CN113887770B - Aero-engine life-span maintenance decision optimization algorithm based on problem decoupling - Google Patents
Aero-engine life-span maintenance decision optimization algorithm based on problem decoupling Download PDFInfo
- Publication number
- CN113887770B CN113887770B CN202010625099.XA CN202010625099A CN113887770B CN 113887770 B CN113887770 B CN 113887770B CN 202010625099 A CN202010625099 A CN 202010625099A CN 113887770 B CN113887770 B CN 113887770B
- Authority
- CN
- China
- Prior art keywords
- maintenance
- life
- engine
- decision
- value
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000012423 maintenance Methods 0.000 title claims abstract description 187
- 238000005457 optimization Methods 0.000 title claims abstract description 29
- 239000002245 particle Substances 0.000 claims abstract description 49
- 238000012545 processing Methods 0.000 claims abstract description 3
- 230000008439 repair process Effects 0.000 claims description 33
- 230000006978 adaptation Effects 0.000 claims description 24
- 238000000034 method Methods 0.000 claims description 24
- 230000003044 adaptive effect Effects 0.000 claims description 13
- 101150067055 minC gene Proteins 0.000 claims description 4
- 230000009191 jumping Effects 0.000 claims description 3
- 210000004027 cell Anatomy 0.000 description 9
- 238000009826 distribution Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 4
- 238000002474 experimental method Methods 0.000 description 4
- 238000011160 research Methods 0.000 description 4
- 238000011084 recovery Methods 0.000 description 3
- 238000011179 visual inspection Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000009827 uniform distribution Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 210000005056 cell body Anatomy 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/06—Multi-objective optimisation, e.g. Pareto optimisation using simulated annealing [SA], ant colony algorithms or genetic algorithms [GA]
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Economics (AREA)
- Evolutionary Computation (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Computer Hardware Design (AREA)
- Life Sciences & Earth Sciences (AREA)
- Quality & Reliability (AREA)
- General Business, Economics & Management (AREA)
- Operations Research (AREA)
- Marketing (AREA)
- Entrepreneurship & Innovation (AREA)
- Game Theory and Decision Science (AREA)
- Geometry (AREA)
- Development Economics (AREA)
- Health & Medical Sciences (AREA)
- Tourism & Hospitality (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Combined Controls Of Internal Combustion Engines (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to the technical field of maintenance strategies of aeroengines, in particular to a problem decoupling-based aeroengine whole-service-life maintenance decision optimization algorithm capable of effectively reducing the maintenance cost of an engine, which is characterized in that decision variables of the whole-service-life maintenance decision problem of the engine are divided into three groups, wherein the first group is the maintenance times of the whole service life of the engine, the second group is the maintenance time of the past time, and the third group is whether each unit body is overhauled and whether each service-life part is replaced or not when the past maintenance is carried out; the first group of decision variables are subjected to traversal processing, the second group of decision variables are solved by adopting a particle swarm optimization algorithm, and the third group of decision variables are solved by adopting an engine maintenance decision algorithm based on particle swarm optimization.
Description
Technical field:
the invention relates to the technical field of aero-engine maintenance strategies, in particular to an aero-engine whole-service-life maintenance decision optimization algorithm based on problem decoupling, which can effectively reduce the maintenance cost of an engine.
The background technology is as follows:
with the development of economy and the continuous improvement of the technical level, air transportation has become the most viable industry. Civil aviation engines are hearts of aircraft and complex equipment with extremely high safety requirements. In order to ensure the operational safety of the engine, it is necessary to constantly monitor and repair it. The maintenance cost of the engine is high. In order to reduce the maintenance costs of the engine, it is necessary to make a reasonable maintenance decision, i.e. to determine the engine maintenance timing and maintenance working range.
In the aspect of engine maintenance time determination, most of the current researches are to build an optimized model taking maintenance time as a decision variable, and obtain the maintenance time by solving the model. The determination of the engine maintenance work range mainly includes the determination of the unit body maintenance level and the determination of the replacement of the life part. At present, the research on the aspect of engine maintenance decision is often performed by dividing the engine maintenance time determination and the maintenance work range determination, but in practice, the maintenance time affects the maintenance work range, and the maintenance work range also affects the maintenance time. Meanwhile, most of the current researches only pay attention to the determination of single maintenance time and maintenance working range of the engine, but in practice, the engine can be subjected to multiple repair in the whole life, and the maintenance time and the maintenance working range of each repair can influence the subsequent maintenance time and maintenance working range, so that the whole life maintenance cost can be influenced.
