Background
The unmanned aerial vehicle system consists of an aircraft and a ground station. The ground station of the unmanned aerial vehicle is mainly used for monitoring the flight task of the aircraft, and is a ground command control center of an unmanned aerial vehicle system. The unmanned aerial vehicle ground station system consists of two parts, namely hardware and software, wherein the software has the functions of flight monitoring, task planning, map navigation, flight data query and processing and the like. At present, the software scale of the unmanned aerial vehicle ground station system has the characteristics of enlargement, complexity, diversification and the like, so that the possibility of fault factors existing in the software research and development process is higher and higher. Software failure can not only cause system failure, equipment damage and the like, cause serious economic loss, but also can endanger personal safety. Therefore, the reliability of the ground station system of the unmanned aerial vehicle needs to be studied.
The reliability of the system is limited and affected by the reliability of all modules. From the software development perspective, in order to ensure the overall reliability of the system, the reliability index of the system must be correctly, scientifically and economically distributed to each component and each module, so that each module can reach the required reliability index. From the perspective of software testing and evaluation, not only the overall reliability of the system needs to be tested, but also the reliability of each module needs to be tested, and particularly, the critical and important modules must meet the specified reliability standards. Therefore, when the reliability of the system is studied, the reliability index of each module must be calculated on the premise of the overall reliability, and the reliability distribution of each module is completed.
The reliability of the unmanned aerial vehicle ground station system software is distributed, and various reliability distribution methods can be adopted to improve the economy and safety of the unmanned aerial vehicle ground station system software in the research and development process, so that the distribution optimization of the reliability of the unmanned aerial vehicle ground station system software is realized. The traditional software reliability allocation adopts an equal allocation method, a proportion allocation method, a grading allocation method and the like. The equal distribution method assumes that the system is formed by connecting a plurality of modules in series, and then distributes the same software reliability index to each module. However, the equal distribution method is only suitable for the initial stage of system development and design, and the definition and the function module of the system are not clear at this time, and the distribution work can be carried out only according to engineering experience of workers and the reliability distribution data of the existing similar system software. Due to the subjective will of workers or the inaccurate reliability allocation of the reference system, the software reliability allocation result is easy to be biased. The proportional allocation method is simple and intuitive, and is suitable for the situation that the new system and the original system have basically the same structure. However, when the original system itself has a situation that the software reliability allocation is not reasonable, the software reliability allocation of the new system may be also not reasonable. The scoring distribution method is to score and comprehensively analyze the influence factors of the system reliability by experienced experts to obtain the relative ratio of the reliability influence factors of the system, and finally complete the distribution of the reliability value of the system. The scoring distribution method is likely to cause the scoring deviation due to different expert scoring standards, and influences the final software reliability distribution of the system.
The unmanned aerial vehicle ground station system U can specifically comprise a debugging module U in function1And a route planning management module U2Flight monitoring control module U3Data transmission module U4And a data management module U5And the like. Wherein, a debugging module U is arranged1The functions of accelerometer calibration, compass calibration, remote controller calibration, mode setting and the like are mainly realized; air route planning management module U2The functions of waypoint editing, map display, route planning, waypoint reading and the like of the aircraft are mainly realized; flight monitoring control module U3Mainly realizes the functions of flight control, flight data display, parameter adjustment, data storage, data analysis, data playback and the like; data transmission module U4The functions of communication between the aircraft and the ground station and the like are realized; data management module U5The functions of storing, replaying, analyzing and the like of the aircraft data in the ground station are mainly realized.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides an unmanned aerial vehicle ground station system software reliability intelligent distribution method based on a particle swarm algorithm and an analytic hierarchy process. According to the method, the influence factors of the reliability of each module in the unmanned aerial vehicle ground station system are analyzed by using an analytic hierarchy process, and finally, a reliability distribution model of the unmanned aerial vehicle ground station system based on the minimization of reliability constraint cost is constructed. The method adopts the particle swarm algorithm to seek the optimal solution value of the reliability distribution model of the unmanned aerial vehicle ground station system under the multi-constraint condition, is easy to realize when solving the reliability distribution problem of the unmanned aerial vehicle ground station system software, and embodies better reliability distribution optimization performance. The particle swarm algorithm simulates the predation behavior of a bird swarm, and the movement of the whole swarm generates an evolution process from disorder to order in a problem solving space by utilizing the sharing of the individual pair information in the swarm, so that the optimal solution is obtained.
