CN106295809B - The embedded real-time diagnosis reasoning algorithm pilot system of aircraft - Google Patents

Abstract

The invention discloses an aircraft embedded real-time diagnosis and reasoning algorithm test system, comprising: a main control device for registering the diagnosis and reasoning algorithm to be verified, and integrating the registered diagnosis and reasoning algorithm to be verified into an embedded target machine, And issue a fault generation command to generate fault data; the data simulation device is used to generate corresponding fault data according to the fault generation command sent by the main control device; the embedded target machine runs the diagnostic reasoning algorithm to be verified to perform Perform reasoning calculations on the fault data, and send the calculation results to the main control device; wherein, the main control device evaluates the performance of the diagnostic reasoning algorithm to be verified by analyzing the calculation results. The invention solves the difficulty existing in the prior art that the real-time diagnostic reasoning algorithm needs to be verified on the aircraft.

Description

Aircraft embedded real-time diagnostic reasoning algorithm test system

technical field

The invention relates to the field of aircraft fault diagnosis, in particular to an aircraft embedded real-time diagnostic reasoning algorithm test system.

Background technique

With the increasing functional complexity and system complexity of new aircraft, the fault diagnosis and health management systems for each system are becoming more and more complex. At the same time, the aircraft health management system puts forward higher requirements for the accuracy and efficiency of the diagnostic reasoning process. Therefore, in the development stage, designers need to face higher performance requirements, which brings more difficulties to development and production.

Fault diagnosis for airborne equipment is a technology involving multiple equipment, from data collection for the equipment under test, to importing equipment status information into the regional manager for diagnostic processing, and finally sending the diagnostic results to the ground for correlation maintenance decisions. This process puts forward high real-time requirements for the corresponding diagnostic reasoning engine algorithm. Therefore, in the development stage of the ground diagnostic reasoning algorithm, designers need to conduct multiple experiments to verify the algorithm to meet the design requirements. In order to carry out the test and analysis work of implementing the diagnosis design in the development stage of the airborne embedded diagnosis system, find design defects timely and effectively and make optimization adjustments, it is very necessary to develop the test analysis system of the aircraft embedded real-time diagnosis design.

Contents of the invention

The purpose of the present invention is to provide a system for testing the aircraft embedded real-time diagnostic reasoning algorithm, so as to obtain an effective and reliable aircraft embedded real-time diagnostic reasoning algorithm.

The test system of aircraft embedded real-time diagnostic reasoning algorithm of the present invention comprises:

The main control device is used to register the diagnostic reasoning algorithm to be verified, integrate the registered diagnostic reasoning algorithm to be verified into the embedded target machine, and issue a fault generation instruction for generating fault data;

The data simulation device is used to generate corresponding fault data according to the fault generation instruction issued by the main control device;

The embedded target machine performs reasoning and calculation on the fault data by running the diagnostic reasoning algorithm to be verified, and sends the calculation result to the main control device;

Wherein, the main control device evaluates the performance of the diagnostic reasoning algorithm to be verified by analyzing the calculation result;

Wherein, the fault generating instruction is a simulated fault generating instruction or an actual fault injection instruction or a historical fault reading instruction, and the fault data is the simulated fault data generated according to the simulated fault generating instruction or the actual fault injection instruction from the test The corresponding actual fault data obtained by the fault injection system or the corresponding historical fault data read from the historical fault database according to the historical fault reading instruction.

Preferably, the embedded target machine includes: a receiving module for receiving the task requirements; a reading module for reading in corresponding fault data from the data simulation device according to the task requirements; It is a calculation module for calculating the corresponding fault data by using the diagnosis reasoning algorithm to be verified.

Preferably, the main control device includes: a matching processing module for matching the number of fault types in the calculation result with the type and number of faults involved in the fault instruction; for matching according to the matching processing The degree is an evaluation module for evaluating the performance of the diagnostic reasoning algorithm to be verified.

Preferably, the embedded target machine further includes a computer resource occupancy module, configured to: receive an occupancy instruction issued by the main control device for occupancy of computing resources; according to an occupancy level included in the occupancy instruction, set the The ratio of non-diagnostic reasoning algorithms to computing resources in the load computing environment; in the case of assigning corresponding computing resources to non-diagnostic reasoning algorithms and diagnostic reasoning algorithms to be verified according to the occupation ratio, the diagnosis to be verified when computing resources are occupied is obtained Infer the solution efficiency and capability of the algorithm and send it to the master control device.

Preferably, the matching processing module of the main control device performs matching processing on the number of fault types in the calculation result and the number of fault types involved in the fault instruction; Evaluate the performance of the diagnostic reasoning algorithm to be verified when it is crowded out.

Preferably, the fault simulation model simulates the fault of the hydraulic servo system, and generates fault data including sudden change in gain of the electronic amplifier, slow change in gain of the electronic amplifier, and leakage in the hydraulic cylinder.

Preferably, the overcrowding module obtains the computing efficiency and capability of the diagnostic inference algorithm to be verified when computing resources are overcrowded through the following processes: making statistics on the occupancy of software and hardware resources during each diagnostic inference task running; The monitoring points are added to the program in the way of stubs, and finally the information of each location is summarized and counted to obtain the performance monitoring information during the task running.

