A digital assembly method for aviation engine fuel main pipe

Abstract

The present invention belongs to the technical field of digital assembly of aircraft engine fuel manifolds, and specifically relates to a digital assembly method for aircraft engine fuel manifolds. The specific technical scheme is: perform a physical flow characteristic test on the fuel manifold and calculate the physical flow unevenness. If the physical flow unevenness meets the requirements, the assembly is completed; if the physical flow unevenness does not meet the requirements, a digital assembly test is carried out through the digital twin model of the fuel manifold. If the prediction accuracy of all nozzles meets the threshold, a new nozzle is placed in the digital twin model of the fuel manifold to replace the nozzle with the largest flow deviation value, and digital assembly is performed and the digital flow unevenness is calculated. If the digital flow unevenness meets the requirements, the new nozzle is actually replaced, and the physical flow characteristic test is re-performed until the physical unevenness meets the requirements and the assembly is completed.

Description

Technical Field

[0001] The invention belongs to the technical field of digital assembly of a fuel manifold of an aircraft engine, and in particular relates to a digital assembly method for a fuel manifold of an aircraft engine. Background Art

[0002] The fuel manifold is the core component of the aircraft engine combustion system. Its core function is to inject and atomize fuel into the engine combustion chamber. Therefore, the various performance indicators of the fuel manifold have a direct impact on whether the fuel in the engine combustion chamber is fully burned, which in turn determines the performance, combustion efficiency, safety and reliability of the combustion chamber, and ultimately plays a decisive role in the operating efficiency, safety and reliability of the entire engine. There are many types of fuel manifolds in modern aircraft gas turbine engines, and they have obvious differences in structure.

[0003] For the assembly of the fuel manifold, the process personnel first need to measure and collect the flow characteristics of each fuel nozzle, specifically testing the fuel outlet flow and spray angle of the fuel nozzle under different fuel pressures. Based on the flow characteristics of the same batch of fuel nozzles, the process personnel use their rich experience to select fuel nozzles with similar flow characteristics and assemble them into a fuel manifold. Then, the fuel manifold is tested for flow at different pressure measurement points, and the unevenness of the fuel manifold is calculated. This parameter is the core performance indicator of the fuel manifold and is the core factor affecting the combustion chamber outlet temperature distribution and turbine working reliability. Therefore, the process personnel need to focus on monitoring it. If the unevenness is within the process specification requirements, the fuel manifold can be used for product delivery. Otherwise, the process personnel need to perform parameter analysis to find single or multiple fuel nozzles that have a large impact on the unevenness, and replace them for reassembly until the unevenness meets the test standard requirements.

[0004] Limited by the existing assembly process level, this method is still the main assembly method in engine factories. It relies heavily on subjective factors, so it is easy to have substandard performance after assembly or even need multiple assembly, which affects the factory production cycle and delays delivery. At the same time, a single fuel main is assembled by more than a dozen or even more than 20 fuel nozzles, and most fuel nozzles are divided into small flow type and large flow type. Therefore, it is often difficult to select a set of fuel nozzles manually. On the production line, this task is generally completed by experienced craftsmen, but it often takes a long time to train such masters. Therefore, when the old masters retire and new people are not trained, the factory will be unqualified. In addition, since the fuel nozzles are randomly selected for assembly by "feeling", the success rate of this assembly method is low, and repeated assembly and re-testing are required frequently, which increases the manpower and time costs, resulting in low factory production efficiency and longer delivery cycles. Summary of the invention

[0005] In order to solve the above technical problems, the present invention proposes a digital assembly method for an aircraft engine fuel manifold, which can realize the digital assembly and unevenness testing of the fuel manifold in the cloud, thereby reducing the number of actual physical tests and improving assembly efficiency.

