Research On Reliability Testing And Evaluation Of Radar System Simulation Software

With the constant maturity of radar theory and the development of simulation technology, the simulation about radar system has been approaching perfection. Simultaneously, the requirements for simulation also are gradually increased. Especially with the simulation structure becoming more and more complex, simulation scale becomes larger and larger, and the quality requirements of the simulation software are getting higher than before, which results in paying increasing attention to the simulation software. Therefore, it is necessary to study the reliability testing method, the reliability evaluation model and the reliability evaluation tool of the radar system simulation software. The contents are as follows.1. Aiming at the saturation effect problem of the traditional reliability testing methods, a mixed testing method is presented to test the reliability of radar system simulation software, which combines the reliability testing based on the operational profile with directed testing. At the same time, a new construction method of operational profile is presented according to the operation and its effect on the rate of the reliability of radar system simulation software.2. In order to deal with the problem of narrow application range of traditional single software reliability model and to meet the need of a large number of training data for standard neural network integrated model, a software reliability integrated prediction model based on grey theory, which is according to the fact that the characteristics of failure data is less of radar system simulation software, is proposed. Firstly, the unbiased grey GM(1,1) model is used to process the original data to weaken the randomness of data and to reduce the demand for the prediction data. Secondly, the end condition of network training is improved so that network error which meets the specified minimum is converted into powers exceeding the specified value before termination of training, which accelerates the training speed and reduces the demand for the training data. The prediction about the number of defects of radar system simulation software shows: the predicted results of integrated prediction model based on grey theory and neural network are more accurate than the traditional reliability model.3. A software reliability model based on functional modules is proposed for the reasons that the characteristics, such as the running time, complexity and the degree of importance of functional modules, are different, so the effects on the reliability of the simulation software are different. The G-O model is used as the basis to establish the functional model, and it is according to the running time, complexity and the degree of importance of functional modules to improve the model. The prediction about the number of defects of radar system simulation software shows that the predicted results of functional model are more accurate than the traditional reliability model.4. A reliability analysis and evaluation tool is designed for radar system simulation software. The tool can be used to implement the management of the failure data and the failure trend analysis, and to execute the reliability prediction that uses the traditional models and the established models based on the characteristics of the radar system simulation. It can also be used to compare the evaluation accuracy of the model which uses the U chart and Y chart. In order to deal with the problem of shortcomings of traditional KS distance which is used to evaluate the model prediction quality, the average KS distance is proposed to evaluate the model prediction quality.