Preferably, the allocation of corresponding computing resources for the non-diagnostic reasoning algorithm is realized by running equivalent computing programs.

Preferably, the running priority of the equivalent calculation program is higher than that of the diagnostic reasoning algorithm to be verified, so that the diagnostic reasoning algorithm to be verified is preferentially run according to the crowding instruction of the main control device.

Preferably, the computing resource occupancy includes CPU usage and memory usage of threads running the equivalent computing program.

The beneficial effect of the present invention is that by verifying the aircraft embedded diagnostic reasoning algorithm on the ground, the verification efficiency of the aircraft embedded diagnostic reasoning algorithm can be improved, the development of the aircraft embedded diagnostic reasoning algorithm can be accelerated, and the aircraft embedded diagnostic reasoning algorithm can be greatly reduced. verification cost.

The technical contents and effects of the present invention will be described in detail below in conjunction with the accompanying drawings.

Description of drawings

Fig. 1 is the schematic diagram of aircraft embedded real-time diagnosis reasoning algorithm test system of the present invention;

Fig. 2 is the hardware architecture diagram of the aircraft embedded real-time diagnostic reasoning algorithm test system of the present invention;

Fig. 3 is the hardware architecture diagram of the data simulator of the present invention;

Fig. 4 is the hardware architecture diagram of the embedded target machine of the present invention;

Fig. 5 is the software architecture diagram of the aircraft embedded diagnosis reasoning algorithm test system of the present invention;

Fig. 6 is the software architecture diagram of the main control device of the aircraft embedded diagnostic reasoning algorithm test system of the present invention;

Fig. 7 is the hardware architecture diagram of the data simulator of the present invention;

Fig. 8 is a software architecture diagram of the embedded target machine of the present invention;

Fig. 9 is a schematic diagram of the first embodiment of the aircraft embedded real-time diagnostic reasoning algorithm test method of the present invention;

Fig. 10 is a schematic diagram of the second embodiment of the aircraft embedded real-time diagnostic reasoning algorithm test method of the present invention;

Fig. 11 is a schematic diagram of the third embodiment of the test method for aircraft embedded real-time diagnostic reasoning algorithm of the present invention.

Detailed ways

Fig. 1 has shown aircraft embedded real-time diagnosis reasoning algorithm test system of the present invention, as shown in Fig. 1, this system comprises:

The main control device is used to register the diagnostic reasoning algorithm to be verified, integrate the registered diagnostic reasoning algorithm to be verified into the embedded target machine, and issue a fault generation instruction for generating fault data;

The data simulation device is used to generate corresponding fault data according to the fault generation instruction issued by the main control device;

The embedded target machine performs reasoning and calculation on the fault data by running the diagnostic reasoning algorithm to be verified, and sends the calculation result to the main control device;

Wherein, the main control device evaluates the performance of the diagnostic reasoning algorithm to be verified by analyzing the calculation result;

Wherein, the fault generating instruction is a simulated fault generating instruction or an actual fault injection instruction or a historical fault reading instruction, and the fault data is the simulated fault data generated according to the simulated fault generating instruction or the actual fault injection instruction from the test The corresponding actual fault data obtained by the fault injection system or the corresponding historical fault data read from the historical fault database according to the historical fault reading instruction.

The above-mentioned embedded target machine includes: a receiving module for receiving task requirements from the main control device, such as an interface module; a reading module for reading in corresponding fault data from the data simulation device according to the task requirements, For example; a calculation module for calculating the corresponding fault data by using the diagnostic reasoning algorithm to be verified.

The above-mentioned main control device includes: a matching processing module for matching the number of fault types in the calculation result with the fault types and numbers involved in the fault instruction; An evaluation module that describes the performance of the diagnostic reasoning algorithm to be verified.

The above-mentioned embedded target machine also includes a computer resource occupancy module, which is used to: receive an occupancy instruction issued by the main control device for occupancy of computing resources; according to an occupancy level included in the occupancy instruction, set The occupation ratio of non-diagnostic reasoning algorithms to computing resources in the airborne computing environment; in the case of assigning corresponding computing resources to the non-diagnostic reasoning algorithm and the diagnostic reasoning algorithm to be verified according to the occupation ratio, it is obtained when the computing resources are occupied to be verified Diagnose the solution efficiency and capability of the reasoning algorithm and send it to the main control device.

The matching processing module of the above-mentioned main control device performs matching processing on the number of fault types in the calculation result and the number of fault types involved in the fault instruction; the evaluation module of the main control device is occupied according to the matching degree of matching processing and computing resources When the calculation efficiency and capability of the diagnostic reasoning algorithm to be verified are evaluated, the performance of the diagnostic reasoning algorithm to be verified is evaluated.

The above-mentioned fault simulation model simulates the fault of the hydraulic servo system, and generates fault data including sudden change in the gain of the electronic amplifier, slow change in the gain of the electronic amplifier, and leakage in the hydraulic cylinder.