[0006] To achieve the above-mentioned purpose of the invention, the technical solution adopted by the present invention is: a digital assembly method for an aircraft engine fuel manifold, selecting m small flow nozzles and n large flow nozzles to form a fuel manifold; performing a physical flow characteristic test on the fuel manifold and calculating the physical flow unevenness. If the physical flow unevenness meets the standard requirements, the assembly is completed; if the physical flow unevenness does not meet the standard requirements, a digital assembly test is carried out through a digital twin model of the fuel manifold. If the model prediction accuracy meets the threshold requirements, a new nozzle is placed in the digital twin model of the fuel manifold to replace the nozzle with the largest flow deviation value, and the fuel manifold is digitally assembled and the digital flow unevenness is calculated. If the digital flow unevenness meets the requirements, the new nozzle is actually replaced, and the physical flow characteristic test is re-performed until the physical unevenness meets the requirements and the assembly is completed.

[0007] Preferably, the calculation formula of the physical / digital flow non-uniformity k is as follows:

[0008]

[0009] In the formula, f′ min is the minimum physical / digital flow in the fuel main; f′ max is the maximum physical / digital flow in the fuel rail.

[0010] Preferably, the flow deviation value σ of the fuel main pipe is calculated as follows:

[0011]

[0012] In the formula, is the flow prediction value of the i-th nozzle; m is the number of small flow nozzles; n is the number of large flow nozzles.

[0013] Preferably: the fuel manifold digital twin model includes a fuel nozzle coding network layer and a fuel manifold decoding network layer, the fuel nozzle coding network layer encodes the flow characteristics of m+n fuel nozzles, transmits them to the fuel manifold decoding network layer for decoding, and predicts the flow characteristics of the fuel manifold.

[0014] Preferably: the fuel nozzle coding network layer includes a secondary oil circuit coding network layer and a main and secondary oil circuit coding network layer, the secondary oil circuit coding network layer input is the secondary oil circuit performance data of the fuel nozzle, and the main and secondary oil circuit coding network layer is the main and secondary oil circuit performance data of the fuel nozzle;

[0015] And / or, the fuel main pipe decoding network layer includes a secondary oil circuit decoding network layer and a main and secondary oil circuit decoding network layer. The secondary oil circuit decoding network layer and the main and secondary oil circuit decoding network layer respectively obtain coding information from the fuel nozzle coding network layer to obtain the predicted flow in the fuel main pipe secondary oil circuit and the main and secondary oil circuit.

[0016] Preferably, the auxiliary oil circuit decoding network layer and the main and auxiliary oil circuit decoding network layer both adopt a fully connected neural network layer, and transmit the encoded information to the fuel main pipe decoding network layer by assigning different weights.

[0017] Preferably: the coding information of the auxiliary oil circuit coding network layer and the main-auxiliary oil circuit coding network layer are γ1 and γ2 respectively, then the inputs of the auxiliary oil circuit decoding network layer and the main-auxiliary oil circuit decoding network layer are ωγ1+(1-ω)γ2 and (1-ω)γ1+ωγ2 respectively; ω is the weight coefficient, and its value is 0-1.

[0018] Preferably, the method comprises the following steps:

[0019] S11, selecting m small flow nozzles with similar flow characteristics;

[0020] S12, selecting n large flow nozzles with similar flow characteristics;

[0021] S13, assembling m small flow nozzles and n large flow nozzles into a fuel main pipe, and performing a physical flow characteristic test of the fuel main pipe;

[0022] S14, calculating the physical flow unevenness of the fuel main pipe;

[0023] S15, judging whether the physical flow non-uniformity meets the standard requirements: if it meets the standard requirements, proceeding to step S21; if it does not meet the standard requirements, proceeding to step S16;

[0024] S16. Conduct digital assembly tests using the digital twin model of the fuel main to obtain digital flow characteristic data and calculate the model prediction accuracy;

[0025] S17, judging whether the model prediction accuracy meets the threshold requirement: if the model prediction accuracy meets the threshold requirement, proceed to step S18; if the model prediction accuracy does not meet the threshold requirement, randomly replace the nozzle and proceed to step S13;

[0026] S18, calculating the flow deviation values ​​of m+n nozzles, and placing a new nozzle into the digital twin model of the fuel manifold to replace the nozzle with the largest flow deviation value;

[0027] S19, performing a digital assembly test on the new fuel main pipe in step S18 and calculating the digital flow unevenness;

[0028] S20, judging whether the digital flow rate unevenness meets the standard requirements: if it meets the standard requirements, the new nozzle is actually replaced and the process goes to step S13; if it does not meet the standard requirements, the process goes to step S18;

[0029] S21. Assembly completed.

