].

[0107] The beneficial effects of this invention are as follows:

[0108] (1) This invention addresses the cavitation failure of reusable rocket turbopumps by developing a turbopump health monitoring system after multiple uses, which is beneficial for research on the fatigue durability and reliability of turbopumps. Unlike most current technologies that only monitor single-operation conditions and steady-state processes, for intelligent diagnosis under reuse conditions, the system cannot simply claim the zero-fault state at the time of manufacture. Instead, it needs to make adaptive decisions based on previous usage data regarding the minor but fatal cavitation failure of the turbopump after multiple uses.

[0109] (2) The turbopump health diagnosis model built in this invention can make full use of ground data to analyze the remaining availability of the current turbopump, and use the distribution of the normal use stage of the reusable system obtained by fitting as the basis for judging the cavitation start point during a single operation. According to the characteristic distribution of each cavitation stage, the optimized VMD decomposition model is used to adaptively match the penalty coefficient α and the number of decomposition modes k according to the reuse characteristics.

[0110] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0111] This embodiment uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for health monitoring of a reusable rocket turbo pump after basic maintenance, characterized in that, The method comprises the following steps: A reliability model of the turbine pump under multiple use conditions is established, and based on the reliability model, the cumulative ground working time-failure data of the turbine pump is used to analyze the normal use stage distribution of the reusable rocket turbine pump in the preventive maintenance cycle; The starting point of cavitation of the reusable rocket in the basic repair state is determined by using the normal use stage distribution; The vibration signal in the cavitation process of the turbine pump is decomposed by using the VMD method to obtain the modal component; The high fitness fault feature component is located by using the correlation coefficient method.

2. The method of health monitoring of a reusable rocket turbopump after basic maintenance according to claim 1, characterized in that, Further comprising: The parameters in the VMD method are optimized by using an intelligent optimization algorithm.

3. The method of health monitoring of a reusable rocket turbopump after basic maintenance of claim 2, wherein, The intelligent optimization algorithm is the raccoon optimization algorithm.

4. The method of health monitoring of a reusable rocket turbopump after basic maintenance of claim 1, wherein, The preventive maintenance cycle is a time cycle for inspection after a preset cumulative working time.

Images