The maintenance decision problem of the whole service life of the engine is to determine whether each unit body is overhauled and each service life part is replaced or not in the maintenance times, the maintenance time and maintenance time in the whole service life of the engine under the constraint of the limit of the maintenance interval of the engine, the limit of the overhaul interval of the unit body, the limit of the service life parts and the like.
If the maintenance times in the whole life of the engine are recorded as m. To reduce the value range of m, the maximum repair interval I of the engine can be determined according to historical repair data or engineer experience max And minimum repair interval I min Further, the minimum maintenance times m is calculated min And maximum maintenance number m max I.e. m min ≤m≤m max . The maintenance time of the engine is recorded as T 1 ,T 2 ,...,T m And I min ≤T k -T k-1 ≤I max Where k=2, 3,..m. The maintenance level of the unit body can be generally divided into: visual inspection, minimal repair, performance recovery, and overhaul. Wherein, visual inspection and minimum repair have no influence on the unit body performance, and the maintenance cost is not greatly different; performance recovery and overhaul can recover unit body performance, and the maintenance cost is not greatly different. To simplify the problem, visual inspection and minimal repair are combined, collectively referred to as minor repair; performance recovery and overhaul are combined, collectively referred to as overhaul. The number of the unit bodies is denoted as p, and the unit bodies are denoted as M i I is more than or equal to 1 and p is more than or equal to p. Based on historical repair data or engineer experienceTo determine the maximum repair interval of the unit bodyThe maintenance work scope determination instruction file issued by the OEM of the engine is also given with the optimal overhaul interval of the unit body +.> Generally determined by the reliability of the parts and taking into account factors such as the opportunity for maintenance [14] . The service time of the unit body after overhaul is recorded as +.>After the unit body is overhauled, the service time is initialized to 0. The repair cost of the unit body is marked as +.>The minor repair cost of the unit body is marked as +.>The unit body is overhauled in advance, so that the performance of the unit body is wasted; the unit cell delay overhaul may result in an increase in unit cell overhaul cost. Therefore, punishment is performed on the unit cell at the time of overhaul at a non-optimal overhaul interval, which is marked as +.>The calculation is performed by using the formula (1). />When the engine is used to the full life period T of the engine lim The engine also has a residual value noted V. The residual value V is determined by the residual value V of the life part LLP And the residual value V of the unit body MOD Constitution, wherein V MOD Can be calculated according to formula (2). The total maintenance cost of all the unit bodies in the whole life cycle is recorded as C MOD As shown in formula (3).
Wherein->Is the unit M at the kth maintenance i Maintenance policy to be taken,/->0 represents a minor repair and 1 represents a major repair.
The lifetime piece is a part having lifetime limitation. Within the life limits, life pieces are extremely reliable and generally do not fail. The operational safety of the engine is affected if the life piece exceeds the life limit. It is assumed that the lifetime piece will not fail within the lifetime limit. The number of life parts is denoted as n, and the number of life parts is denoted as L j J is more than or equal to 1 and n is more than or equal to n. Life part L j The life limit of (2) is recorded asThe service life of the service life part is recorded as +.>The remaining life is recorded as->After replacement of the life part, its service time is initialized to 0. The replacement cost of the life part is recorded as +.>When the engine is used to the full life period T of the engine lim Residual value V of life part LLP As shown in formula (4). The total replacement cost of all life parts in the whole life period is recorded as C LLP As shown in formula (5).
Wherein->Is the life part L in the kth maintenance j Maintenance policy to be taken,/->0 represents no replacement and 1 represents replacement.
In summary, the engine life-wide maintenance decision problem can be formally expressed as an optimization model shown in equation (6). Wherein m, T 1 、T 2 、…、T m 、For decision variables, the minimum total maintenance cost of the engine during the whole life is an objective function, as shown in a formula (6),
minC(s)＝m·C 0 +C LLP +C MOD
a solution vector formed by decision variables;
m: the number of maintenance times of the whole service life;
C 0 : the maintenance and fixation cost of the engine in factories;
T lim : the full life of the engine;
C LLP : total replacement cost over the life of all life pieces;
C MOD : total maintenance costs over the life of all the unit cells.
The problem of maintenance decision of the whole service life of the engine belongs to the problem of combination optimization, and a polynomial time algorithm for solving the optimal solution is difficult to find. From the definition field of each decision variable in the above equation (6), it is not difficult to find the solution space scale of the problem as shown in equation (7).