[0030] Accordingly: an electronic device comprising:

[0031] one or more processors;

[0032] A storage device for storing one or more programs;

[0033] When the one or more programs are executed by the one or more processors, the one or more processors implement the digital assembly method.

[0034] Correspondingly: a computer-readable medium, wherein the computer-readable medium stores a computer program, and when the computer program is executed by a processor, the digital assembly method is implemented.

[0035] Compared with the prior art, the present invention has the following beneficial effects:

[0036] 1. The present invention conducts digital assembly tests through the digital twin model of the fuel main. If the prediction accuracy of all nozzles meets the threshold, a new nozzle is placed in the digital twin model of the fuel main to replace the nozzle with the largest flow deviation value, and digital assembly is performed and the digital flow unevenness is calculated. If the digital flow unevenness meets the requirements, the new nozzle is actually replaced, and the physical flow characteristic test is re-performed until the physical unevenness meets the requirements and the assembly is completed. The digital assembly of the fuel main can be realized on the cloud such as the server, which ultimately reduces the number of physical assembly and its corresponding test experiments, thereby reducing the waste of manpower and equipment, and improving the production efficiency, product constants and revenue of the existing fuel main assembly workshop.

[0037] 2. The digital twin model of the fuel manifold of the present invention includes a fuel nozzle coding network layer and a fuel manifold decoding network layer. The fuel nozzle coding network layer encodes the flow characteristics of m+n fuel nozzles, transmits them to the fuel manifold decoding network layer for decoding, and predicts the flow characteristics of the fuel manifold. The model trains and learns the correlation of all fuel manifolds according to the actual fuel manifold assembly process, and introduces the circumferential position of the fuel nozzle into the model through position coding. In addition, the attention mechanism can effectively find the characteristics of large flow nozzles in small flow nozzles, so the model has higher physical interpretability and model accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 It is a structural schematic diagram of the fuel main pipe of the present invention;

[0039] Figure 2 is a flow chart of the digital assembly method of the present invention;

[0040] Figure 3 This is a structural diagram of the digital twin model of the fuel main pipe of the present invention;

[0041] Figure 4 This is a diagram of the encoding structure of the Transformer model in Example 2 of the present invention. DETAILED DESCRIPTION

[0042] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Unless otherwise specified, the technical means used in the embodiments are conventional means well known to those skilled in the art.

[0043] It should be noted that the fuel manifold used in the digital assembly method for the fuel manifold of an aircraft engine of the present invention is a fuel manifold of an annular structure, and its structural schematic diagram is as follows: Figure 1 As shown, a number of fuel nozzles are symmetrically distributed on an annular pipe. The fuel is transmitted to each fuel nozzle through the fuel inlet in the annular pipe. Then the fuel nozzle atomizes and sprays the fuel through the swirler inside it, and then it is ignited and burned in the combustion chamber.

[0044] like Figure 2-3As shown, the present invention discloses a digital assembly method for an aircraft engine fuel main pipe, the core idea of ​​which is as follows: m small flow nozzles and n large flow nozzles with similar flow characteristics are selected to form a fuel main pipe; a physical flow characteristic test is performed on the fuel main pipe and the physical flow unevenness is calculated. If the physical flow unevenness meets the standard requirements, the assembly is completed; if the physical flow unevenness does not meet the standard requirements, a digital assembly test is carried out through a digital twin model of the fuel main pipe. If the prediction accuracy of all nozzles meets the threshold requirements, a new nozzle is placed in the digital twin model of the fuel main pipe to replace the nozzle with the largest flow deviation value, and digital assembly is performed and the digital flow unevenness is calculated. If the digital flow unevenness meets the requirements, the new nozzle is actually replaced, and the physical flow characteristic test is re-performed until the physical unevenness meets the requirements and the assembly is completed.

