Research On Fault Diagnosis And Recovery Method Of Temperature Control Sensor In Dry-type Transformer

With the continuous improvement of science and technology,China is moving towards high quality development,urban and rural power supply load is increasing,the use of dry-type transformers will become increasingly popular demand.The normal operation of transformers directly affects the entire power system.In order to ensure that the transformer is stable and working properly,it needs to be monitored uninterruptedly,and manual monitoring is too inefficient,so a special dry-type transformer temperature controller is needed to monitor and adjust the temperature of the transformer automatically.According to the site research and analysis of demand,a set of dry-type transformer temperature controller was developed,in order to ensure monitoring accuracy in line with national standards,using three groups of PT1000 temperature sensors redundant arrangement.And control the fan start for dry-type transformer internal cooling,if the temperature is too high then send out an alarm,more than the upper temperature limit then trip power off,to prevent the transformer due to high temperature and fire.However,the sensors used in existing dry-type transformer temperature monitoring devices have little self-confirmation,which means that the sensor’s operating status is always considered normal.As a result,in the event of a sensor failure,its output will be seriously skewed.If the monitor generates false alarms as a result,it may affect the normal operation of the power system and may lead to catastrophic consequences.Aiming at the problem that the dry-type transformer temperature monitoring sensor fails during the monitoring process and leads to wrong data output,resulting in the wrong response of the monitor and monitoring background.This study proposes a sensor fault diagnosis,location and recovery algorithm based on principal component analysis(PCA),sparrow search optimized long-short memory neural network(SSALSTM)and fault tree.Realize fault sensor location,fault diagnosis and fault sensor data recovery under dynamic process.First,use the temperature monitor developed in this research to collect the temperature data inside the dry-type transformer in real time;preprocess the collected data to form a training set for the experiment;use the training set to build a fault diagnosis model based on PCA and data recovery based on SSALSTM model;secondly,the fault type and cause of the sensor were analyzed,and a fault location model based on the fault tree was constructed;the temperature data collected by the thermostat was superimposed on the typical sensor fault state to form a simulation test set;the test set was used to carry out simulation experiments to verify The accuracy of fault diagnosis and the accuracy of data recovery;finally,the sensor fault diagnosis and recovery algorithm is constructed by combining PCA,fault tree and SSA-LSTM;through field experiments,it is verified whether the algorithm can ensure that the monitor will not go wrong even if there is a sensor fault Response to the question.The experimental results show that:(1)the monitor adopts redundant arrangement of sensors,and the temperature difference between the three groups of sensors conforms to the national standard,which can guarantee the monitoring accuracy well against each other.(2)The monitor can monitor the working temperature of the drytype transformer in real time to ensure that it works within the rated temperature range,effectively preventing the internal temperature of the transformer from becoming too high and causing accidents.(3)The fault diagnosis and recovery algorithm proposed for temperature control sensor faults is effective in diagnosing different faults of single or multiple sensors,with a fault recognition rate of over 96% and a diagnosis time of less than 1ms;it also has high prediction accuracy and good recovery effect,with an error within 0.1°C and strong generalization performance;field experiments have verified that the algorithm can significantly improve the stability of the monitor,even if the sensor fails The algorithm can ensure that the dry-type transformer works within the normal temperature range even if the sensor fails.