The above-mentioned crowding module obtains the computing efficiency and capability of the diagnostic reasoning algorithm to be verified when computing resources are crowded out through the following processing:

Make statistics on the occupancy of software and hardware resources during each diagnostic reasoning task run;

Add monitoring points to the program by inserting stubs, and finally collect statistics on each location information to obtain performance monitoring information during task running.

The present invention allocates corresponding computing resources for non-diagnostic reasoning algorithms by running equivalent computing programs. Wherein, the running priority of the equivalent calculation program is higher than that of the diagnostic reasoning algorithm to be verified, so that the diagnostic reasoning algorithm to be verified is preferentially run according to the crowding instruction of the main control device.

The computing resource occupation mentioned in the present invention includes the CPU occupancy rate and the memory occupancy rate of the thread running the equivalent calculation program.

The function, hardware architecture, and software architecture of the aircraft embedded real-time diagnostic reasoning algorithm test system of the present invention will be described in detail below in conjunction with the accompanying drawings. It should be pointed out that these descriptions are only used to explain the present invention, not to limit the present invention.

Functionally, the main control device can be divided into an operation task configuration function module, a task operation status monitoring and control module, a verification and evaluation function module, and as an embedded target computer upper computer to perform embedded target machine related configuration monitoring operations. The data simulation device can be divided into several modules for data acquisition, data simulation, data management and data transmission from the function. As a simulation of the onboard computing environment, the embedded target machine is responsible for running diagnostic reasoning algorithms to be verified, equivalent resource occupation computing tasks, and task running status monitoring tasks.

As the task initiator and control terminal of the test analysis system, the main control device first needs to configure the running task. Task running configuration needs to make necessary settings for the currently established test analysis task. The test analysis system supports the management and indexing of objects to be verified and algorithms through the database. Before the task runs, it is necessary to register the object to be verified and the corresponding algorithm, provide the name, model and modeling/operating conditions of the object to be tested, the algorithm function to be tested and analyzed, the name of the algorithm, the verification index and the name of the algorithm model. After the above-mentioned specific object information is confirmed, before each new task creation operation, input the detailed information of each task for the object to be tested and analyzed, and complete the new task creation.

The task running status monitoring and control module monitors the task running stage and provides feedback through the interactive interface during the task running process, and at the same time provides the authority and function to control the task running status in real time, and performs the operation of continuing or stopping the task. In the task running status monitoring and control module, the current task queue status can also be displayed through the user interface, and different currently running tasks can be selected for control.

The verification and evaluation function module realizes the analysis and evaluation of the performance evaluation ability, fault diagnosis ability and fault prediction ability of the test analysis and diagnosis reasoning algorithm. According to the selected verification algorithm, the measurement and evaluation indicators are selected for verification and analysis, and finally the results are compared with the regulations. Values are compared to make a qualified judgment, and a verification analysis conclusion is drawn.

The main control part also serves as the upper computer of the embedded real-time processor, and needs to configure and download the embedded real-time operating system Vxworks in the embedded target machine through the Workbench environment. On the basis of the completion of the operating system configuration, the diagnostic reasoning algorithm registered in the task information database needs to be added to the algorithm library of the embedded real-time processor after debugging in the Workbench environment on the premise of meeting the relevant requirements of the test analysis system In , form a correspondence with the task information database to support the calling requirements of the test analysis task.

The current test analysis system is mainly aimed at the verification and analysis of diagnostic reasoning algorithms for aircraft electromechanical systems and avionics systems. Therefore, the data simulation device is mainly designed based on the data acquisition and generation methods of the above objects, including rotary test bench data acquisition and hydraulic servo system simulation. Data generation, typical avionics system object state diagram simulation model data generation, and data acquisition methods such as public data.

From the perspective of functional division, the data simulation device needs to have the functions of data acquisition, data simulation, data management and data transmission. The data acquisition function needs to be supported by data acquisition hardware. The current data acquisition uses sensors to inject data from the fault injection system of the test bench. Acquisition, the signal collected by the sensor is transmitted to the data simulation device through the data acquisition card for storage. Data simulation provides a software environment for fault injection of simulation objects in electromechanical systems and avionics systems by providing a simulation software environment to generate fault data and store them in the monitoring and diagnostic data simulator. Data management provides data file management or database management environment for the above-mentioned collected or simulated generated data to perform data archiving and other operations, and record data-related information (collection time, working conditions, sampling rate, sampling time, etc.). At the same time, on the basis of the data filing operation, the corresponding data can be extracted from the database and a data file meeting the format requirements can be generated through the data generation command received from the main control device. Data transmission refers to supporting the transmission of data that meets the task requirements generated by the optical fiber network card between the data simulation device and the embedded target machine.

In addition, the data simulation device needs to realize the function of data verification, load the required data into the database through the pre-sampling instruction received from the main control device, and manually verify the corresponding technical requirements such as data format, integrity and correctness, Generate a data verification result report, and obtain correct and complete fault data files through multiple interactive verifications with the main control computer.