The method specifically comprises the following steps:
(1) and initializing parameters, including the expected reliability value of the ground system of the unmanned aerial vehicle, the value range of the reliability value of each module in the ground station system of the unmanned aerial vehicle, the particle size in the particle swarm algorithm, the search space dimension, the iteration times and the like.
(2) And solving the weight W of reliability distribution of each module of the unmanned aerial vehicle ground station system under the influence factors of the severity K, the complexity H, the importance F, the mean fault interval time MTBF, the running time T and the like according to an analytic hierarchy process.
(3) Obtaining each module U of unmanned aerial vehicle ground station systemjCost function C of reliability assignment values ofj。
(4) And constructing a reliability distribution model of the unmanned aerial vehicle ground station system based on reliability constraint cost minimization.
(5) According to a reliability distribution model of an unmanned aerial vehicle ground station system, constructing a fitness value function fit (X) of any particle i in a particle swarmi)。
(6) Obtaining the position of any particle i in the particle swarm according to the t iteration
Speed of rotation
Obtaining the position of any particle i in the particle swarm obtained by the t +1 th iteration
Speed of rotation
(7) And (6) executing the loop until the maximum iteration number T is reached, and ending the loop.
(8) After the particle swarm optimization is finished, PgTThe optimal positions that all particles in the particle swarm have undergone after T iterations. According to PgT=(R1,R2,R3,R4,R5) Then is waited forUnmanned aerial vehicle ground station system in set up debugging module U1And a route planning management module U2Flight monitoring control module U3Data transmission module U4Data management module U5The reliability of (2) is assigned an optimal value.
The method adopts a particle swarm algorithm and a reliability distribution model based on an analytic hierarchy process to initialize a set debugging module U of an unmanned aerial vehicle ground station system1And a route planning management module U2Flight monitoring control module U3Data transmission module U4Data management module U5And (3) constructing a reliability distribution model based on reliability constraint cost minimization according to the cost function of the unmanned aerial vehicle ground station system, constructing a fitness function of any particle in the particle swarm according to the composition relation between any particle in the particle swarm and the reliability distribution value of each module in the unmanned aerial vehicle ground station system, and finally distributing a reasonable reliability value for each module in the unmanned aerial vehicle ground station system through a multi-time particle swarm iterative optimization algorithm.
The particle swarm algorithm starts from a random solution, evaluates the quality of the current solution through the fitness value calculated by the fitness function of any particle in the algorithm, iteratively searches for particles meeting various constraint conditions, and finally searches for the globally optimal solution. The particle swarm algorithm simulates the predation behavior of a bird swarm to a certain extent, and the movement of the whole swarm is subjected to an evolution process from disorder to order in a problem solving space by utilizing the sharing of individuals in the swarm to information, so that an optimal solution is obtained. The analytic hierarchy process is a software reliability dividing method for dividing an unmanned aerial vehicle ground station system into functional modules according to functions and then distributing reliability indexes to the modules. The reliability distribution model based on the analytic hierarchy process is simple and practical, decomposes a complex system, enables the thinking process of people to be mathematic and systematized, is convenient for people to accept, and can solve the problem that multiple targets are difficult to be completely quantized into a multi-level single-target problem. The reliability intelligent distribution algorithm based on the particle swarm algorithm and the analytic hierarchy process has the advantages of good stability, high convergence speed and high precision, and the total development cost is gradually reduced along with the increase of evolution algebra. Compared with the traditional reliability distribution methods such as an equal distribution method, a proportional distribution method, a scoring distribution method and the like, the method has the advantages of simple algorithm, high solving speed and wide adaptability when the problem of complex system reliability distribution is faced, can complete intelligent optimal search and solution of reliability distribution under various constraint conditions, and has better optimization performance.
Detailed Description
As shown in fig. 2, the method adopts a particle swarm algorithm and a reliability distribution model based on an analytic hierarchy process to intelligently optimize and solve the reliability distribution values of a plurality of modules of the unmanned aerial vehicle ground station system software, and finds the intelligent reliability distribution of the unmanned aerial vehicle ground station system meeting various reliability distribution constraint conditions. The specific implementation method comprises the following steps:
(1) and initializing parameters, including the expected reliability value of the ground system of the unmanned aerial vehicle, the value range of the reliability value of each module in the ground station system of the unmanned aerial vehicle, the particle size in the particle swarm algorithm, the search space dimension, the iteration times and the like.
