CN113255116B

CN113255116B - Split parallel simulation method for modeling of aircraft electromechanical system

Info

Publication number: CN113255116B
Application number: CN202110512439.2A
Authority: CN
Inventors: 饶玉章; 李成功; 夏成海
Original assignee: Sichuan Zhizhou Technology Co ltd
Current assignee: Sichuan Zhizhou Technology Co ltd
Priority date: 2021-05-11
Filing date: 2021-05-11
Publication date: 2023-01-03
Anticipated expiration: 2041-05-11
Also published as: CN113255116A

Abstract

The invention discloses a split parallel simulation method for modeling an aircraft electromechanical system, which analyzes modal complex eigenvalue and frequency response of the aircraft electromechanical system model through a frequency domain analysis principle, calculates and finds the split position of the system model, splits the system model into a plurality of models with independent input quantity and output quantity, strips and extracts the input quantity and the output quantity of each split model to form a single data variable, defines one end of each split model as an output end and the other end as an input end, respectively arranges an input module and an output module at the input end and the output end, writes variable data, adds a capacitive link or an inertial weight link, simultaneously adds an adaptive module, integrates and unifies models modeled by different software to realize parallel simulation without simplifying the models or sacrificing the simulation precision, so that the architecture of the system becomes clear.

Description

Split parallel simulation method for modeling of aircraft electromechanical system

Technical Field

The invention relates to the field of airplane simulation, in particular to a split parallel simulation method for modeling an electromechanical system of an airplane.

Background

The existing system simulation, especially the electromechanical system of an airplane, generally adopts a professional modeling tool for modeling, for example, a hydraulic system adopts AMESim, MWorks, DSHplus and the like; the environment control system adopts FloMASTER, dymola, AMESim and the like, the fuel system adopts AMESim, dymola, floMASTER and the like, the transmission system adopts SimulinX, AMESim and the like, the dynamic simulation adopts ADAMS, virtual. Labmotion, simpack and the like, the needed system simulation model is more and more complex along with the requirement of the simulation, on one hand, the related professional subjects are more and more, on the other hand, the units of the model are more and more complicated, the calculation efficiency is more and more slow, the calculation is not easy to converge, the system simulation requires the cooperation and parallel simulation of multiple subjects while the simulation efficiency and the simulation precision are improved, at present, different systems of the aircraft electromechanical system are modeled by using different modeling software, the related fields are more, the established models are more complex, the physical coupling is stronger and the parallel simulation is difficult, and the aircraft electromechanical system is seriously influenced.

Disclosure of Invention

Aiming at the problems, the invention provides a split parallel simulation method for modeling an electromechanical system of an airplane, which has the advantages of integrating different software models and realizing collaborative parallel simulation.

The technical scheme of the invention is as follows:

a split parallel simulation method for modeling an aircraft electromechanical system, the steps comprising:

s1, analyzing modal complex eigenvalues and frequency responses of an aircraft electromechanical system model through a frequency domain analysis principle, calculating and finding detachable positions of the system model, detaching the system model into a plurality of models with independent input interface modules and output interface modules, wherein each input interface module and each output interface module can independently output input quantity and output quantity;

s2, stripping and extracting the input quantity and the output quantity of each split model to obtain a single data variable;

s3, defining one end of each split model as an output end, defining the other end of each split model as an input end, connecting an output interface module at the output end, extracting the stripped output quantity in real time, creating variable data identical to the output quantity, writing the variable data into the input quantity of the output end in real time, connecting an input end interface module at the input end, extracting the stripped input quantity in real time, creating variable data identical to the input quantity, and writing the variable data into the input quantity of the input end in real time;

s4, adding a capacitive link or inertial weight at the output interface module and the input interface module of each split model;

s5, adding an adaptation module in the split model, wherein the adaptation module is used for solver numerical range adaptation and frequency adaptation;

and S6, performing cooperative completion of simulation iteration combination calculation on one or more input interface modules, one or more output interface modules and one or more adapter modules of the split model to obtain a final model.

In S1, the frequency domain analysis of the electromechanical system of the aircraft first needs to be linearized, and generally converted into a jacobian matrix:

where F' (x), the derivative of the function at point x, is the Jacobian matrix JF (x);

x∈Rn；

f (x) belongs to Rm, and JF (x) is m multiplied by n;

and then performing modal complex eigenvalue and frequency response analysis on the established main nonlinear working condition moments of the airplane electromechanical system model, wherein the analysis adopts a computer numerical integration method.

