In connection with the significant progress in the improvement of sensitive elements (gyroscopes, accelerometers, magnetometers), as well as the increased possibilities of information processing by computational means, it became necessary to use modern algorithms for constructing correction systems for the gyromagnetic compass.
 The object of research is the correction circuits: with an adaptable proportional-integral-differential (PID) controller, with a Fuzzy controller, with a Kalman filter for correcting the gyroscopic meter according to the magnetic heading sensor. In this case, the well-known scheme of a gyromagnetic compass with a torque proportional-integral (PI) correction is taken as the basis for comparison. Smoothing of the oscillatory error of a magnetic compass, which may be predominant, is considered. The subject of research is the accuracy characteristics in the steady state.
 The study is carried out by simulating the operation of a gyromagnetic compass. The root mean square error (rms) and the mean value of the error are taken as the correction characteristics. At the same time, the transitional process of the initial exhibition is also controlled.
 In the algorithms under study, new solutions are applied.
 In the differential PID channel, a quarter-period delay of the dominant oscillations is applied, and the gain of the differential channel is adjusted according to the oscillation frequency. This setting allows you to almost completely smooth out the oscillatory error of the gyromagnetic compass by compensating for the oscillatory error with the received signal in the differential circuit.
 In the Kalman filter scheme, the resulting heading error estimate is in antiphase with the error. After the introduction of a delay in the estimate for half a period of fluctuations, the estimate almost completely corresponds to the error. As a result, the oscillatory error can be almost completely eliminated in the instrumental heading value.
 The article shows that all three investigated correction schemes show better characteristics in comparison with the known basic scheme. The highest accuracy can be achieved when applying the Kalman filter with the necessary settings for the perturbation, observation and initial error matrices.
 The simplest to implement is a circuit with an adaptable PID controller. Its characteristics are close to the scheme with the Kalman filter.