The embedded target machine is mainly divided into three parts in terms of functions, which are running diagnostic reasoning algorithms to be verified, computing tasks occupied by equivalent resources, and task running status monitoring.

The diagnostic inference algorithm to be verified is compiled and debugged in the Workbench environment of the host computer, and becomes a regular diagnostic inference engine that meets the requirements of the verification system, and the program is integrated into the overall program framework of the embedded target computer to support calling. The embedded target machine reads in the data to be processed by analyzing the task requirements forwarded by the monitoring and diagnostic data simulator, and calls the required algorithm for calculation to obtain the calculation result. The introduction of equivalent resource occupancy computing tasks is to restore the occupancy of computing resources by non-diagnostic reasoning algorithms in real airborne computing environments, and set the proportion of resource occupancy by analyzing the occupancy levels in task requirements. In the embedded real-time operating system vxworks, the priority of different tasks determines the power to use resources. Among the occupation tasks introduced, the computing resources of diagnostic reasoning tasks are preempted by injecting high-priority tasks to achieve equivalent The effect of resource consumption. The function of task running status monitoring is to make statistics on the occupation of software and hardware resources during the running of each diagnostic reasoning task, add monitoring points in the program by inserting stubs, and finally summarize and count the location information to obtain the task running period performance monitoring information.

The functional division of the test analysis system determines that the system hardware architecture cannot be realized without communication, so it is necessary to design a communication module that meets the system requirements. The communication module needs to support two types of communication methods, which are the task information data transmission part based on Ethernet and the diagnostic reasoning data transmission part based on fiber optic network card. Among them, the task information transmission part will undertake the task command transmission between the main control module, data simulator, embedded target machine and evaluation module, and realize a complete control command transmission data chain. Diagnosis and reasoning data transmission part is transmitted through the reflective memory fiber card.

Fig. 2 shows the hardware architecture of the aircraft embedded real-time diagnostic reasoning algorithm test system of the present invention. As shown in Figure 3, the overall hardware architecture of the system includes a main control computer as a main control device, a data simulation device and an embedded target machine. Among them, the main control software and verification and evaluation software are both running on the main control computer, and the main control computer adopts a high-performance workstation platform. Functions such as collection of verification data and simulation are realized in the data simulation device, and the data simulator is capable of transmitting verification data to the embedded real-time processor through optical fiber. The data simulator adopts a high-performance industrial computer platform that supports multi-channel acquisition cards/various types of board card expansion. The embedded target machine adopts the embedded target machine based on the airborne ICP architecture, which supports the verification data sent by the data simulator through the optical fiber, and realizes the simulation of the airborne computing environment.

Figure 4 shows the architecture of the data simulation device. The data simulation device needs to have high computing and processing capabilities to support the normal and efficient operation of various types of simulation models, and has various types of hardware expansion interfaces to connect different types of data acquisition cards to achieve support. The function of data collection, as well as the function of having a fiber optic network card interface and an Ethernet card interface to support various data transmission and interaction requirements.

The embedded real-time target machine is a single-chip microcomputer system based on a unified optical fiber network design. The system consists of an integrated motherboard and multiple LRM modules. All modules are integrated in the motherboard. If possible, spare module slots are provided for system function expansion. Its interior mainly includes a power module unit, a core computing unit, an optical fiber interface, an Ethernet interface, and a storage unit.

In order to meet the operating efficiency and good user experience of the test analysis system, the main control computer adopts a general-purpose high-performance workstation. At the same time, the main control computer has a reserved PCI interface to realize the communication with the data simulator and the embedded real-time processor through multi-channel Ethernet and the function of the host computer as the embedded real-time processor.

Fig. 5 shows the software architecture of the aircraft embedded real-time diagnosis reasoning algorithm test system of the present invention.

Figure 6 shows the software structure of the main control computer, which is divided into three parts from the function: task establishment, process control, verification and evaluation.

The task creation function is mainly realized through the verification object selection module, fault injection module and algorithm registration download module. First of all, for the algorithm verified for the first time, the algorithm information is registered to the main control software database through the algorithm registration and download module, and downloaded to the target computer through Workbench. The user selects object information through the verification object selection module, including algorithm application objects, operating environment parameters, and algorithm parameters. Then, select the source type of data, the size of the data set, the distribution of samples, the setting of fault trend, etc. through the fault injection module, and determine the equivalent task occupancy level that needs to be produced. Finally, the software translates the user's authentication setup requirements into control instructions.

The process control function is mainly realized through the cooperative control module and the communication module. First, put the started verification task into the verification task queue, and mark its status bit, then, by monitoring the working conditions of each system node, according to the priority and status of the task, send the task information to the idle state system node. Through the stack of verification tasks outside the system node, the work efficiency of the system node is maximized and the verification process is optimized. At the same time, the instructions and data information of each node of the verification task are recorded in the data storage module, so as to realize the full tracking and reproducibility of the verification task and ensure the objectivity of the verification task.

