Research And Implementation Of Temperature Sensor Failure Detection And Temperature Prediction In Intelligent Storage Scenarios

Grain is the material on which mankind depends for survival and an important strategic material for the country.The development of agriculture must be placed in the first place of all economic development.At present,food crises are frequently occurring in various countries in the world.Our country can maintain a certain amount of grain stocks under the condition of self-sufficiency.Grain security has an important impact on the happiness of the people and the prosperity of the country,and it is directly related to social stability.In recent years,my country’s grain output has increased year by year,and the state has put forward higher requirements for grain storage,and proposed a strategy of “hiding grain on the ground and storing grain on technology”.In the intelligent warehouse management system,there are still problems such as poor management of measurement and control equipment and incomplete monitoring system functions.In order to solve these problems,research on grain situation information and establish related models,optimize the detection method of sensor failure hidden danger in intelligent warehouse management system,and realize the research and realization of temperature prediction in the case of sensor failure.The main research work of this thesis is as follows:(1)In view of the hidden trouble of the temperature sensor in the intelligent warehouse management system,the characteristics of the temperature sensor and grain storage application scenarios are analyzed,and the historical data of the target sensor and neighboring sensors are used as the basis for fault detection,and the use of extremes is proposed.The Gradient Boosting Algorithm(Extreme Gradient Boosting Decision Tree,XGBoost)establishes a sensor failure detection model,organically fuses the sensor failure detection in the storage management system and machine learning related knowledge,and finally detects the sensor failure risk.The method is optimized.The experimental results show that the XGBoost algorithm has a strong advantage in establishing a hidden fault detection method for sensors,and the accuracy of the diagnosis of hidden fault sensors can be as high as 95.4%.This method allows the staff to judge the working status of the sensor through the management system,which is beneficial to eliminate the hidden dangers of the food storage safety and reduce the food loss caused by the sensor failure.(2)In view of the problem that the temperature sensor of the granary cannot be detected,the characteristics of the storage application scene are analyzed,and the historical data of the target sensor is used as the basis for prediction,and the decision tree regression algorithm and the support vector regression(Support Vector Regression,SVR)algorithm,establishes the future temperature prediction model in the case of sensor failure,combines the future temperature prediction in the case of sensor failure with the time series prediction method in machine learning,and realizes the prediction of the future temperature of the failed sensor.The experimental results show that the future temperature prediction method established by the SVR algorithm performs better in the case of sensor failure,and its fitting accuracy and predicted temperature trend are more accurate,which solves the problem of the inability to monitor the temperature of the grain pile due to the temperature sensor failure.(3)Aiming at the problem that the grain pile temperature cannot be monitored when the temperature sensor fails,a micro server with the function of predicting the temperature of the fault sensor is designed and implemented.The server combines the support vector regression prediction model,uses the Golang programming language,and uses Json-formatted messages to interact with the client.The functions of the network layer,business layer and data layer are designed,and the fault sensor temperature is successfully deployed and operated.Forecast microserver.The server can meet the needs of the current intelligent warehouse management system,and make up for the inability to monitor the grain condition information in the case of sensor failures,and the designed micro-server and the grain condition monitoring system are relatively independent,which improves the reuse and scalability of the system Sex.In summary,this article focuses on the problem of the inability to obtain the temperature information of the grain pile caused by the hidden trouble of the temperature sensor in the intelligent warehouse management system.With the help of the spatio-temporal correlation between the granary sensor and the detected historical data,a series of studies are carried out to put forward the hidden danger of sensor failure.The improved detection scheme improves the diagnostic accuracy of sensors with hidden dangers;at the same time,a method for predicting future temperature in the case of sensor failure is proposed,and a micro-server with future temperature prediction function in the case of sensor failure is designed and implemented,which makes up for the deficiencies of the intelligent storage system,Has high scalability,can be connected to other distributed grain condition monitoring and control services,and has potential use value and application prospects.