The reliability value of the unmanned aerial vehicle ground station system U is R, and the expected reliability value is Rs. Setting a debug module U1Has a reliability assigned value of R1And a route planning management module U2Has a reliability assigned value of R2Flight monitoring control module U3Has a reliability assigned value of R3Data transmission module U4Has a reliability assigned value of R4Data management module U5Has a reliability assigned value of R5。R1、R2、R3R4、R5Respectively is a preset [ R ]1min,R1max]、[R2min,R2max]、[R3min,R3max]、[R4min,R4max]、[R5min,R5max]。
In the particle swarm optimization, each particle is equivalent to a solution of a solving system, and an optimal solution is found through calculation and multiple iterations. The particle group consists of M particles, each particle has a search space of N dimensions, and the position X of any particle i in the particle groupi=(Xi1,Xi2,…XiN) Velocity Vi=(Vi1,Vi2,…,ViN). The fitness value of particle i is defined by the function fit (X)i) Solving, the step length adjustment factor of the particle i to the best position direction is constant c1The step length adjustment of the particle i to the global best position direction of the particle swarm is a constant c2The iteration number of the particle swarm is T, and the maximum iteration number is T.
Position X of arbitrary particle i in the particle populationiComposed of the reliability distribution values of 5 modules of the unmanned aerial vehicle ground station system, namely Xi=(R1,R2,R3,R4,R5) And N is 5. Position X of arbitrary particle iiA possible value of the reliability distribution value of 5 modules of the unmanned aerial vehicle ground station system, namely the fitness function fit (X) of any particle ii) Assignment of an objective function, fitR (R), by the reliability of 5 modules of an unmanned aerial vehicle ground station system1,R2,R3,R4,R5) Given, i.e. fit (X)i)=fitR(R1,R2,R3,R4,R5). After T iterations through the particle swarm, the maximum fitness value fit (X) in the particle swarm is obtainedi) Position X of particle i of (2)iThe optimal value of reliability distribution of 5 modules of the unmanned aerial vehicle ground station system is required.
(2) And solving the weight W of reliability distribution of each module of the unmanned aerial vehicle ground station system under the influence factors of the severity K, the complexity H, the importance F, the mean fault interval time MTBF and the running time T according to an analytic hierarchy process.
The analytic hierarchy process is a software reliability dividing method for dividing an unmanned aerial vehicle ground station system into functional modules according to functions and then distributing reliability indexes to the modules. In a software reliability allocation task of an unmanned aerial vehicle ground station system, the reliability of each functional module of the unmanned aerial vehicle ground station system is affected by multiple factors such as the severity K, the complexity H, the importance F, the mean time between failure MTBF, the operation time T and the like. Meanwhile, the measurement units of the influencing factors are obviously inconsistent, and the measurement values need to be normalized, so that the influencing factors are comparable.
The severity K represents the severity of a failure mode existing in each module in the unmanned aerial vehicle ground station system; the complexity H represents the complexity of each module in the unmanned aerial vehicle ground station system, and the current general information entropy is used for quantitative description. The information entropy reflects the utilization rate of the module quantitatively, the larger the utilization rate is, the smaller the uncertainty degree is, the smaller the information quantity contained in the module is, and the lower the complexity is; the importance degree F represents the influence degree of each module in the unmanned aerial vehicle ground station system on the functions of the unmanned aerial vehicle ground station system after the modules fail, and the influence degree is larger when the importance degree is larger; mean Time Between Failures (MTBF) represents the failure rate of each module in the ground station system of the unmanned aerial vehicle, and the reciprocal of the module failure rate is the mean time between failures of the modules; and the operation time T represents the proportion of the operation time of each module in the unmanned aerial vehicle ground station system to the total operation time of the unmanned aerial vehicle ground station system.
Firstly, a judgment weight vector D of the reliability distribution of the unmanned aerial vehicle ground station system by the influence factors such as the severity K, the complexity H, the importance F, the mean time between failure MTBF, the running time T and the like is obtained.
