Research On The Evaluation Method Of Ship Fire Monitoring Software Based On Random Forest

When the ship is in navigation or mooring state,the marine fire monitoring software is an important part of the fire monitoring system.Its main function is to monitor the whole ship's fire and fire monitoring equipment,and display alarms for abnormal conditions.The software adopts comprehensive configuration technology,and the monitoring points cover fire monitoring equipment status monitoring,temperature monitoring,combustible gas monitoring and other types.The monitoring points of different cabins and passages form a large number of measurement point sets,and there are multiple related business logic processing relationships inside.In the current evaluation method of traditional marine fire monitoring software,the verification of the monitoring function requires the evaluation personnel to manually drive the data storage value,which can only change the state of a single measurement point at a time.In the evaluation of functional logic,there are special data such as illegal and abnormal error processes that cannot be output,which cannot fully meet the requirements of software evaluation.This paper proposes a method for evaluating marine firefighting monitoring software based on random forest,which mainly includes the generation and decision-making of random forest evaluation data model,in view of the problems of huge magnitude of measurement points,large variety of measurement points,and complicated logic of measuring point business processing.Two aspects of automatic evaluation.Through the in-depth study of this method and the verification of the actual application effect,compared with the traditional manual evaluation mode,the function,performance and boundary data evaluation of the measured point data of the software under test can be quickly completed,and the effect of shortening the consumption time of the evaluation and improving the efficiency of the evaluation.The main work of'the thesis is as follows:1.Analyze the principle of the algorithm and summarize the advantages and disadvantages of the current random forest algorithm,mainly including the random forest generation algorithm and the decision generation algorithm.This paper sorts out the commonly used evaluation methods of marine fire monitoring software,and mainly evaluates the measurement point function(including normal and abnormal functions),performance,and analog boundary of the software.2.Aiming at the key problems of marine fire monitoring software evaluation,this paper proposes a more intelligent and automated evaluation method,that is,a decision-making software evaluation method based on random forest.First,a random forest model is generated from a large amount of evaluation data related to the software to be tested.Then,according to the input of the evaluation personnel,the evaluation decision is automatically generated from the evaluation data model,and the evaluation data is formed into an evaluation drive script that meets the evaluation business.Through the automatic execution of the script,the evaluation of the software to be tested is quickly completed while covering the evaluation needs jobs.3.Design and build a method application verification system.The system is based on the idea of three-tier architecture,using random forest technology,and developed using Winform technology.Combined with the random forest evaluation data model generation and decision-making automatic evaluation algorithm,on the basis of satisfying the evaluation range of the software,performance,and boundary of the software to be tested,it can shorten the time-consuming of the evaluation work and improve the efficiency of the evaluation work.The feasibility and advanced nature of this method are proved by the system operation effect.