Reciprocating engine fault diagnosis method based on multi-source angular domain envelope-operating characteristic joint diagram

Abstract

A reciprocating engine fault diagnosis method based on a multi-source angular domain envelope-operating characteristic joint diagram comprises the steps that firstly, vibration signals of the same key positions of all cylinders of an engine and key phase signals of a crankshaft are collected, and all pulse triggering positions in the key phase signals are obtained; secondly, envelope processing and truncation processing are carried out on the vibration signals, an angular domain abscissa sequence is constructed, and conversion of the vibration signals from a time domain to an angular domain is achieved; sequencing the signals of all the measuring points according to the ignition sequence of all the cylinders of the engine; and finally, in combination with the working characteristics of the engine, constructing a multi-source angular domain envelope-working characteristic joint diagram, and marking the working state of the part corresponding to each impact moment. According to the method, deep fusion of the multi-source vibration signals, the angular domain information and the engine working characteristic information is realized, the engine vibration impact characteristics can be quickly and accurately associated with the working states of the key parts, and the engine fault analysis efficiency and accuracy are remarkably improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of reciprocating engine fault diagnosis, and particularly relates to a reciprocating engine fault diagnosis method based on a multi-source angular domain envelope - working characteristic joint diagram. Background Technique

[0002] As the core power source for driving equipment operation, reciprocating engines play a crucial role in many key fields such as petrochemical industry and national defense industry. The quality of their operation directly affects the overall economic benefits of enterprises. Compared with conventional machinery, reciprocating engines not only have a precise structure and high integration, but also have many excitation sources. Moreover, the vibration responses of their various components show a high degree of coupling and instantaneous impact characteristics, and are mixed with significant non-stationary and non-periodic signal components, further increasing the complexity of fault analysis. Therefore, in an environment of multiple interferences, how to quickly and accurately identify and extract the vibration impact characteristics of reciprocating engines has become a major technical problem in the field of fault monitoring and diagnosis of such equipment. Scholars at home and abroad have carried out a series of studies on this, aiming to explore more effective fault diagnosis methods to improve the efficiency and accuracy of fault analysis.

[0003] Currently, most reciprocating engine fault diagnosis methods are based on the extraction and analysis of impact characteristics from engine vibration signals. The conventional approach mainly focuses on signal acquisition, decomposition, and feature extraction (Li Guobin, Guan Delin, etc. Research on the extraction of diesel engine vibration signal characteristics based on wavelet packet transform and singular value decomposition [J], Vibration and Shock, 2011, (08): 149 - 152). Its main steps are as follows: First, obtain vibration signals through precise sampling technology and apply wavelet transform for noise reduction to improve signal quality; then, use singular value decomposition technology to construct a feature parameter system to extract key information from complex signals; finally, comprehensively infer the working state of the reciprocating engine based on these feature parameters. However, such conventional methods have significant limitations: On the one hand, it ignores the internal relationship between multiple cylinders of the engine, especially the key timing information of the ignition sequence, resulting in the difficulty of accurately matching the impact characteristics of the engine vibration signal with the ignition action, making the corresponding relationship between these characteristics and the actual operating state of the engine unclear; on the other hand, this type of method focuses too much on the signal noise reduction and decomposition links, but fails to clearly combine the decomposed features with the specific working cycle characteristics of the engine, making the feature extraction process lack pertinence and affecting the accuracy of fault diagnosis.

[0004] Regarding the intricate structure of reciprocating engines and the inherent characteristics of their multi-source vibration responses, how to construct a map that can closely integrate multi-source vibration signals with the working cycle characteristics of the engine, and then quickly and clearly describe the precise correspondence between vibration shock characteristics and the actions of equipment components, has become the core and key problem in improving the efficiency and accuracy of reciprocating engine fault diagnosis. Summary of the Invention

[0005] To overcome the above-mentioned shortcomings of the prior art, the purpose of the present invention is to provide a reciprocating engine fault diagnosis method based on a multi-source angular domain envelope - working characteristic joint diagram. By using multi-source information such as vibration signals, key phase signals, and engine working characteristics, a multi-source angular domain envelope - working characteristic joint diagram is constructed, realizing the rapid and accurate correlation between the engine vibration shock characteristics and the working states of key components, and providing important technical support for engine condition assessment and fault diagnosis.