As can be seen from equation (7), the solution space of the problem is extremely large in scale. Factors affecting the scale of the solution space are the number of repairs m, the number of life parts n, the number of unit bodies p and the total life T of the engine lim . It is almost impossible to obtain an optimal solution by traversing the solution space in a fully traversal method.
The invention comprises the following steps:
aiming at the defects and shortcomings in the prior art, the invention provides an aeroengine life-span maintenance decision optimization algorithm based on problem decoupling, which can effectively reduce the maintenance cost of an engine.
The invention can be achieved by the following measures:
the aviation engine whole life maintenance decision optimization algorithm based on problem decoupling is characterized in that decision variables of an engine whole life maintenance decision problem are divided into three groups, wherein the first group is maintenance times in the whole life of the engine, the second group is maintenance time in the past, and the third group is whether each unit body is overhauled or not and whether each life part is replaced or not in the past maintenance; the first group of decision variables are subjected to traversal processing, the second group of decision variables are solved by adopting a particle swarm optimization algorithm, and the third group of decision variables are processed by the following steps:
step 1: estimating the minimum repair number m of the engine according to the repair interval of the engine min And the maximum repair number m max ；
Step 2: setting particle swarm size S and engine maintenance interval I min ,I max ]Maximum number of iterations N, learning factor c 1 And c 2 Weights w, r 1 、r 2 Number of engine repairs m=m min ；
Step 3: particle dimension q=m, randomly initializing S particles, including individual particlesPosition x i ＝{T i1 ,T i2 ,...,T im Sum of velocity v i ＝{v i1 ,v i2 ,...,v im -number of iterations n=0;
step 4: according to the position x of each particle i Solving methods of life part optimal replacement strategy and unit body optimal maintenance strategy, and calculating total replacement cost C in the whole life period of all life parts LLP,min (x i ) And total maintenance cost C during the whole life of all the unit bodies MOD,min (x i ) Thereby obtaining the total maintenance cost C (x i ) And the total maintenance cost C (x i ) As a negative number of the particle;
step 5: comparing the current adaptive value with the maximum adaptive value of the particles, and if the current adaptive value is larger, using the current adaptive value as the maximum adaptive value of the individual history of the particles, and updating the best position of the individual history by using the current position, otherwise, not updating;
step 6: for each particle, its historical maximum fitness value is compared to the global maximum fitness value within the population,
if the current adaptive value of the particle is larger, updating the current adaptive value of the particle to the global maximum adaptive value, updating the current position to the global historical best position, and otherwise, not updating;
step 7: updating the position and the speed of the particles, wherein the iteration times n=n+1;
step 8: judging whether the iteration times are more than N times, and if not, jumping to the step 4. If the number exceeds the number, recording the global maximum adaptation value and the global historical best position in N iterations, wherein the maintenance times are m=m+1;
step 9: judging whether the maintenance times m exceeds m max If not, turning to step 3, if yes, comparing the magnitudes of global maximum adaptation values in different maintenance times, and outputting the maximum adaptation value and maintenance times m corresponding to the maximum adaptation value * And maintenance times m * The global best position of the vehicle is corresponding to the lowest negative value of the total maintenance cost, and the global best position is corresponding to the optimal maintenance time;
step 10: and under the condition of determining the maintenance time, obtaining the optimal replacement strategy of the life part and the optimal maintenance strategy of the unit body according to the solving method of the optimal replacement strategy of the life part and the optimal maintenance strategy of the unit body.
In the invention, when the first group of decision variables are determined, the engine life-span maintenance decision optimization model established by taking the past maintenance time of the second group of decision variables as the decision variables is shown as a formula (8),
minC(T 1 ,T 2 ,…,T m )＝m·C 0 +C LLP,min (T 1 ,T 2 ,…,T m )
+C MOD,min (T 1 ,T 2 ,…,T m )
C LLP,min (T 1 ,T 2 ,...,T m ): when the decision variable is confirmed in the past maintenance time, the total replacement cost of all life parts in the whole life period is the minimum value;
C MOD,min (T 1 ,T 2 ,...,T m ): and when the decision variable is determined by the maintenance time, the total maintenance cost of all the unit bodies in the whole life cycle is the minimum.
The determination rule of the life part optimal replacement strategy in the present invention can be formally expressed as formula (9):
wherein:-representing a kth service life part L j Whether to replace or not, a value of 1 represents replacement, and a value of 0 represents no replacement.