In the step S2, an API function is used for stripping and extracting the input quantity and the output quantity of the model.

In S3, the output interface module and the input interface module adopt the following method to ensure that the split model equation is complete and can be compiled, and the method is as follows:

1) And calling an Add input put () function to Add a co-simulation input port for the model. And calling an Add output put () function to Add a co-simulation output port for the model. The required mapping variables are added. Thus, the consistency of the variable number and the equation number is ensured;

2) The input module and the output module adopt a compiler which is the same as the source tool software to compile during packaging, and are hung in an application library of the tool software, so that unified compilation after connection is ensured.

In the step S4, a capacitive link or an inertial weight is added to increase the convergence efficiency of the model solution, and the existing capacitive link or the existing inertial weight link needs to unify all capacitive links or inertial weights at the interface.

In S5, the adaptation module is a second-order transfer function, then frequency parameters and amplitude parameters are opened, and then an application library is packaged into a tool software platform in the same way as the input and output modules and hung in the tool software platform to serve as a library of tool software, and the application library is called when a model is built. When the tool software solver solves the module, the solver has to reduce the maximum simulation step length and adapt to the precision range due to the convergence requirement.

The invention has the beneficial effects that:

1. the model which is difficult to carry out real-time simulation or semi-physical simulation before is used in a real-time simulation environment, the model is not required to be simplified, and the simulation precision is not sacrificed;

2. by splitting the model, the model which is too large in the prior art and has very low calculation efficiency even can be stably and efficiently calculated due to unconvergence;

3. by splitting the model, the rigidity matrix of the model is divided into modules, and the calculation efficiency is accelerated by several times;

4. the model collaborative simulation between different versions is easy through splitting;

5. the model cross-platform collaborative simulation is easy to realize through splitting;

6. the system model is modularly split, so that the architecture of the system becomes clear.

Drawings

Fig. 1 is a schematic split flow diagram of a split parallel simulation method for modeling an aircraft electromechanical system according to an embodiment of the present invention.

Detailed Description

The embodiments of the present invention will be further described with reference to the accompanying drawings.

Example (b):

as shown in fig. 1, a split parallel simulation method for modeling an electromechanical system of an aircraft includes the following steps:

s1, analyzing modal complex eigenvalues and frequency responses of an aircraft electromechanical system model through a frequency domain analysis principle, calculating and finding detachable positions of the system model, splitting the system model into a plurality of models with independent input quantity and output quantity;

the frequency domain analysis of the electromechanical system of the airplane firstly needs to be linearized and generally converted into a Jacobian matrix:

x∈Rn；

f (x) belongs to Rm, and JF (x) is m multiplied by n;

In order to guarantee the stability of numerical integration, the simulation step size needs to be limited to the order of the minimum time constant (equivalent to the reciprocal of the maximum eigenvalue norm). This is because, according to the derivation, the product of the numerical integration step size and the modulus of the feature root is a single digit.

The approximate stable condition can be deduced through a system characteristic equation, taking an Euler method as an example:

y'＝A y；

[y(n+1)-y(n)]/dt＝A y(n)；

y(n+1)＝(1+A*dt)y(n)；

the stable convergence condition of the series is |1+ A dt | <1; consider a as a component of a vector of eigenvalues (A1, A2.., am).

I1 Aidt I <1, i belongs to [1, m ]

Further | Ai × dt | <2 can be derived.

And (4) expanding the constraint condition, and limiting the simulation step length to be on the order of magnitude of the reciprocal of the maximum eigenvalue modulus when a stable approximate condition is obtained.

The minimum time constant in the model equation can be considered to be empirically determined, and the step size = (0.2 to 0.05) Tmin. Wherein Tmin is the smallest time constant in the state space, i.e. the state with the fastest change.

The step size may also be set according to the shear frequency point of the frequency response, taking into account the relation between the time constant and the frequency response. In most cases, according to the result of system identification, the open-loop frequency response of the object can be acquired more directly. Therefore, the method has more practical engineering significance.

At this time, step size =1/[ (5-20) wc ], where wc is the shear frequency in the open loop frequency response.

