Temperature prediction model generation method for logging instrument

The invention discloses a temperature prediction model generation method for a petroleum logging instrument, and belongs to the field of petroleum logging technologies and instruments. Comprising the following steps: designing a temperature prediction model based on an improved adaptive extended Kalman filter, and taking a system state transition matrix as a system identification parameter so as to improve prediction precision; simplifying a filtering gain matrix and a state one-step prediction covariance matrix in the temperature prediction model by using a Newton downhill method to reduce the calculation complexity; and implementing the simplified prediction algorithm in embedded equipment of the logging instrument, and uploading a real-time prediction result to a ground workstation. The method can accurately predict the temperature change trend of the logging instrument in dozens of minutes in the future, can timely avoid the damage of a circuit in the instrument caused by high temperature, and reduces the