According to the invention, the optimal maintenance strategy of the unit body is solved by sequentially solving the optimal replacement strategy of each unit body. For single unit, its optimal maintenance strategyThe value space size is 2 m Because the maintenance times of the engine in the whole life period are very small, the value space of the optimal maintenance strategy of the unit body is also very small, and therefore, the optimal maintenance strategy of the single unit body can be solved by adopting a traversing method.
Compared with the prior art, the method simplifies the algorithm complexity and effectively reduces the maintenance cost of the engine.
Description of the drawings:
FIG. 1 is a schematic diagram of a life part replacement rule in the present invention.
FIG. 2 is a graph of the number of life pieces versus algorithm solving time in example 1 of the present invention.
FIG. 3 is a graph showing the number of unit cells versus algorithm solving time in example 1 of the present invention.
FIG. 4 is a convergence diagram of the particle swarm in example 2 of the present invention.
FIG. 5 is a diagram showing the result of solving the lifetime component in example 2 of the present invention.
FIG. 6 is a diagram showing the result of solving the unit cell in example 2 of the present invention.
The specific embodiment is as follows:
the invention will be further described with reference to the drawings and examples.
Aiming at the problems in the background technology, the invention seeks a method for solving the problem with better solution. If intelligent optimization algorithms such as particle swarm optimization are directly adopted to directly solve the problem, due to the huge size of the solution space, a satisfactory solution is difficult to find even if a large amount of time is spent. Therefore, the invention firstly decouples the engine maintenance decision problem, reduces the solution space scale of the problem, and searches for the better solution of the problem on the basis.
There is some inherent link between decision variables of the engine life-time maintenance decision problem. If some decision variables are determined, other decision variables can find the optimal solution by adopting some quick methods, so that the size of the solution space to be traversed can be greatly reduced. Based on this, the invention divides the decision variables of the engine life-time maintenance decision problem into three groups. The first group is the maintenance times in the whole life of the engine, the second group is the maintenance time of the past, and the third group is whether each unit body is overhauled and each life part is replaced or not in the past maintenance. Of course, if a traversal method is adopted for each group of decision variables to find the optimal solution of the problem, the size of the solution space which needs to be traversed does not change from that of the non-grouping solution.
Analysis finds that the value range of the first group of decision variables is smaller for the maintenance decision problem of the whole service life of the engine, so that the group of decision variables can be traversed; the value range of the second group of decision variables is large, and a rapid method is difficult to find to solve the optimal solution, so that the group of decision variables can be solved by adopting a particle swarm optimization algorithm; the third set of decision variables has a large range of values, but it has been found through research that when the first set of decision variables and the second set of decision variables are determined, the set of decision variables can find the optimal solution by using some fast methods, and the specific method is referred to in the third section of the present invention.
In summary, when the first set of decision variables is determined, an engine life-time maintenance decision optimization model is established by taking the past maintenance time of the second set of decision variables as shown in formula (8).
minC(T 1 ,T 2 ,…,T m )＝m·C 0 +C LLP,min (T 1 ,T 2 ,…,T m )
+C MOD,min (T 1 ,T 2 ,…,T m )
C LLP,min (T 1 ,T 2 ,...,T m ): when the decision variable is confirmed in the past maintenance time, the total replacement cost of all life parts in the whole life period is the minimum value;
C MOD,min (T 1 ,T 2 ,...,T m ): and when the decision variable is determined by the maintenance time, the total maintenance cost of all the unit bodies in the whole life cycle is the minimum.
When the maintenance times and the past maintenance time in the whole life period of the aeroengine are determined, the determination rules of the optimal replacement strategy of the life piece are not difficult to find, and are as follows: if a life part is not replaced during a repair, which leads to the life part exceeding the life of the life part during the next repair, the life part must be replaced during the repair, otherwise, the replacement is not needed. As shown in fig. 1 below.
The determination rule of the lifetime piece optimal replacement policy can be formally expressed as formula (9).
Wherein:-representing a kth service life part L j Whether to replace or not, the value of 1 represents replacement,
a value of 0 indicates no replacement.
Notably, engines often have multiple life pieces. When the maintenance times and the past maintenance time in the whole life of the engine are determined, the optimal replacement strategies of all the life parts have independence, i.e. the optimal replacement strategies of all the life parts are not affected. Therefore, the optimal replacement strategy of each life part can be sequentially determined according to the determination rule of the optimal replacement strategy of the life part, and the solving efficiency can be improved.