For a modeling object which changes rapidly, a time constant is generally smaller, a frequency response is higher, and due to the stability influence of simulation data, the coupling degree is very high, so that the modeling object cannot be split. On the contrary, for a modeling object with slow change, generally, the time constant is large, the frequency response is low, the splitting can be considered here, the modal complex eigenvalue and the frequency response of the aircraft electromechanical system model system are analyzed in the frequency domain, the relative frequency response at the cavity and the short axis is low, and the main point is selected for the splitting position. But the split position cannot be determined immediately. The simulation step length calculated in the process is larger than the parallel distributed collaborative simulation step length, and the step length is the position where the splitting can be carried out.

S2, stripping and extracting the input quantity and the output quantity of each split model to form a single data variable, calling an API (application program interface) function of simulation tool software through C/C + +, extracting the physical quantity, calling an Add input put () function to obtain model port input interface information and storing the model port input interface information into a data stream file, calling an Add output put () function to obtain model port output interface information and storing the model port output interface information into the data stream file, wherein the data file structurally comprises initialization, cyclic calculation and exit stages. When the port information is acquired, add output () or Add input () type functions can be directly written in the initialization paragraph, and the data types in table 1 are formed after stripping:

TABLE 1

The API function operates as follows:

a) And calling an Add input put () function to Add a co-simulation input port for the model. And calling an Add output put () function to Add a co-simulation output port for the model. Calling the AddParameter () function adds a co-simulation parameter port to the model. Calling an AddSignal () function to add a collaborative simulation signal port to the model;

b) Calling a GetCS input putData () function in the while loop body to obtain corresponding input port data on the simulation soft bus;

c) Calling a GetCSParaData () function to obtain corresponding parameter port data on the simulation soft bus;

d) And the split model code is in a ToDo: the model run is controlled to operate under the input g;

e) The transfer SetCS output putData () function of the split model code sends the output port data to the simulation soft bus;

s3, defining one end of each split model as an output end, defining the other end as an input end, connecting an output interface module at the output end, extracting the stripped output quantity in real time, creating variable data the same as the output quantity, writing the variable data into the input quantity of the output end, connecting an input end interface module at the input end, extracting the stripped input quantity in real time, creating variable data the same as the input quantity, and writing the variable data into the input quantity of the input end, wherein the output interface module and the input interface module adopt the following method to ensure that the split model equation is complete and can be compiled, and the method comprises the following steps:

1) And calling an Add input put () function to Add a co-simulation input port for the model. And calling Add output put () function to Add a co-simulation output port for the model. The required mapping variables are added. Thus, the consistency of the variable number and the equation number is ensured;

2) The input module and the output module adopt a compiler which is the same as the source tool software to compile during packaging, and are hung in an application library of the tool software, so that unified compilation can be ensured after connection;

s4, adding a capacitive link or inertial weight at the output interface module and the input interface module of each split model, so that the convergence efficiency of model solution can be effectively increased, and all capacitive links or inertial weights at the interface need to be put together in order in the existing capacitive link or inertial weight link;

capacitive links, as illustrated below:

for the dynamic closed cavity, when the pressure in the closed cavity is P, the volume is V; at increasing pressure P + Δ P, the volume is V- Δ V, whereby:

ΔP＝Ee×Δq×t/V

in the formula: ee: effective bulk modulus of elasticity

V: the total volume of the enclosed volume;

Δ q: during time t, the difference between the flow rates of the liquid into and out of the closed volume (no longer a change in volume)

Δ P: during time t, the pressure of the enclosed volume changes (either increases or decreases);

and (ap/t) is known as the pressure fly-up rate (change in pressure per unit time). It can be seen that there are three main factors, namely Ee, V and Δ q, that affect the pressure change in the closed volume.