The verification evaluation function is mainly realized through the verification evaluation knowledge base module and the verification evaluation and result analysis module. The user first selects and creates a number of verification indicators based on the knowledge in the verification evaluation knowledge base, and their weights in the final total performance score to form a verification index scheme for the task. When the system detects that the task is in the state of algorithm completion, it will automatically call the verification evaluation and result analysis module to calculate the value of each index, compare it with the value required by the user, and give a verification evaluation conclusion, which can also be based on the weight of each index , giving a comprehensive performance score, which is convenient for horizontal comparison of similar algorithms.

Figure 7 shows the software architecture of the data simulation device, and its functional structure can be divided into a display control module, a data acquisition module, a data processing module and an interface module.

Display control module: responsible for the queue process control of the entire data generation, sending different types of data generation tasks to different interfaces. Convert the control instructions sent by the main control computer into the collection requirements of the test bench or the generation requirements of simulation models and historical data, and call the simulation model software or the data reorganization program based on MATLAB.

Data acquisition module: including three independent sub-modules of the acquisition program of the test bench, the calling program of the simulation model and the data reorganization program based on MATLAB. Receive the data acquisition instruction sent by the display control module, call the corresponding sub-module, and send the generated data to the data processing module.

Data processing module: Receive the original collected data sent by the data acquisition module, perform normalization processing according to the operation requirements of the algorithm, and send the generated data to the data storage module.

Data storage module: receive the data sent by the data processing module, store it as historical data, and send it to the interface module. At the same time, it is also provided as a data source to the historical data reorganization module.

Interface module: divided into two modes: optical fiber communication and Ethernet communication. The bottom send function based on C voice is responsible for sending the data to the embedded target machine through the fiber card. The interactive display module based on LabVIEW is responsible for displaying the information of the generated data to the user, and sending the data to the virtual embedded environment through Ethernet.

Fig. 8 shows the software of the embedded target machine of the present invention. The most important feature of the real-time operating system of the airborne embedded target machine is its high real-time performance, and it is different from the software installation method in the general computer operating system environment. The operating system and software loading need to be loaded through the Workbench in the host computer. The environment is downloaded, which limits the real-time loading of the algorithm to be tested and verified. Considering the optimization of the test analysis process, the method of loading multiple algorithms to be verified into the embedded real-time processor, and using the designed algorithm call module to dynamically call and configure the algorithms to be verified can greatly improve the test analysis. process efficiency. On the basis of the completion of the algorithm selection, the equivalent resource occupation tasks of the corresponding level are introduced, and the algorithm and the equivalent resource occupation program are run at the same time. At the same time, the task monitoring module also starts to perform monitoring tasks. After the diagnostic reasoning algorithm is completed, the result information is sent to the main control computer.

Since the embedded diagnostic reasoning calculation and task calculation in the actual airborne system mostly share the same computing resources, the change of task calculation will have a direct impact on the speed of diagnostic reasoning. In order to make the environment of the built test verification system close to the actual airborne computing environment, and avoid the phenomenon that diagnostic reasoning calculations can exclusively share CPU and storage resources, resulting in faster real-time diagnostic reasoning and differences from the real system, it is necessary to design in the system, etc. Efficient computing task generation module.

The equivalent computing task generation module designs tunable task computing threads by simulating the status and degree of CPU, memory and other resources occupied by task computing, which is used to simulate the occupation of CPU, memory and other resources equivalent to the actual system. Introduced into the embedded operating system target machine, it runs simultaneously with the diagnostic reasoning calculation thread, and simulates the occupation of CPU resources by task calculation, so as to match the real-time performance of the real system. Real-time deployment through real-time tuning and matching threads to simulate changes in CPU, memory and other resources in the real environment.

In order to realize the design of equivalent computing tasks, it is necessary to design a program to quantify the overall computing power, and then start the equivalent task computing program to occupy high-priority CPU and memory resources, forcing the computing power of diagnostic reasoning to be suppressed In order to simulate the actual situation of the real system.

There are mainly three threads running on the real-time processor: resource monitoring, real-time diagnostic processing and equivalent task calculation. What the system actually needs to simulate is that the computing resources of the real-time diagnostic processing program are occupied by different programs, so it is necessary to design a thread that can adjust the running priority of the program to plan the computing resources of the real-time processor.

Here, it is necessary to design an equivalent task calculation program whose running priority is higher than that of the real-time diagnostic processing program, so that the calculation efficiency and capability of the diagnostic calculation process can be truly simulated when the computing resources are occupied to varying degrees.

In fact, the most critical part of real-time tuning and matching is computing resource monitoring.

When the main frequency of the processor and the size of the memory are fixed, it is necessary to create a thread that can monitor the CPU usage and memory usage in real time to observe the CPU and memory usage, and then check the CPU and memory usage in different degrees. The diagnosis process and diagnostic ability of the case are tested in order to simulate the actual situation.

Further, it can be considered to use the floating-point computing power value to more accurately quantify the computing power of the real-time processor, and the stable computing times when it is fully loaded can be set as 100% load, and 0 when it is empty.

By monitoring the CPU and memory usage in real time, and pre-allocating computing resources for equivalent task calculations, the computing resources in the diagnosis and solving process can be controlled, and the real processor environment can be simulated indirectly.