When the maintenance times and the past maintenance opportunities in the whole life of the engine are determined, it is difficult to find a solution method for the optimal maintenance strategy of the unit body. Like the life piece, the engine generally has a plurality of unit bodies. When the maintenance times and the past maintenance time in the whole service life of the engine are determined, the optimal replacement strategy of each unit body has independence. Therefore, the optimal replacement strategy of each unit body can be solved in sequence. For a single unit body, the value space of the optimal maintenance strategy is 2 m . Because the maintenance times in the whole life of the engine are very small, the value space of the unit body optimal maintenance strategy is also very small. Thus, the traversing method can be adopted for the optimal maintenance of the single unit bodyAnd solving the strategy.
And according to the established engine full-life maintenance decision optimization model, the third partial-life part optimal replacement strategy and the solving method of the unit body optimal maintenance strategy. And solving the model by combining a particle swarm optimization algorithm. The algorithm performs the steps as follows.
Step 1: estimating the minimum repair number m of the engine according to the repair interval of the engine min And the maximum repair number m max 。
Step 2: setting particle swarm size S and engine maintenance interval I min ,I max ]Maximum number of iterations N, learning factor c 1 And c 2 Weights w, r 1 、r 2 Number of engine repairs m=m min 。
Step 3: particle dimension q=m, randomly initializing S particles, including the position x of each particle i ＝{T i1 ,T i2 ,...,T im Sum of velocity v i ＝{v i1 ,v i2 ,...,v im Number of iterations n=0.
Step 4: according to the position x of each particle i Solving methods of life part optimal replacement strategy and unit body optimal maintenance strategy, and calculating total replacement cost C in the whole life period of all life parts LLP,min (x i ) And total maintenance cost C during the whole life of all the unit bodies MOD,min (x i ). Thereby obtaining the total maintenance cost C (x) i ) And the total maintenance cost C (x i ) As the adaptation value of the particle.
Step 5: for each particle. The current fitness value is compared with the maximum fitness value experienced and if it is larger, it is used as the individual historical maximum fitness value for the particle, and the individual historical best location is updated with the current location. Otherwise, not updating.
Step 6: and comparing the historical maximum adaptation value of each particle with the global maximum adaptation value in the group, and if the historical maximum adaptation value is larger, updating the current adaptation value of the particle to the global maximum adaptation value, and updating the current position to the global historical best position. Otherwise, not updating.
Step 7: the position and velocity of the particles are updated, the number of iterations n=n+1.
Step 8: judging whether the iteration times are more than N times, and if not, jumping to the step 4. And if the total maximum adaptation value exceeds the total maximum adaptation value, recording the global historical best position in N iterations. Maintenance times m=m+1.
Step 9: judging whether the maintenance times m exceeds m max If not, go to step 3. And if the number of the maintenance times exceeds the number of the maintenance times, comparing the magnitude of the global maximum adaptation value. Outputting the maintenance times m corresponding to the maximum adaptation value * And maintenance times m * Is the global best position in the past. The largest adaptation value corresponds to the negative value with the smallest total maintenance cost, and the global best position corresponds to the optimal maintenance opportunity.
Step 10: and under the condition of determining the maintenance time, obtaining the optimal replacement strategy of the life part and the optimal maintenance strategy of the unit body according to the solving method of the optimal replacement strategy of the life part and the optimal maintenance strategy of the unit body.
Example 1:
the algorithm provided by the invention is verified and evaluated by adopting the initial states of the randomly generated multiple groups of engines. To solve the particle group into knots
The result is compared to the traversal optimal solution. Let the number of cell bodies p e {1,2,3,4,5,10,15}, the number of lifetime pieces n e {2,3,4,5,10,15,20}. Full life T of engine lim The maintenance interval is within interval [80,160 =600 flight cycles]Rounding off on the flight cycle. Life of life pieceObeying an even distribution over the flight cycle of the interval (200, 300), the life part replacement cost +.>Obeying an even distribution over the interval $ (7000,15000), maximum repair interval +.>Compliance zone (150, 350) of the flight cycleEven distribution, optimal overhaul interval->Compliance zone->Evenly distributed on the flight cycle, overhaul cost->Uniform distribution over dollars in compliance area (7000,15000), cost of minor repair +.>Obeying section->Uniform distribution over dollars. When randomly initializing the engine state, the initial service time T of the engine 0 Obeying an even distribution over the flight cycle of the interval (1, 200), the initial service time of the life part +.>Obeying section->Even distribution over the flight cycle, initial service time of the unit cell +.>Obeying section->Even distribution over the flight cycle. Taking p for each (p, n)<n, 20 questions were randomly generated for the experiment. The particle swarm algorithm provided by the invention is adopted on a common computer to solve each group of problems. And solving the optimal solution of each group of problems by adopting a traversing method, and recording the average relative deviation R of each group of solving results of the particle swarm algorithm and the optimal solution and the maximum relative deviation R of each group of solving results and the optimal solution. Table 1 shows the results of the experiment. Watch with a watch1, the average relative deviation R between the solving result obtained by the method and the optimal solution is kept within 2%, and the maximum relative deviation R is kept within 9%, which indicates that the better maintenance strategy of the aero-engine can be obtained by solving the whole life maintenance decision optimization model of the aero-engine by using the particle swarm algorithm.