According to the influence degrees of the influence factors such as the severity K, the complexity H, the importance F, the mean fault interval time MTBF, the running time T and the like on the reliability distribution of each module of the unmanned aerial vehicle ground station system, the influence magnitudes pk, ph, pf, pmtbf and pt of the influence factors such as the severity K, the complexity H, the importance F, the mean fault interval time MTBF, the running time T and the like on the reliability distribution of the unmanned aerial vehicle ground station system can be determined, and a judgment matrix A of the reliability distribution of the unmanned aerial vehicle ground station system by the influence factors is constructed according to the formulas (1), (2) and (3); judging a feature vector corresponding to the maximum feature value of the matrix A, wherein the feature vector is a judgment weight vector D of the influence factors of the severity K, the complexity H, the importance F, the mean time between failure MTBF and the running time T on the reliability distribution of the ground station system of the unmanned aerial vehicle; in the judgment weight vector D, the dK represents the influence weight of the severity K on the reliability distribution of the unmanned aerial vehicle ground station system, the dH represents the influence weight of the complexity H on the reliability distribution of the unmanned aerial vehicle ground station system, the dF represents the influence weight of the importance F on the reliability distribution of the unmanned aerial vehicle ground station system, the dMTBF represents the influence weight of the mean fault interval time MTBF on the reliability distribution of the unmanned aerial vehicle ground station system, and the dT represents the influence weight of the running time T on the reliability distribution of the unmanned aerial vehicle ground station system.
p=[pk ph pf pmtbf pt] (1)
aστ=pσ/pτ (2)
D=[dK dH dF dMTBF dT] (4)
Secondly, influence unit vectors rk, rh, rf, rmtbf and rt of influence factors such as severity K, complexity H, importance F, mean time between failure MTBF, running time T and the like on reliability distribution of each module of the ground station system of the unmanned aerial vehicle are obtained.
Obtaining module U in unmanned aerial vehicle ground station systemjSeverity of (k)jBy normalization to obtain rkjAnd therefore, a unit vector rk of the severity influence of reliability distribution of each module of the unmanned aerial vehicle ground station system is obtained. Wherein, betajIs a module UjThe loss probability is the probability that the unmanned aerial vehicle ground station system fails when the module fails, and the value of the loss probability is [0,1 ]]The larger the value is, the larger the probability of failure of the unmanned aerial vehicle ground station system is; alpha is alphajIs a module UjThe sum of the probability of various faults relative to all faults is determined according to the module UjObtaining the operation condition of the functional components;
kj=βj*αj*(1-Rjmin)*106,j∈[1,N] (5)
rk=[rk1 … rkj … rkN],j∈[1,N] (7)
obtaining module U in unmanned aerial vehicle ground station systemjComplexity h ofjObtaining rh by normalizationjAnd therefore, a unit vector rh of the complexity influence of reliability distribution of each module of the unmanned aerial vehicle ground station system is obtained. Wherein o isjIs a module UjThe usage rate of (2);
hj=-log2(oj)*oj,j∈[1,N] (8)
rh=·rh1 … rhj… rhN],j∈[1,N] (10)
obtaining module U in unmanned aerial vehicle ground station systemjDegree of importance fjObtaining rf by normalizationjTherefore, an importance influence unit vector rf of reliability distribution of each module of the unmanned aerial vehicle ground station system is obtained. Wherein, FsFailure model for unmanned aerial vehicle ground station system, fiIs a module UiThe failure model of (2); determining 3 minimal cut sets { U } by combining an unmanned aerial vehicle ground station system according to the existing universal fault tree analysis method1U2U4U5},{U1U4U5},{U1U3U4U5}. The minimal cut set is defined as a module set which can cause the unmanned aerial vehicle ground station system to fail, so as to obtain a failure model Fs;
Fs=f1f2f4f5+f1f4f5+f1f3f4f5 (12)
rf=[rf1 … rfj … rfN],j∈[1,N] (14)
Obtaining module U in unmanned aerial vehicle ground station systemjMean time between failures (mtbf)jNormalized to get rmtbfjTherefore, an average fault interval time influence unit vector rmtbf of reliability distribution of each module of the unmanned aerial vehicle ground station system is obtained.
mtbfj=1/(1-Rjmin),j∈[1,N] (15)
rmtbf=[rmtbf1 … rmtbfj … rmtbfN],j∈[1,N] (17)
Obtaining module U in unmanned aerial vehicle ground station systemjRun time scaling factor tjThe rt is obtained by normalizationjAnd therefore, a unit vector rt of the influence of the running time of reliability distribution of each module of the unmanned aerial vehicle ground station system is obtained. Wherein G isjFor module U in unmanned aerial vehicle ground station systemjG is the total operating time of the ground station system of the unmanned aerial vehicle;
tj=Gj/G,j∈[1,N] (18)
rt=[rt1 rt2 rt3 rt4 rt5],j∈[1,N] (20)
finally, according to the judgment weight D, the unit vector rk of the severity influence, the unit vector rh of the complexity influence and the influence of the importanceAnd solving the reliability distribution weight W of each module of the ground station system of the unmanned aerial vehicle by using the force unit vector rf, the mean fault interval time influence unit vector rmtbf, the running time influence unit vector rt and the like. Wherein, wjFor module U in unmanned aerial vehicle ground station systemjAnd the influence weight of reliability distribution under the influence factors of the severity K, the complexity H, the importance F, the mean time between failure MTBF, the running time T and the like.