The equivalent task calculation is designed as a stable calculation program that can adjust the calculation intensity. Its task is to adjust its calculation intensity in advance so that it can stably occupy the preset real-time processor computing resources. It is preliminarily estimated that it is designed into six gears according to the intensity level of the computing resources occupied, which are 100% (completely occupying real-time processor computing resources), 80%, 60%, 40%, 20% and 0% (not occupying at all) computing resources). The default design is 0%, which means that the real-time diagnostic processing program monopolizes the computing resources of the real-time processor, and then its gear can be adjusted according to the needs to achieve different intensities of resource occupation.

The equivalent task calculation program itself should be a calculation program that can adjust the calculation intensity. During its operation, real-time processor computing resources can be occupied relatively stably according to the preset gear. In the design, computing resources should be continuously monitored according to the resource monitoring program, and the equivalent task computing program should be adjusted to enable stable computing resources. occupy. The system operation process control module, the equivalent calculation task generation module and the diagnosis and reasoning engine algorithm evaluation module all run in the main control computer.

Fig. 9 shows the first embodiment of the test method of the aircraft embedded real-time diagnostic reasoning algorithm of the present invention, which includes: the main control device registers the diagnostic reasoning algorithm to be verified; the main control device integrates the registered diagnostic reasoning algorithm to be verified into the embedded target machine; the data simulation device starts the fault simulation model according to the simulation fault instruction of the main control device to generate the simulation fault data, and generates corresponding simulation fault data; the embedded target machine utilizes the diagnosis to be verified The reasoning algorithm performs reasoning calculations on the simulated fault data, and sends the calculation results to the main control device; the main control device evaluates the performance of the diagnostic reasoning algorithm to be verified by analyzing the calculation results.

Using the diagnostic inference algorithm to be verified to perform inferential calculation on the simulated fault data includes: receiving task requirements from the main control device; reading in corresponding simulated fault data from the data simulation device according to the task requirements; using diagnostic reasoning to be verified An algorithm calculates the corresponding simulated fault data.

The main control device evaluates the performance of the diagnostic reasoning algorithm to be verified by analyzing the calculation result, including: matching the fault type and quantity in the calculation result with the fault type and quantity involved in the simulated fault instruction Processing: Evaluate the performance of the diagnostic reasoning algorithm to be verified according to the matching degree of the matching processing.

Performing reasoning calculation on the simulated fault data by using the diagnosis reasoning algorithm to be verified further includes: receiving an occupation instruction including an occupation level from the computing resources of the main control device forwarded by the data simulation device; Occupancy level, setting the occupation ratio of non-diagnostic reasoning algorithms to computing resources in the airborne computing environment; in the case of assigning corresponding computing resources to non-diagnostic reasoning algorithms and diagnostic reasoning algorithms to be verified according to the occupation ratio, the calculation resource When it is crowded, the calculation efficiency and capability of the diagnostic reasoning algorithm are to be verified and sent to the main control device.

The main control device evaluates the performance of the diagnostic inference algorithm to be verified by analyzing the calculation results, including: matching the number of fault types in the calculation results with the number of fault types involved in the simulated fault instruction; The performance of the diagnostic reasoning algorithm to be verified is evaluated according to the matching degree of the matching process and the computing efficiency and capability of the diagnostic reasoning algorithm to be verified when computing resources are occupied.

The fault simulation model simulates the faults of the hydraulic servo system, and generates fault data including sudden changes in the gain of the electronic amplifier, slow changes in the gain of the electronic amplifier, and leakage in the hydraulic cylinder.

Obtaining the computing efficiency and capability of the diagnostic inference algorithm to be verified when computing resources are occupied includes: making statistics on the occupancy of software and hardware resources during the operation of each diagnostic inference task; adding monitoring points in the program by inserting stubs, and finally The performance monitoring information during the running of the task is obtained by summarizing the location information.

Allocating corresponding computing resources for non-diagnostic reasoning algorithms is realized by running equivalent computing programs.

The running priority of the equivalent calculation program is higher than that of the diagnostic reasoning algorithm to be verified, so that the diagnostic reasoning algorithm to be verified is preferentially run according to the crowding instruction of the main control device.

Computing resource occupancy includes CPU usage and memory usage of threads running the equivalent computing program.

Fig. 10 shows the second embodiment of the test method of the aircraft embedded real-time diagnostic reasoning algorithm of the present invention, which includes: the main control device registers the diagnostic reasoning algorithm to be verified; the main control device integrates the registered diagnostic reasoning algorithm to be verified into the embedded target machine; the data simulation device obtains the corresponding actual fault data from the fault injection system of the test bench according to the fault injection instruction injected into the actual fault data sent by the main control device; the embedded target machine utilizes the diagnosis to be verified The reasoning algorithm performs reasoning calculation on the actual fault data, and sends the calculation result to the main control device; the main control device evaluates the performance of the diagnosis reasoning algorithm to be verified by analyzing the calculation result.

Using the diagnostic reasoning algorithm to be verified to perform inferential calculation on the actual fault data includes: receiving the task requirements of the main control device; reading in the corresponding actual fault data from the data simulation device according to the task requirements; An inference algorithm performs calculations on said corresponding actual fault data.