Table 1 experimental results
Table 1 Experimental result
In order to explore the relation between the number of life parts n, the number of unit bodies p and the algorithm solving time t, two groups of experiments are carried out. The first set lets the number of engine units p=3, the number of life pieces n= {3,4,5,10,15,20}. The second set makes the number of engine life parts n=20, the number of unit bodies p= {3,4,5,10,15}, and 100 problems generated randomly for each set (p, n) are solved, and the solving time t of the algorithm is recorded respectively, as shown in fig. 2 and 3.
As can be seen from fig. 2 and 3, as the number of life parts n or the number of unit bodies p increases, the algorithm can still be solved, which illustrates that the method provided by the invention is also applicable to the problem of full life maintenance decision of the engine with more life parts and unit bodies. However, as the number of unit cells or life pieces increases, the algorithm solution time increases.
Example 2:
taking an engine of an airline company as an example, the particle swarm optimization algorithm provided by the invention is adopted to solve the whole life maintenance decision optimization model of the engine.
The engine of this model comprises 17 units and 20 life pieces, table 2 is a list of units and the current use time of each unit, and table 3 is a list of life pieces and the current use time of each life piece. Fixing of engine for each factory repairThe maintenance cost is set to be 12 ten thousand dollars, and the whole service life T of the engine lim =60000 flight cycles, the current total flight cycle T of the engine 0 =10907 flight cycle. The particle swarm optimization method provided by the invention is adopted for solving, the particle swarm size is set to 3000, and the maximum iteration frequency is set to 200.
Table 2 list of unit cell
Table 2 Modules list
TABLE 3 life part list
The algorithm totals 1393.7s, with a total maintenance cost of 1366.933 ten thousand dollars. It can be seen from fig. 4 that the algorithm begins to converge when iterating around 75 times and a better solution is obtained when iterating around 125 times. The invention shows that the optimization model which is established by the invention and takes the maintenance time as a decision variable and the lowest total maintenance cost as an objective function can play the roles of reducing the complexity, the operation time and the maintenance cost. L (L) 1
According to the result obtained by the algorithm, the maintenance times in the whole life period is 4, and the maintenance time and the maintenance working range are shown in fig. 5 and 6.
In fig. 4 and 5, 72,8072,20072,36072 is the maintenance timing, L 1 ,L 2 ...,L 20 Numbering, M, of life parts 1 ,M 2 ,...,M 20 Numbering the unit bodies. The box corresponding to the life piece is filled with oblique lines to represent that the life piece is replaced, and the unfilled life piece is not replaced. And filling the corresponding boxes of the unit bodies with the grids to carry out major repair on the unit bodies, and carrying out minor repair on the unit bodies which are not filled with the grids. Although in the third sumThe life part is greatly replaced during the fourth maintenance, but the life part replacement can be calculated to meet the replacement rule of the life part. The result shows that the better maintenance strategy of the aeroengine can be obtained by solving the maintenance decision optimization model of the whole life of the aeroengine by using the particle swarm algorithm.
In order to determine the maintenance time and the maintenance working range of the aeroengine, a maintenance decision model taking the maintenance time as a decision variable is established, and a solving strategy of the model is provided. The model was evaluated using a numerical experiment. Experimental results show that a better solving result can be obtained, and the method is also suitable for occasions with more unit bodies and service life numbers. And finally, verifying the model by adopting actual data of a certain engine of the airline company. The verification result shows that the model can provide decision support for the maintenance time and the maintenance working range of the engine.