[0045] Furthermore, the calculation formula of the physical / digital flow non-uniformity k is as follows:

[0046]

[0047] In the formula, f′ min is the minimum physical / digital flow in the fuel main; f′ max is the maximum physical / digital flow in the fuel main. That is, the calculation method of digital flow unevenness is the same as that of physical flow unevenness. When calculating, the minimum physical flow is replaced by the minimum digital flow, and the maximum physical flow is replaced by the maximum digital flow.

[0048] Furthermore, the model prediction accuracy of the fuel manifold digital twin model can be expressed by the flow prediction accuracy of the fuel manifold nozzle, or by the flow unevenness prediction accuracy of the fuel manifold.

[0049] The calculation formula of the flow prediction accuracy ε of the fuel manifold nozzle is as follows:

[0050]

[0051] In the formula, is the flow prediction value of the fuel manifold digital twin model for m+n nozzles, f′ i is the physical flow value of the i-th nozzle.

[0052] The calculation formula for the prediction accuracy ε1 of the fuel main flow unevenness is as follows:

[0053]

[0054] Where k1 is the digital flow unevenness, that is, the flow unevenness predicted by the digital twin model of the fuel manifold; k0 is the physical flow unevenness, that is, the flow unevenness calculated by the physical flow characteristic test.

[0055] Furthermore, the flow deviation value σ of the fuel main pipe is calculated as follows:

[0056]

[0057] In the formula, is the flow prediction value of the i-th nozzle; m is the number of small flow nozzles; n is the number of large flow nozzles.

[0058] Furthermore, the fuel manifold digital twin model includes a fuel nozzle coding network layer and a fuel manifold decoding network layer. The fuel nozzle coding network layer encodes the flow characteristics of m+n fuel nozzles, transmits them to the fuel manifold decoding network layer for decoding, and predicts the flow characteristics of the fuel manifold.

[0059] Furthermore, the fuel nozzle coding network layer includes a secondary oil circuit coding network layer and a main-secondary oil circuit coding network layer, the secondary oil circuit coding network layer input is the secondary oil circuit performance data of the fuel nozzle, and the main-secondary oil circuit coding network layer is the main-secondary oil circuit performance data of the fuel nozzle; the fuel main pipe decoding network layer includes a secondary oil circuit decoding network layer and a main-secondary oil circuit decoding network layer, the secondary oil circuit decoding network layer and the main-secondary oil circuit decoding network layer respectively obtain coding information from the fuel nozzle coding network layer, and respectively obtain the predicted flow in the two cases of the fuel main pipe secondary oil circuit and the main-secondary oil circuit.

[0060] Furthermore, the auxiliary oil circuit decoding network layer and the main and auxiliary oil circuit decoding network layer both adopt a fully connected neural network layer, and transmit the encoded information to the fuel main pipe decoding network layer by assigning different weights.

[0061] Furthermore, the coding information of the auxiliary oil circuit coding network layer and the main and auxiliary oil circuit coding network layer are γ1, γ ２ , then the inputs of the auxiliary oil circuit decoding network layer and the main and auxiliary oil circuit decoding network layer are ωγ1+(1-ω)γ2, (1-ω)γ1+ωγ ２ ω is the weight coefficient, which ranges from 0 to 1.

[0062] The present invention also discloses an electronic device, comprising: one or more processors; a storage device for storing one or more programs; when the one or more programs are executed by the one or more processors, the one or more processors implement the digital assembly method for the fuel main pipe of an aircraft engine. The electronic device in this embodiment may include but is not limited to mobile terminals such as mobile phones, notebook computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc.