The main control device evaluates the performance of the diagnostic reasoning algorithm to be verified by analyzing the calculation result, including: matching the number of fault types in the calculation result with the number of fault types involved in the fault injection instruction; According to the matching degree of the matching processing, the performance of the diagnostic reasoning algorithm to be verified is evaluated.

Performing inferential calculation on the actual fault data by using the diagnostic reasoning algorithm to be verified includes: receiving an occupation instruction including an occupation level forwarded by the data simulation device from the computing resources of the main control device; Level, setting the occupation ratio of non-diagnostic reasoning algorithms to computing resources in the airborne computing environment; in the case of assigning corresponding computing resources to non-diagnostic reasoning algorithms and diagnostic reasoning algorithms to be verified according to the occupation ratio, the calculation resources are obtained. When crowded, the calculation efficiency and capability of the diagnostic reasoning algorithm are to be verified and sent to the main control device.

The main control device evaluates the performance of the diagnostic reasoning algorithm to be verified by analyzing the calculation results, including: matching the number of fault types in the calculation results with the fault types and numbers involved in the fault injection instruction ; Evaluate the performance of the diagnostic reasoning algorithm to be verified according to the matching degree of matching processing and the computing efficiency and capability of the diagnostic reasoning algorithm to be verified when computing resources are occupied.

The main control device communicates with the data simulation device and the embedded target machine through LAN transmission, and the data simulation device communicates with the embedded target machine through optical fiber.

Obtaining the computing efficiency and capability of the diagnostic inference algorithm to be verified when computing resources are occupied includes: making statistics on the occupancy of software and hardware resources during the operation of each diagnostic inference task; adding monitoring points in the program by inserting stubs, and finally The performance monitoring information during the running of the task is obtained by summarizing the location information.

Allocating corresponding computing resources for non-diagnostic reasoning algorithms is realized by running equivalent computing programs. Wherein, the running priority of the equivalent calculation program is higher than that of the diagnostic reasoning algorithm to be verified, so that the diagnostic reasoning algorithm to be verified is preferentially run according to the crowding instruction of the main control device.

Computing resource occupancy includes CPU usage and memory usage of threads running the equivalent computing program.

Fig. 11 shows the third embodiment of the test method of the aircraft embedded real-time diagnostic reasoning algorithm of the present invention, which includes: the main control device registers the diagnostic reasoning algorithm to be verified; the main control device integrates the registered diagnostic reasoning algorithm to be verified into In the embedded target machine; the data simulation device reads out the corresponding historical fault data from its historical fault database according to the historical fault reading instruction sent by the main control device; the embedded target machine utilizes the diagnostic reasoning algorithm to be verified to The historical fault data is inferred and calculated, and the calculation result is sent to the main control device; the main control device evaluates the performance of the diagnostic reasoning algorithm to be verified by analyzing the calculation result.

Using the diagnostic inference algorithm to be verified to infer and calculate the historical fault data includes: receiving the task requirements of the main control device; according to the task requirements, reading in the corresponding historical fault data from the data simulation device; using the diagnostic inference to be verified The algorithm performs calculations on the corresponding actual fault data.

The main control device evaluates the performance of the diagnostic inference algorithm to be verified by analyzing the calculation results, including: matching the number of fault types in the calculation results with the fault types and quantities involved in the historical fault readout instruction; The matching degree of processing is used to evaluate the performance of the diagnostic reasoning algorithm to be verified.

Using the diagnostic inference algorithm to be verified to perform inferential calculation on the historical fault data includes: receiving an occupancy instruction containing an occupancy level from the computing resources of the main control device forwarded by the data simulation device; Level, setting the occupation ratio of non-diagnostic reasoning algorithms to computing resources in the airborne computing environment; in the case of assigning corresponding computing resources to non-diagnostic reasoning algorithms and diagnostic reasoning algorithms to be verified according to the occupation ratio, the calculation resources are obtained. When crowded, the calculation efficiency and capability of the diagnostic reasoning algorithm are to be verified and sent to the main control device.

The main control device evaluates the performance of the diagnostic reasoning algorithm to be verified by analyzing the calculation result, including: matching the fault type and quantity in the calculation result with the fault type and quantity involved in the fault injection instruction deal with;

The performance of the diagnostic reasoning algorithm to be verified is evaluated according to the matching degree of the matching process and the computing efficiency and capability of the diagnostic reasoning algorithm to be verified when computing resources are occupied.

The main control device communicates with the data simulation device and the embedded target machine through LAN transmission, and the data simulation device communicates with the embedded target machine through optical fiber.

Obtaining the computing efficiency and capability of the diagnostic inference algorithm to be verified when computing resources are occupied includes: making statistics on the occupancy of software and hardware resources during the operation of each diagnostic inference task; adding monitoring points in the program by inserting stubs, and finally The performance monitoring information during the running of the task is obtained by summarizing the location information.