As V increases, Δ P becomes smaller, that is, the frequency response becomes slower;

inertial mass, illustrated below:

for the short axis, the moment T is equal to the moment of inertia I multiplied by the angular acceleration α: t = I × α, finding the relationship of the moment angles in an integral manner:

T＝I*d^2θ/dt^2

in the formula: t: moment of force;

i: moment of inertia;

d ^2 theta/dt ^2: change of angle during time t;

also as I increases, the rotational frequency response of the shaft slows;

and S5, connecting an adaptation module at the output end of the split model, wherein the adaptation module is used for numerical range adaptation and frequency adaptation, the adaptation module is a second-order transfer function, then opening frequency parameters and amplitude parameters, and then packaging into an application library in the same way as the input module and the output module, hanging in a tool software platform, serving as a tool software library and calling when the model is built. When the tool software solver solves the module, the solver has to reduce the maximum simulation step length and adapt to the precision range due to the convergence requirement, the adaptation module has two adaptation modes, one is numerical range adaptation, the problem of large and small number matching calculation precision loss during calculation of the solver is solved, and the other is frequency adaptation, and the problem of step length self-adaptation during calculation of the solver is solved;

and S6, performing cooperative completion of variable data of input quantity and output quantity of one or more split models and the adaptive model to perform simulation iterative merging calculation, wherein n split models can be respectively put into n computers to perform parallel distributed simulation. The original model adopts a computer, the split system model carries out collaborative simulation calculation through the computer 1, the computer 2 \8230andthe computer n, the data stream of the parallel distributed simulation task split model is designed to be (control stream, clock stream and data stream) only belonging to the same domain, and the domains are isolated, so that different simulation tasks occupy different regions of a simulation soft bus and are not influenced mutually, and multi-user concurrent simulation is realized through the design of the data domain of the distributed simulation soft bus, and the final system model is obtained.

The above-mentioned embodiments only express the specific embodiments of the present invention, and the description thereof is more specific and detailed, but not construed as limiting the scope of the present invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the inventive concept, which falls within the scope of the present invention.

Claims

1. A split parallel simulation method for modeling an electromechanical system of an airplane is characterized by comprising the following steps:

s1, analyzing modal complex eigenvalues and frequency responses of an aircraft electromechanical system model through a frequency domain analysis principle, calculating and finding detachable positions of the system model, splitting the detachable positions into a plurality of models with independent input quantity and output quantity;

s3, defining one end of each split model as an output end, defining the other end of each split model as an input end, connecting an output interface module at the output end, extracting the stripped output quantity in real time, creating variable data identical to the output quantity, writing the variable data into the input quantity of the output end, connecting an input end interface module at the input end, extracting the stripped input quantity in real time, creating variable data identical to the input quantity, and writing the variable data into the input quantity of the input end;

s5, adding a solver adaptation module in the split model, wherein the adaptation module is used for numerical value range adaptation and frequency adaptation;

and S6, performing cooperative completion of simulation iterative computation on the output interface module, the input interface module and the adaptation module of the single or multiple split models together to obtain a final model.

2. The method of claim 1, wherein in S1, the frequency domain analysis of the electromechanical system of the aircraft, which first needs to be linearized, is converted into a jacobian matrix:

x∈Rn；

f (x) belongs to Rm, and JF (x) is m multiplied by n;

and then performing modal complex eigenvalue and frequency response analysis on the established main nonlinear working condition moment of the airplane electromechanical system model, wherein the analysis adopts a computer numerical integration method.

3. The method for split parallel simulation of modeling of aircraft electromechanical systems according to claim 1, wherein in S2, API functions are used for stripping and extracting the input quantity and the output quantity of the model.

4. The method for split parallel simulation of aircraft electromechanical system modeling according to claim 1, wherein in S3, the output interface module and the input interface module ensure that the split model equation is complete and can be compiled by the following method:

1) Calling an Add input put () function to Add a collaborative simulation input port for the model; calling an Add output put () function to Add a collaborative simulation output port for the model; newly adding required mapping variables; thus, the consistency of the variable number and the equation number is ensured;

2) The input module and the output module adopt a compiler which is the same as the source tool software to compile during packaging, and are hung in an application library of the tool software, so that unified compilation can be ensured after connection.

5. The method according to claim 1, wherein in S4, a capacitive link or an inertial weight is added to increase a convergence efficiency of model solution, and all capacitive links or inertial weights at an interface need to be integrated and unified.

6. The method for split parallel simulation of aircraft electromechanical system modeling according to claim 1, wherein in S5, the adaptation module is a second-order transfer function, then opens the frequency parameter and the amplitude parameter, then encapsulates the frequency parameter and the amplitude parameter into an application library in the same way as the input and output modules, hangs the application library in a tool software platform, serves as a tool software library, and is called when the model is built; when the tool software solver solves the module, the solver has to reduce the maximum simulation step length and adapt to the precision range due to the convergence requirement.