[0063] The present invention also discloses a computer readable medium, wherein the computer program is stored in the computer readable medium, and when the computer program is executed by a processor, the computer program implements a digital assembly method for an aircraft engine fuel manifold as described above. The embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a non-transitory computer readable medium, and the computer program includes a method for executing a process Figure 2 The program code of the method shown.

[0064] It should be noted that the computer-readable medium of the present disclosure may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0065] In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, apparatus, or device. In the present disclosure, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in combination with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0066] In some embodiments, the client and the server may communicate using any currently known or future developed network protocol such as HTTP (HyperText Transfer Protocol), and may be interconnected with any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network ("LAN"), a wide area network ("WAN"), an internet (e.g., the Internet), and a peer-to-peer network (e.g., an ad hoc peer-to-peer network), as well as any currently known or future developed network.

[0067] The computer-readable medium may be included in the electronic device, or may exist independently without being incorporated into the electronic device.

[0068] Embodiment 1: Digital assembly method of fuel main pipe

[0069] Several small flow nozzles and large flow nozzles of the same batch of fuel manifolds are tested. It is assumed that a fuel manifold is composed of m small flow nozzles and n large flow nozzles. This embodiment introduces the specific process of the digital assembly method of the fuel manifold for a fuel manifold with m+n nozzles, such as Figure 2 As shown, the following steps are included:

[0070] S11. m small flow nozzles with similar flow characteristics are selected from the fuel main pipes of the same batch. It should be noted that similar flow characteristics refer to close flow values ​​at the same working condition measurement point, and the specific degree of closeness can be determined according to the actual situation.

[0071] S12. Select n large flow nozzles with similar flow characteristics from the fuel main pipes of the same batch.

[0072] S13, assembling m small flow nozzles and n large flow nozzles into a fuel main pipe, and performing the first physical flow characteristic test of the fuel main pipe.

[0073] S14. Calculate the physical flow non-uniformity of the fuel main pipe.

[0074] Here, it is assumed that the flow rates of m+n fuel nozzles after assembly are f′1…f′ m , f′ m+1 …f′ m+n , the first 1, ..., m is the physical flow rate of the small flow nozzle, and the last m+1, ..., m+n is the physical flow rate of the large flow nozzle, among which the largest flow rate is f' max , where the minimum flow is f′ min , then the calculation formula of physical non-uniformity is:

[0075]

[0076] It should be noted that for different types of fuel mains, the physical unevenness and digital unevenness standard requirements they meet may be different, but the physical unevenness and digital unevenness standard requirements of the same fuel main should be the same.

[0077] S15. Determine whether the physical flow non-uniformity meets the standard requirements: if it meets the standard requirements, proceed to step S21; if it does not meet the standard requirements, proceed to step S16.

[0078] Specifically, assuming that the standard unevenness limit is δ, k≤δ must be satisfied for the performance of the assembled fuel main pipe to be qualified.

[0079] S16. Conduct digital assembly tests using the digital twin model of the fuel main to obtain digital flow characteristic data and calculate the model prediction accuracy.

[0080] This embodiment uses the flow prediction accuracy of the fuel manifold nozzle as the model prediction accuracy. For the digital assembly test, the flow characteristic parameters of the m small flow nozzles in step S11 and the n large flow nozzles in step S12 are used as the input of the digital twin model of the fuel manifold to predict the flow values ​​of each fuel nozzle of the fuel manifold after assembly. It is assumed that the flow prediction values ​​are The calculation formula for prediction accuracy is:

[0081]

[0082] If the prediction accuracy meets the threshold requirement, it proves that the digital twin model of the fuel main is credible, and the subsequent steps can be entered. If it does not meet the threshold requirement, the traditional method is continued to be used to replace the fuel nozzle and enter S13.

[0083] In particular, the threshold value here can be set according to actual conditions, and a recommended value of 5% is given here.

[0084] S17. Determine whether the flow prediction accuracy of the m+n nozzles meets the threshold requirement: if the flow prediction accuracy ε of the m+n nozzles all meet the threshold requirement, proceed to step S18; if the flow prediction accuracy of one or more nozzles does not meet the threshold requirement, randomly replace one or more nozzles and proceed to step S13.

[0085] It should be noted that the flow prediction accuracy here can be the maximum value of the flow prediction accuracy of all nozzles. When the maximum value does not meet the threshold requirement, the worker randomly replaces the fuel nozzle with a nozzle based on his or her own assembly experience. The flow prediction accuracy can also be compared with the flow prediction accuracy of all nozzles one by one. When one or more flow prediction accuracies do not meet the threshold requirement, the worker randomly replaces one or more fuel nozzles based on his or her own assembly experience.

[0086] The method for replacing the nozzle here can be: the worker randomly replaces the fuel nozzle based on his own assembly experience, and reassembles it into a fuel main pipe and performs a physical flow characteristic test.

[0087] S18. Calculate the flow deviation values ​​of m+n nozzles, and place a new nozzle into the digital twin model of the fuel main to replace the nozzle with the largest flow deviation value.

[0088] The calculation formula of the flow deviation value σ of the fuel main pipe is as follows:

[0089]

[0090] In the formula, is the flow prediction value of the i-th nozzle; m is the number of small flow nozzles; n is the number of large flow nozzles. Here, the position corresponding to the i-th nozzle is the target nozzle replacement position.

[0091] S19, placing the new fuel manifold in step S18 into the digital twin model of the fuel manifold, performing a digital assembly test, obtaining the predicted flow distribution of the fuel manifold, and calculating the digital flow unevenness. It should be noted that the replacement here does not refer to an actual replacement, but a virtual replacement of the fuel manifold in the server, that is, replacing the flow characteristic data of the corresponding nozzle of the digital twin model of the fuel manifold.

[0092] S20. Determine whether the digital flow unevenness meets the standard requirements: If it does, actually replace the new nozzle and go to step S13 to perform physical replacement and assembly, and conduct a physical flow characteristic test until the physical unevenness meets the standard requirements; if it does not meet the standard requirements, go to step S18.

[0093] S21. Assembly completed.

[0094] Example 2: Digital Twin Model of Fuel Manifold

[0095] Before introducing the digital twin model of the fuel manifold, it is necessary to briefly summarize the performance test process of the fuel nozzle and the fuel manifold. First, for a single fuel nozzle, its flow characteristic test is to test the fuel outflow rate and spray angle of the nozzle under different fuel pressures. It should be noted that under low pressure, the fuel nozzle only opens the secondary oil circuit; while under high pressure, the fuel nozzle opens both the main and secondary oil circuits. Therefore, the performance data of the fuel nozzle is assumed to be [[f l ,θ l ],[f h ,θ h ]], note here [f l ,θ l ] and [f h,θ h ] are the fuel flow rate and spray angle data when the auxiliary oil circuit is opened and the main and auxiliary oil circuits are opened at the same time, and they are data vectors containing several pressure measurement points.

[0096] After the fuel nozzle is assembled, due to the change in structure, the flow rate of the fuel nozzle will change under the same pressure measurement point conditions. Similarly, the performance data of the fuel nozzle in the fuel main pipe under low and high fuel pressures is [f′ l , f′ h ].

[0097] In addition to the main and auxiliary oil circuits, there are two types of fuel with large and small flow rates in the assembly process of the fuel manifold. The present invention is based on the test conditions, assembly conditions and actual use conditions of the above fuel manifold and its fuel nozzles to construct the following Figure 3 The digital twin of the fuel manifold is shown. Figure 3 It can be seen that the whole model is divided into two parts: the fuel nozzle encoder network and the fuel manifold decoder network. The fuel nozzle encoder network mainly encodes the flow characteristics of m+n fuel nozzles and transmits them to the fuel manifold decoder for decoding, so as to predict the flow characteristics of the fuel manifold.

[0098] In particular, the fuel nozzle encoder network layer mainly encodes the flow characteristics of each fuel nozzle on the fuel main pipe before installation, and passes it to the fuel main pipe decoding network layer, which is composed of the auxiliary oil circuit encoding network layer and the main and auxiliary oil circuit encoding network layer. Its input is composed of the fuel nozzle auxiliary oil circuit performance data and the fuel nozzle main and auxiliary oil circuit performance data, that is, [f l ,θ l ] and [f h ,θ h The fuel manifold decoder mainly decodes the flow characteristics of the fuel nozzle, which includes two parts: the auxiliary oil circuit decoding network layer and the main and auxiliary oil circuit decoding network layer. The two network layers obtain the coding information from the fuel nozzle coding network layer respectively, so as to obtain the flow prediction distribution of the auxiliary oil circuit of the fuel manifold and the main and auxiliary oil circuit, that is, [f′ l , f′ h ]. In particular, [f l ,θ l ] and [f h ,θ h ] is the input of the model, [f′ l , f′ h ] is the output of the model.

[0099] In particular, the auxiliary oil circuit decoding network layer and the main and auxiliary oil circuit decoding network layer are fully connected neural network layers, respectively, and their inputs are composed of the auxiliary oil circuit encoding network layer and the main and auxiliary oil circuit encoding network layer. Specifically, the encoding information is passed to the two decoding network layers by assigning different weights. Assuming that the encoding information of the auxiliary oil circuit encoding network layer and the main and auxiliary oil circuit encoding network layer is γ1 and γ2, the inputs of the auxiliary oil circuit decoding network layer and the main and auxiliary oil circuit decoding network layer are ωγ1+(1-ω)γ2 and (1-ω)γ1+ωγ2. ω is a weight coefficient, and its value is 0-1. Here, ω is recommended to take a large value. In the decoding network layer, the encoder of the corresponding oil circuit plays a decisive role. For the preferred technical solution, it is recommended that ω take 0.9 in this embodiment.

[0100] In particular, in the fuel main pipe, the performance of each fuel nozzle is interrelated, thus forming the overall flow characteristics and unevenness of the fuel main pipe. Compared with the traditional neural network, the self-attention method of calculating the correlation between each input element is more suitable for the scenario described in the present invention. Therefore, the encoding network layer of the auxiliary oil circuit and the main auxiliary oil circuit of the present invention adopts the encoding layer in Transform, and its structure is as follows: Figure 4 As shown in the figure, the performance parameters of m+n fuel nozzles are introduced into the circumferential position information of the fuel nozzle through the nozzle position encoding layer, and then enter the multi-head attention layer, normalization layer, feedforward neural network layer and the second normalization layer respectively to complete the encoding of the fuel nozzle flow characteristics.

[0101] The embodiments described above are only descriptions of the preferred modes of the present invention and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various deformations, modifications, and substitutions made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. A digital assembly method for an aircraft engine fuel manifold, characterized in that: Select m small-flow nozzles and n large-flow nozzles to form a fuel manifold; conduct a physical flow characteristic test on the fuel manifold and calculate the physical flow unevenness. If the physical flow unevenness meets the standard requirements, the assembly is completed; if the physical flow unevenness does not meet the standard requirements, a digital assembly test is carried out through the digital twin model of the fuel manifold. If the model prediction accuracy meets the threshold requirement, a new nozzle is placed in the digital twin model of the fuel manifold to replace the nozzle with the largest flow deviation value. A digital assembly test of the fuel manifold is carried out and the digital flow unevenness is calculated. If the digital flow unevenness meets the requirements, the new nozzle is actually replaced and the physical flow characteristic test is carried out again until the physical unevenness meets the requirements and the assembly is completed.

2. A digital assembly method for an aircraft engine fuel manifold according to claim 1, characterized in that: The calculation formula of the physical / digital flow non-uniformity k is as follows: In the formula, f′ min is the minimum physical / digital flow in the fuel main; f′ max is the maximum physical / digital flow in the fuel rail.

3. The digital assembly method for an aircraft engine fuel manifold according to claim 1, characterized in that: The calculation formula of the flow deviation value σ of the fuel main pipe is as follows: In the formula, is the flow prediction value of the i-th nozzle; m is the number of small flow nozzles; n is the number of large flow nozzles.

4. The digital assembly method for an aircraft engine fuel manifold according to claim 1, characterized in that: The fuel manifold digital twin model includes a fuel nozzle coding network layer and a fuel manifold decoding network layer. The fuel nozzle coding network layer encodes the flow characteristics of m+n fuel nozzles, transmits them to the fuel manifold decoding network layer for decoding, and predicts the flow characteristics of the fuel manifold.

5. A digital assembly method for an aircraft engine fuel manifold according to claim 4, characterized in that: The fuel nozzle coding network layer includes a secondary oil circuit coding network layer and a main and secondary oil circuit coding network layer. The input of the secondary oil circuit coding network layer is the secondary oil circuit performance data of the fuel nozzle, and the input of the main and secondary oil circuit coding network layer is the main and secondary oil circuit performance data of the fuel nozzle; And / or, the fuel main line decoding network layer includes a secondary oil circuit decoding network layer and a main and secondary oil circuit decoding network layer. The secondary oil circuit decoding network layer and the main and secondary oil circuit decoding network layer obtain coding information from the fuel nozzle coding network layer to obtain the predicted flow in the fuel main line secondary oil circuit and the main and secondary oil circuit respectively.

6. A digital assembly method for an aircraft engine fuel manifold according to claim 5, characterized in that: The auxiliary oil circuit decoding network layer and the main and auxiliary oil circuit decoding network layer both adopt a fully connected neural network layer, and transmit the encoded information to the fuel main pipe decoding network layer by assigning different weights.

7. A digital assembly method for an aircraft engine fuel manifold according to claim 6, characterized in that: The coding information of the auxiliary oil circuit coding network layer and the main and auxiliary oil circuit coding network layer are γ1 and γ2 respectively, then the inputs of the auxiliary oil circuit decoding network layer and the main and auxiliary oil circuit decoding network layer are ωγ1+(1-ω)γ2 and (1-ω)γ1+ωγ2 respectively; ω is the weight coefficient, and its value is 0-1.

8. The digital assembly method for an aircraft engine fuel manifold according to claim 1, characterized in that: The steps include: S11, selecting m small flow nozzles with similar flow characteristics; S12, selecting n large flow nozzles with similar flow characteristics; S13, assembling m small flow nozzles and n large flow nozzles into a fuel main pipe, and performing a physical flow characteristic test of the fuel main pipe; S14, calculating the physical flow unevenness of the fuel main pipe; S15, judging whether the physical flow non-uniformity meets the standard requirements: if it meets the standard requirements, proceeding to step S21; if it does not meet the standard requirements, proceeding to step S16; S16. Conduct digital assembly tests using the digital twin model of the fuel main to obtain digital flow characteristic data and calculate the model prediction accuracy; S17, judging whether the model prediction accuracy meets the threshold requirement: if the model prediction accuracy meets the threshold requirement, proceed to step S18; if the model prediction accuracy does not meet the threshold requirement, randomly replace the nozzle and proceed to step S13; S18, calculating the flow deviation values ​​of m+n nozzles, and placing a new nozzle into the digital twin model of the fuel manifold to replace the nozzle with the largest flow deviation value; S19, performing a digital assembly test on the new fuel main pipe in step S18 and calculating the digital flow unevenness; S20, judging whether the digital flow rate non-uniformity meets the standard requirements: if it meets the standard requirements, the new nozzle is actually replaced and the process goes to step S13; If the standard requirements are not met, proceed to step S18; S21. Assembly completed.

9. An electronic device, characterized in that: include: one or more processors; A storage device for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the digital assembly method according to any one of claims 1 to 8.

10. A computer readable medium storing a computer program, characterized in that: When the computer program is executed by a processor, the digital assembly method according to any one of claims 1 to 8 is implemented.

Images