Allocating corresponding computing resources for non-diagnostic reasoning algorithms is realized by running equivalent computing programs. Wherein, the running priority of the equivalent calculation program is higher than that of the diagnostic reasoning algorithm to be verified, so that the diagnostic reasoning algorithm to be verified is preferentially run according to the crowding instruction of the main control device. Computing resource occupancy includes CPU usage and memory usage of threads running the equivalent computing program.

The embedded real-time diagnostic reasoning algorithm demonstration system of the present invention lays a foundation for the test flow design of the test analysis system of aircraft embedded real-time diagnostic design. The method ensures the orderly conduct of the test.

Although the present invention has been described in detail above, the present invention is not limited thereto, and various modifications can be made by those skilled in the art based on the principle of the present invention. Therefore, any modifications made according to the principles of the present invention should be understood as falling within the protection scope of the present invention.

Claims (10)

1. a kind of test system of aircraft embedded real-time diagnosis reasoning algorithm, is used for verifying aircraft embedded diagnosis reasoning algorithm on the ground, it is characterized in that, described test system comprises: The main control device is used to register the diagnostic reasoning algorithm to be verified, integrate the registered diagnostic reasoning algorithm to be verified into the embedded target machine, and issue a fault generation instruction for generating fault data; A data simulation device connected to the main control device with the functions of collecting actual fault data, historical fault storage and fault simulation is used to generate corresponding fault data according to the fault generation instruction issued by the main control device; The embedded target machine connected to the main control device and the data simulation device is used to receive the registered diagnostic reasoning algorithm to be verified from the main control device, and receive the Fault data; by running the diagnosis reasoning algorithm to be verified, inferential calculation is performed on the received fault data, and the calculation result containing the number of fault types is sent to the main control device; Wherein, the main control device evaluates the performance of the diagnostic reasoning algorithm to be verified by analyzing the calculation result containing the number of fault types from the embedded target machine; Wherein, the data simulation device generating corresponding fault data according to the fault generation instruction issued by the main control device includes: When the fault generation command sent by the main control device is a simulated fault generation command, the simulated fault data generated by the data simulation device according to the simulated fault generation command; When the fault generation instruction issued by the main control device is an actual fault injection instruction, the corresponding actual fault data obtained by the data simulation device from the test bench fault injection system according to the actual fault injection instruction; When the fault generation command issued by the main control device is a historical fault reading command, the data simulation device reads corresponding historical fault data from the historical fault database according to the historical fault reading command. 2. test system according to claim 1, is characterized in that, described embedded target machine comprises: A receiving module for receiving task requirements; A reading module for reading corresponding fault data from the data simulation device according to the task requirements; A calculation module for calculating the corresponding fault data by using the diagnostic reasoning algorithm to be verified. 3. The test system according to claim 2, wherein said main control device comprises: A matching processing module for matching the number of fault types in the calculation result with the type and number of faults involved in the fault generation instruction; An evaluation module for evaluating the performance of the diagnostic reasoning algorithm to be verified according to the matching degree of the matching processing. 4. test system according to claim 2, it is characterized in that, described embedded target machine also comprises computer resource crowding-out module, is used for: receiving a crowd command sent by the main control device for crowding out computing resources; According to the occupation level included in the occupation instruction, set the occupation ratio of the non-diagnostic reasoning algorithm to the computing resource in the airborne computing environment; In the case of allocating corresponding computing resources to the non-diagnostic reasoning algorithm and the diagnostic reasoning algorithm to be verified according to the occupation ratio, the calculation efficiency and capability of the diagnostic reasoning algorithm to be verified when the computing resources are occupied are obtained, and sent to the main control device. 5. The test system according to claim 4, wherein the matching processing module of the main control device matches the number of fault types involved in the calculation result with the number of fault types involved in the fault generation instruction; The evaluation module of the control device evaluates the performance of the diagnostic reasoning algorithm to be verified according to the matching degree of the matching process and the computing efficiency and capability of the diagnostic reasoning algorithm to be verified when computing resources are occupied. 6. The test system according to claim 1, characterized in that, the failure simulation model of the data simulation device simulates the failure of the hydraulic servo system, and generates a sudden change in the gain of the electronic amplifier, a slow change in the gain of the electronic amplifier, and a failure in the hydraulic cylinder. Leaked fault data. 7. The test system according to claim 4, characterized in that, the described crowding module obtains the computing efficiency and capability of the diagnostic reasoning algorithm to be verified when computing resources are crowded by the following processing: Make statistics on the occupancy of software and hardware resources during each diagnostic reasoning task run; Add monitoring points to the program by inserting stubs, and finally collect statistics on each location information to obtain performance monitoring information during task running. 8. The test system according to claim 4, wherein the allocation of corresponding computing resources for the non-diagnostic reasoning algorithm is realized by running an equivalent computing program. 9. The test system according to claim 8, wherein the running priority of the equivalent calculation program is higher than that of the diagnostic reasoning algorithm to be verified, so that according to the crowding instruction of the main control device, priority Run the diagnostic inference algorithm to be verified. 10 . The test system according to claim 9 , wherein the computing resource occupancy includes CPU occupancy rate and memory occupancy rate of threads running the equivalent calculation program. 11 .