(3) Obtaining each module U of unmanned aerial vehicle ground station systemjCost function C of reliability assignment values ofj。
According to wjObtaining module U in unmanned aerial vehicle ground station systemjCan reach the highest reliability value RjmaxDegree of feasibility yjBuilding each module U in the unmanned aerial vehicle ground station systemjCost function C ofj. Degree of feasibility yjSmaller the representation of the lifting module UjReliability R ofjThe more difficult, the required cost CjThe higher the feasibility yjThe larger the lifting module UjReliability R ofjThe easier, the required cost CjThe lower.
yj=1-wj,j∈[1,N] (22)
(4) And constructing a reliability distribution model of the unmanned aerial vehicle ground station system based on reliability constraint cost minimization.
The reliability distribution model of the unmanned aerial vehicle ground station system comprises 1 minimum cost function and a plurality of reliability constraint conditions. Cost function Cost of unmanned aerial vehicle ground station system is calculated by each module U in unmanned aerial vehicle ground station systemjIs assigned a value RjCost function C ofjThe method comprises the following steps:
each module U in unmanned aerial vehicle ground station systemjIs assigned a value RjThere is a minimum value RjminAnd maximum value RjmaxThe constraint of (2). Equipment debugging module U in known unmanned aerial vehicle ground station system1Is assigned a value R1And a route planning management module U2Is assigned a value R2Flight monitoring control module U3Is assigned a value R3Data transmission module U4Is assigned a value R4Data management module U5Is assigned a value R5Under the condition, according to the structural diagram 1 of the unmanned aerial vehicle ground station system, the reliability value R of the current unmanned aerial vehicle ground station system is obtained by combining the existing general reliability value calculation method of the series-parallel connection system, and the reliability value R must be not less than the reliability value R expected to be reached by the unmanned aerial vehicle ground station systems。
R=[1-(1-R1R2)*(1-R1)*(1-R1R3)]*R4*R5 (25)
Therefore, a reliability distribution model of the unmanned aerial vehicle ground station system based on reliability constraint cost minimization can be obtained:
minimizing the cost function:
reliability constraint conditions:
(5) according to a reliability distribution model of an unmanned aerial vehicle ground station system, constructing a fitness value function fit (X) of any particle i in a particle swarmi)。
The reliability distribution model of the unmanned aerial vehicle ground station system based on the reliability constraint cost minimization comprises a minimized cost function and a plurality of constraint conditions of reliability values. And converting the software reliability distribution problem with constraint conditions into an unconstrained software reliability distribution problem by adopting the conventional universal penalty function method, and solving the unconstrained software reliability distribution problem.
Constructed reliability assignment objective function, fitR (R)1,R2,R3,R4,R5) Comprises the following steps:
wherein Q isiFor the penalty factor, i is 1 … 3. Fitness function fit (X) of any particle i in the populationi) The following equation (28) can be used to obtain:
fit(Xi)=fitR(R1,R2,R3,R4,R5) (29)
(6) according to the position of any particle i in the j dimension in the particle swarm obtained by the t iteration
Speed of rotation
The position of any particle i in the j dimension in the particle swarm obtained by the (t + 1) th iteration is obtained
Speed of rotation
Wherein r is
1、r
2Is [0,1 ]]A random number of intervals;
the optimal position of the particle i in the particle swarm after t iterations;
the optimal positions that all particles in the particle swarm have undergone after t iterations.
(7) And (6) executing the loop until the maximum iteration number T is reached, and ending the loop.
(8) After the particle swarm optimization is finished, PgTThe optimal positions that all particles in the particle swarm have undergone after T iterations. According to PgT=(R1,R2,R3,R4,R5) Then the setting debugging module U in the unmanned aerial vehicle ground station system to be solved is obtained1And a route planning management module U2Flight monitoring control module U3Data transmission module U4Data management module U5The reliability of (2) is assigned an optimal value.