Claims (3)
1. The aviation engine whole life maintenance decision optimization algorithm based on problem decoupling is characterized in that decision variables of an engine whole life maintenance decision problem are divided into three groups, wherein the first group is maintenance times in the whole life of the engine, the second group is maintenance time in the past, and the third group is whether each unit body is overhauled or not and whether each life part is replaced or not in the past maintenance; the first group of decision variables are subjected to traversal processing, the second group of decision variables are solved by adopting a particle swarm optimization algorithm, and the third group of decision variables are processed by the following steps:
step 1: estimating the minimum repair number m of the engine according to the repair interval of the engine min And the maximum repair number m max ；
Step 2: setting particle swarm size S and engine maintenance interval I min ,I max ]Maximum number of iterations N, learning factor c 1 And c 2 Weights w, r 1 、r 2 Number of engine repairs m=m min ；
Step 3: particle dimension q=m, randomly initializing S particles, including the position x of each particle i ＝{T i1 ,T i2 ,...,T im Sum ofVelocity v i ＝{v i1 ,v i2 ,...,v im -number of iterations n=0;
step 4: according to the position x of each particle i Solving methods of life part optimal replacement strategy and unit body optimal maintenance strategy, and calculating total replacement cost C in the whole life period of all life parts LLP,min (x i ) And total maintenance cost C during the whole life of all the unit bodies MOD,min (x i ) Thereby obtaining the total maintenance cost C (x i ) And the total maintenance cost C (x i ) As a negative number of the particle;
step 5: comparing the current adaptive value with the maximum adaptive value of the particles, and if the current adaptive value is larger, using the current adaptive value as the maximum adaptive value of the individual history of the particles, and updating the best position of the individual history by using the current position, otherwise, not updating;
step 6: comparing the historical maximum adaptation value of each particle with the global maximum adaptation value in the group, if the historical maximum adaptation value is larger, updating the current adaptation value of the particle to the global maximum adaptation value, and updating the current position to the global historical best position, otherwise, not updating;
step 7: updating the position and the speed of the particles, wherein the iteration times n=n+1;
step 8: judging whether the iteration times exceeds N times, if not, jumping to the step 4, and if so, recording the global maximum adaptation value and the global historical best position in the N iterations, wherein the maintenance times are m=m+1;
step 9: judging whether the maintenance times m exceeds m max If not, turning to step 3, if yes, comparing the magnitudes of global maximum adaptation values in different maintenance times, and outputting the maximum adaptation value and maintenance times m corresponding to the maximum adaptation value * And maintenance times m * The global best position of the vehicle is corresponding to the lowest negative value of the total maintenance cost, and the global best position is corresponding to the optimal maintenance time;
step 10: under the condition of determining maintenance time, obtaining the optimal replacement strategy of the life part and the optimal maintenance strategy of the unit body according to the solving method of the optimal replacement strategy of the life part and the optimal maintenance strategy of the unit body;
when the first group of decision variables are determined, an engine life-span maintenance decision optimization model established by taking the past maintenance time of the second group of decision variables as the decision variables is shown as (8),
minC(T 1 ,T 2 ,…,T m )＝m·C 0 +C LLP,min (T 1 ,T 2 ,…,T m )+C MOD,min (T 1 ,T 2 ,…,T m )
C LLP,min (T 1 ,T 2 ,...,T m ): when the decision variable is confirmed in the past maintenance time, the total replacement cost of all life parts in the whole life period is the minimum value;
C MOD,min (T 1 ,T 2 ,...,T m ): and when the decision variable is determined by the maintenance time, the total maintenance cost of all the unit bodies in the whole life cycle is the minimum.
2. The problem decoupling-based aircraft engine life-wide maintenance decision optimization algorithm of claim 1, wherein the determination rule of the life-wide replacement strategy is formally expressed as formula (9):
wherein:-representing a kth service life part L j Whether to replace or not, a value of 1 represents replacement, and a value of 0 represents no replacement.