[0006] To achieve the above purpose, the technical solution adopted by the present invention is as follows:

[0007] A reciprocating engine fault diagnosis method based on a multi-source angular domain envelope - working characteristic joint diagram, comprising the following steps:

[0008] 1) Set the signal sampling frequency as f s , collect the vibration signals of the same key parts of each cylinder of the engine and the key phase signal of the crankshaft. The key part is the side of the cylinder head or cylinder block; store the data in the matrix U i×j , where: i = 1, 2, 3,..., M, j = 1, 2, 3,..., N, M represents the channel number, the first M - 1 channels are all vibration signals, the Mth channel is the key phase signal, and N represents the signal sampling length;

[0009] 2) Take the signal of the Mth row in the matrix U i×j , and use the key phase signal initial phase point adaptive extraction method to obtain the positions corresponding to all pulse trigger moments in this key phase signal. The result is stored in the array O[k], where k = 1, 2, 3,..., L;

[0010] 3) Perform envelope processing on the first M - 1 rows of data in the matrix U i×j , save the envelope signal to the matrix X, and perform key phase truncation on all envelope signals. The specific operation: take all the data from the O[1]th column to the last column in X to obtain the sub-matrix Y a×b = X[:, O[1]:N], where: a = 1, 2, 3,..., M - 1, b = 1, 2, 3,..., N - O[1]+1;

[0011] 4) Calculate the average sampling length N T within one cycle, and the process is as follows: take the data lengths of n cycles to calculate N T= round[(O[n + 1] - O[1]) / n], where the function round represents rounding to the nearest integer; calculate the abscissa sequence of the angular domain The process is as follows: Calculate the sampling interval of the angular domain Then, through the formula Obtain the abscissa sequence of the angular domain

[0012] 5) According to the ignition sequence of each cylinder of the engine, use the array h[g] to record the cylinder numbers in sequence, where: g = 1, 2, 3,..., M - 1, take h[1] = 1, and rearrange the data of each row in the sub - matrix Y a×b in the order of the cylinder ignition sequence to obtain a new matrix Z = Y[h[g], :];

[0013] 6) Draw the original multi - source angular domain envelope - working characteristic joint diagram. The specific process is as follows: First, establish a coordinate system with the abscissa of the angular domain as the horizontal axis and the signal envelope amplitude as the vertical axis; then, draw the waveform diagrams of the signals of each row in the matrix Z from top to bottom along the vertical axis, and the horizontal axis centers of each waveform are arranged at equal intervals, with the interval being P; finally, based on the working cycle characteristics of the engine, mark the working states corresponding to each impact at the angular domain positions corresponding to the signals in the first row of the matrix Z, and establish the mapping relationship between the impact moment and the engine working state;

[0014] 7) Draw the multi - source angular domain envelope - working characteristic joint diagram after phase compensation. The specific process is as follows: According to the principle that the working states of each cylinder are the same at the same angular domain position, translate the signals of each row in the matrix Z to obtain the matrix Z′, and then repeat step 6) to obtain the multi - source angular domain envelope - working characteristic joint diagram after phase compensation;

[0015] 8) Use the original or phase - compensated multi - source angular domain envelope - working characteristic joint diagram to compare and analyze the vibration impact amplitudes when the same actions of each component of the engine occur, identify the abnormal impacts among them, locate the fault occurrence position according to the angular domain phase characteristics of the abnormal impacts, and analyze and evaluate the health state of the engine.

[0016] The determination principle of the interval P in step 6) is as follows: Calculate the maximum value of all elements in the matrix Z, and take P = α × max[Z], where the value range of α is 1.0 - 2.0.

[0017] The specific method of translating the signals of each row in step 7) is as follows: Take s = 1, 2,..., M - 1, calculate according to N0[s] = N T ×(s - 1) / 4 to obtain the translation amount of each row of the signal to the left, and translate the signals of all rows in the matrix Z to obtain the matrix Z′.

[0018] Compared with the prior art, the beneficial effects of the present invention are:

[0019] Through the deep fusion of multi-source vibration signals, angular domain information, and engine operating characteristic information, the present invention constructs a multi-source angular domain envelope-operating characteristic joint diagram, marks the operating states of components corresponding to each impact moment, and establishes a mapping relationship between the impact moment and the engine operating state. Therefore, this method can quickly and accurately associate the engine vibration impact characteristics with the operating states of key components, significantly improving the efficiency and accuracy of engine fault analysis and providing reliable technical support for engine fault diagnosis. Description of the Drawings

[0020] Figure 1 is a flowchart of an embodiment of the present invention.

[0021] Figure 2 is a layout diagram of vibration and key phase measurement points on the engine cylinder head in the embodiment.

[0022] Figure 3 is the original multi-source angular domain envelope-operating characteristic joint diagram of the cylinder head in the embodiment.

[0023] Figure 4 is the multi-source angular domain envelope-operating characteristic joint diagram of the cylinder head after phase compensation in the embodiment. Detailed Embodiment

[0024] The present invention will be further elaborated in detail below in conjunction with the embodiments and the drawings:

[0025] Embodiment: Analyze the G3608 type reciprocating gas engine of a certain gas production plant. This equipment is provided by Caterpillar Inc. of the United States, with the equipment model G3608TALE, a rated power of 1767KW, a designed speed of 1000rpm, and is equipped with 8 cylinders. Figure 2 For the layout of vibration and key phase measurement points of this equipment, the vibration measurement points are arranged as follows: A vibration measurement point is arranged on the intake side of each cylinder head of the engine, numbered J1 to J8 in sequence, and the acceleration sensor model is CTC AC192; the key phase measurement points are arranged as follows: When the engine is turned to the top dead center of compression of cylinder No. 1, a reflective tape is pasted on the surface of the coupling at the driving end of the engine, and the photoelectric sensor is aligned with the reflective tape to collect the key phase signal. The photoelectric sensor model is ROS-P. During the data acquisition process, the operating speed of the equipment is stable at 950rpm.

[0026] Refer to Figure 1 , a reciprocating engine fault diagnosis method based on a multi-source angular domain envelope-operating characteristic joint diagram, includes the following steps:

[0027] 1) Set the signal sampling frequency to f s= 12,800 Hz, collect the vibration signals of each cylinder head of the engine and the key-phase signal of the crankshaft, and store the data in matrix U i×j where: i = 1, 2, 3, …, M, j = 1, 2, 3, …, N, M = 9 represents the channel number, the first 8 channels are all vibration signals, the 9th channel is the key-phase signal, and the signal sampling length is N = 256,000;

[0028] 2) Take the signal of the 9th row in matrix U i×j and use the adaptive extraction method of the key-phase signal initial phase point to obtain the positions corresponding to all pulse trigger moments in this key-phase signal, and store the results in array O[k], where k = 1, 2, 3, ..., 299, and the results are shown in Table 1;

[0029] Table 1 Positions corresponding to pulse trigger moments

[0030] k 1 2 3 … 298 299 O[k] 801 1656 2509 … 254452 255305

[0031] 3) Perform envelope processing on the first 8 rows of data in matrix U i×j and save the envelope signals to matrix X. Perform key-phase truncation on all envelope signals. The specific operation is as follows: Take all the data from the 801st column to the last column in X to obtain sub-matrix Y a×b = X[:, 801:25600], where: a = 1, 2, 3, …, 8, b = 1, 2, 3, …, 24,800;

[0032] 4) Calculate the average sampling length N within one period T , and the process is as follows: Take the data lengths of 8 periods to calculate N T = round[(7632 - 801) / 8] = 854; Calculate the angular domain abscissa sequence The process is as follows: First, calculate the angular domain sampling interval Then, through the formula obtain the abscissa sequence of the angular domain

[0033] 5) According to the ignition sequence of each cylinder of the engine, use array h[g] to record the cylinder numbers in turn, and the results are shown in Table 2, where: g = 1, 2, 3, ..., 8, take h[1] = 1, and rearrange the data of each row in matrix Y according to the cylinder ignition sequence to obtain a new matrix Z = Y[h[g], :];

[0034] Table 2 Ignition sequence array of each cylinder of the engine

[0035] g 1 2 3 4 5 6 7 8 h[g] 1 6 2 5 8 3 7 4

[0036] 6) Draw the original multi-source angular domain envelope - working characteristic joint diagram. The specific process is as follows: First, use the angular domain abscissa A coordinate system is established with the horizontal axis as the horizontal axis and the signal envelope amplitude as the vertical axis; then, the waveforms of the signals in each row of the matrix Z are drawn from top to bottom along the vertical axis, and the centers of the horizontal axes of the waveforms are arranged at equal intervals, with a spacing of P = 400 m / s. 2 Finally, based on the engine working cycle characteristics, the first row of signals in the matrix Z corresponds to the angular domain position, and the working state corresponding to each impact is marked to establish a mapping relationship between the impact moment and the engine working state, such as Figure 3 As shown, Figure 3 The envelope signals of 8 vibration measurement points of the engine cylinder in 4 working cycles are shown. The signals are arranged in the order of ignition of the engine cylinder, namely J1-J6-J2-J5-J8-J3-J7-J4. In a single working cycle of each cylinder of the engine, the envelope signals of each measurement point show obvious multi-impact phenomenon. The main impact signals correspond to the working states of the intake valve seat and exhaust valve seat, among which the intake valve seat impact amplitude is the most prominent. For two adjacent envelope signals, the phase difference of the impact signals caused by the same component is 90°, which is consistent with the ignition phase difference of each cylinder. It can be seen that Figure 3 It can fully integrate the engine's multi-source vibration signals, angular domain information, and operating characteristics information, allowing vibration engineers to quickly and accurately associate engine vibration impact characteristics with the operating conditions of key components, improving the efficiency and accuracy of engine fault analysis;

[0037] 7) Draw the multi-source angular domain envelope-operating characteristic joint diagram after phase compensation. The specific process is as follows: According to the principle that the working state of each cylinder is consistent at the same angular domain position, the row signals in the matrix Z are translated to obtain the matrix Z′, and then step 6) is repeated to obtain the multi-source angular domain envelope-operating characteristic joint diagram after phase compensation, as shown in Figure 4 As shown in the figure, the main difference from the original multi-source angular domain envelope-operating characteristic joint diagram is that the joint diagram after phase compensation eliminates the phase difference between the impact signals, which indicates that the working states of the cylinders corresponding to the vibration signals of each measuring point at the same angular domain position are the same. Therefore, Figure 4 It is convenient to compare the vibration impact values ​​of 8 cylinders in the same working state. For example, the impact of exhaust valve seating of 1# and 5# cylinders is more obvious than that of other cylinders.

[0038] 8) Using the original multi-source angular domain envelope-working characteristic joint diagram, the impact amplitude of the intake valve of each cylinder at the moment of seating is compared and analyzed: the vibration impact amplitude of the 1#, 5#, and 8# cylinders is significantly larger, reflecting that the changes in the valve stem protrusion of the intake valves of these cylinders are significantly larger; using the multi-source angular domain envelope-working characteristic joint diagram after phase compensation, the impact amplitude of the exhaust valve of each cylinder at the moment of seating is compared and analyzed: the vibration impact amplitude of the 1# and 5# cylinders is significantly larger, reflecting that the changes in the valve stem protrusion of the exhaust valves of these cylinders are significantly larger.

[0039] Disassembly verification results: The valve stem protrusion data of the intake valves of cylinders 1#, 5#, and 8# are shown in Table 3, indicating that the intake valves of the cylinders are worn more significantly; the valve stem protrusion data of the exhaust valves of cylinders 1# and 5# are shown in Table 4, indicating that the exhaust valves of the cylinders are worn more significantly.

[0040] Table 3 Detection results of valve stem protrusion of intake valves (unit: inch)

[0041]

[0042] Table 4 Detection results of valve stem protrusion of exhaust valves (unit: inch)

[0043]

Claims

1. A reciprocating engine fault diagnosis method based on a multi-source angular domain envelope-operating characteristic joint diagram, characterized in that: The following steps are involved: 1) Set the signal sampling frequency to f s , collect the vibration signals of the same key parts of each cylinder of the engine and the key phase signals of the crankshaft, the key parts are the cylinder head or the side of the cylinder block; store the data in the matrix U i×j In which: i = 1, 2, 3, ..., M, j = 1, 2, 3, ..., N, M represents the channel number, the first M-1 channels are vibration signals, the Mth channel is the key phase signal, and N represents the signal sampling length; 2) Take the matrix U i×j The Mth row signal in the key phase signal is extracted by using the key phase signal initial phase point adaptive extraction method to obtain the positions corresponding to all pulse triggering moments in the key phase signal, and the results are stored in the array O[k], k = 1, 2, 3, ..., L; 3) For the matrix U i×j The first M-1 rows of data are all processed by envelope processing, and the envelope signals are saved in the matrix X. All envelope signals are truncated by key phase. The specific operation is: take all the data from the O[1]th column to the last column in X to obtain the submatrix Y a×b =X[:,O[1]:N], where: a=1,2,3,…,M-1, b=1,2,3,…,NO[1]+1; 4) Calculate the average sampling length N within a cycle T , the process is as follows: take the data length of n cycles to calculate N T =round[(O[n+1]-O[1]) / n], the function round means rounding to the nearest integer; calculate the horizontal coordinate sequence of the angular domain The process is as follows: Calculate the angular domain sampling interval Then through the formula Get the horizontal coordinate sequence of the angle domain 5) According to the ignition sequence of each cylinder of the engine, use the array h[g] to record the cylinder numbers in sequence, where: g = 1, 2, 3, ..., M-1, take h[1] = 1, and the submatrix Y a×b The rows of data in are rearranged according to the cylinder ignition order to obtain a new matrix Z=Y[h[g],:]; 6) Draw the original multi-source angular domain envelope-operating characteristic joint diagram. The specific process is as follows: First, take the angular domain horizontal coordinate A coordinate system is established with the horizontal axis as the horizontal axis and the signal envelope amplitude as the vertical axis; then, the waveforms of the signals in each row of the matrix Z are drawn from top to bottom along the vertical axis, and the centers of the horizontal axes of the waveforms are arranged at equal intervals, with the interval being P; finally, based on the engine working cycle characteristics, the working state corresponding to each impact is marked at the angular domain position corresponding to the first row of signals in the matrix Z, and a mapping relationship between the impact moment and the engine working state is established; 7) Draw a joint diagram of multi-source angular domain envelope-operating characteristics after phase compensation. The specific process is as follows: according to the principle that the working state of each cylinder is consistent at the same angular domain position, each row signal in the matrix Z is translated to obtain the matrix Z′, and then repeat step 6) to obtain the joint diagram of multi-source angular domain envelope-operating characteristics after phase compensation; 8) Using the original or phase-compensated multi-source angular domain envelope-operating characteristic joint diagram, compare and analyze the vibration impact amplitudes of the engine components during the same operation, identify abnormal impacts, locate the fault location based on the angular domain phase characteristics of the abnormal impact, and analyze and evaluate the health status of the engine.

2. The method according to claim 1, characterized in that The principle for determining the spacing P in step 6) is as follows: calculate the maximum value of all elements in the matrix Z, and take P = α×max[Z], where the value range of α is 1.0 to 2.

0.

3. The method according to claim 2, characterized in that The specific method of shifting each row signal in step 7) is as follows: Take s = 1, 2, ... M-1, and press N0[s] = N T ×(s-1) / 4 is calculated to obtain the leftward shift of each row of signals, and the signals of all rows in the matrix Z are shifted to obtain the matrix Z′.

Images