3. An aeroengine life cycle based on problem decoupling as claimed in claim 1The maintenance decision optimization algorithm is characterized in that the optimal maintenance strategy of the unit body is obtained by solving the optimal replacement strategy of each unit body in sequence, and the value space of the optimal maintenance strategy of the single unit body is 2 m Because the maintenance times of the engine in the whole life period are very small, the value space of the optimal maintenance strategy of the unit body is also very small, and therefore, the optimal maintenance strategy of the single unit body can be solved by adopting a traversing method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010625099.XA CN113887770B (en) | 2020-07-01 | 2020-07-01 | Aero-engine life-span maintenance decision optimization algorithm based on problem decoupling |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010625099.XA CN113887770B (en) | 2020-07-01 | 2020-07-01 | Aero-engine life-span maintenance decision optimization algorithm based on problem decoupling |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113887770A CN113887770A (en) | 2022-01-04 |
| CN113887770B true CN113887770B (en) | 2024-04-16 |
Family
ID=79012169
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010625099.XA Active CN113887770B (en) | 2020-07-01 | 2020-07-01 | Aero-engine life-span maintenance decision optimization algorithm based on problem decoupling |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113887770B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104568344A (en) * | 2014-10-09 | 2015-04-29 | 芜湖扬宇机电技术开发有限公司 | EGR (Exhaust Gas Recirculation) pipeline fault detection diagnosis method |
| CN106600072A (en) * | 2016-12-21 | 2017-04-26 | 中国航空工业集团公司雷华电子技术研究所 | Maintenance decision optimization algorithm based on failure prediction |
| CN109685224A (en) * | 2018-12-20 | 2019-04-26 | 威海众成信息科技股份有限公司 | A kind of management method for formulating engine maintenance decision according to lifetime piece |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080162235A1 (en) * | 2006-12-30 | 2008-07-03 | United Technologies Corporation | Method and Means for Optimizing Maintenance Work Schedules |
-
2020
- 2020-07-01 CN CN202010625099.XA patent/CN113887770B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104568344A (en) * | 2014-10-09 | 2015-04-29 | 芜湖扬宇机电技术开发有限公司 | EGR (Exhaust Gas Recirculation) pipeline fault detection diagnosis method |
| CN106600072A (en) * | 2016-12-21 | 2017-04-26 | 中国航空工业集团公司雷华电子技术研究所 | Maintenance decision optimization algorithm based on failure prediction |
| CN109685224A (en) * | 2018-12-20 | 2019-04-26 | 威海众成信息科技股份有限公司 | A kind of management method for formulating engine maintenance decision according to lifetime piece |
Non-Patent Citations (1)
| Title |
|---|
| 基于蛙跳退火粒子群算法的民航发动机单元体修理级别决策及成本优化;张青;郑岩;计算机应用;20201231(012);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113887770A (en) | 2022-01-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107730014B (en) | CBM-based fleet maintenance decision method | |
| EP3751147B1 (en) | Computer-implemented method for determining compressor operability | |
| CN110245842B (en) | Production line risk scheduling method for sudden large faults of equipment | |
| Lu et al. | Gas path on-line fault diagnostics using a nonlinear integrated model for gas turbine engines | |
| Wang et al. | A DRL-aided multi-layer stability model calibration platform considering multiple events | |
| CN106355253B (en) | Method for determining optimal operation performance interval of aircraft engine | |
| CN113887770B (en) | Aero-engine life-span maintenance decision optimization algorithm based on problem decoupling | |
| WO2018169722A1 (en) | Fatigue crack growth prediction | |
| Almgren et al. | Optimization models for improving periodic maintenance schedules by utilizing opportunities | |
| EP3751148A1 (en) | Computer-implemented method for controlling a gas turbine engine | |
| Keskin | Process integration and automated multi-objective optimization supporting aerodynamic compressor design | |
| Meyer et al. | Prediction of turbine rotor blade forcing due to in-service stator vane trailing edge damage | |
| CN109978187B (en) | Maintenance decision method for bleed air pressure regulating valve of airplane | |
| Jones et al. | A preliminary engine design process for an affordable capability | |
| CN113962447B (en) | Complex equipment batch long-term maintenance plan optimization method based on improved particle swarm algorithm | |
| CN113010981B (en) | Maintenance decision method for low-pressure air entraining valve of aircraft engine | |
| EP3751146A1 (en) | Computer-implemented method for training a machine learning algorithm | |
| Estébanez et al. | An integrated aerospace requirement setting and risk analysis tool for life cycle cost reduction and system design improvement | |
| Debbah et al. | GA/PSO robust sliding mode control of aerodynamics in gas turbine | |
| Xu et al. | Aircraft engine maintenance based on reinforcement learning | |
| Chen et al. | A whole life cycle maintenance strategy of civil aircraft precooler based on case-based reasoning and probability model | |
| Chen et al. | Study on boiler combustion optimization based on sparse least squares support vector machine | |
| Liu et al. | Research on Group Preventive Maintenance Method Based on Components Dependence | |
| CN117540840A (en) | Equipment life-span maintenance decision optimization method under exit condition | |
| Deng et al. | Research on Optimal Scheduling of Battlefield Maintenance Tasks Considering Residual Life |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |