The problems of the analysis of temperature errors fiber optic gyros (FOG) and problem of synthesis on minimization of these errors are put and decided. The mathematical models of thermal processes, temperature perturbances and thermal drift of fiber optic gyros are constructed and investigated. The capability is rotined and the conditions of the deterministic chaos originating in output signal FOG are detected at application to units temperature stabilization by Peltyer's thermocouples. Qualitative and quantitative estimations of drift parameters are obtained. The practical guidelines on decreasing of thermal drift FOG are developed.

When a gyro changes over from operating condition to non-operating condition, a rotor performs landing on supports. If the support was de-energized without preliminary deceleration of rotor rotation, the landing process is accompanied with intensive shock interactions between the rotor and its supports. The possibility of fail-safe rotor landing is discussed, assuming that rotor landing falls into emergency type if linear velocity of rotor center displacement exceeded an experimental value. It is supposed that interaction of the rotor and its supports can only result in one of two alternative events: either the gyroscope retains its performance capabilities or its reuse is impossible. The problem of probability determination is divided into two parts: simulation of a single landing process with random parameters; obtaining of a stochastic criterion. A model of shock interaction between the rotor and the support is used, basing on hypotheses of Routh and Newton. Formulas are derived for calculation of the proposed criterion value by the method of statistic tests. An illustrative example is shown for comparing a number of options of four supports position with equiprobable orientation of angular momentum vectors, initial linear velocity of rotor displacement and inertial load. However laws of the indicated vector distribution, amount of supports and their possible positional relationship, as well as shock-frictional properties of supports do not restrict application of a stochastic criterion for other types of gyro devices. The result verifies possibility to use fail-safe landing probability as a criterion of selecting a support disposition variant.

Construction principles and algorithms of correction of the analytical gyro vertical of shorten structure are considered. These units find successful use as a part of the track measuring systems. It is possible to use for this purpose a strapdown attitude orientation system. It is natural, that it will provide the maximum accuracy and will have the maximum cost. Another way is to use an analytical gyro vertical of shorten structure, containing only one or two laser gyros. In this case an additional information from accelerometers, odometer and another sensors of the measuring complex can be used to provide necessary accuracy. Article gives algorithms of the analytical gyro verticals and its errors estimations.

The developments of top foreign firms manufacturing periscopes are considered as examples for the trends of the submarine equipment with periscopes. High level is marked for periscope integration with optronic and radio observation systems mapping the external information, and appearance of optronic masts being new periscope constructions.

The following tendencies are among the main tendencies of Strapdown Inertial Navigation Systems (SINS) development:
the miniaturization of all the components of SINS from sensors to board computer;
the use of a redundant number of sensors and development of fault-tolerant SINS;
the increase of the frequency of data processing in SINS.

If these tendencies are taken into consideration many widely used approaches to the development of various components of SINS are to be revised. For special purpose SINS processors it means first of all the development of technique for rational architectures construction.

The analysis of SINS algorithms has been resulted in selection of calculating operation like (1). On one hand this operation is a base one for most of known SINS algorithms. On the other hand it determines a set of possible architecture solutions. Operation (1) concerns directly the architecture of Processor Elements (PE) that are the base of developed Function-Oriented Processors (FOP) SINS. PE architecture consists of sum unit (SUM), multiplier (MULT) and memory (MEM). To compare different PE architectures between each other parameters of time complexity LT and data processing frequency F are introduced. The form of LT and F estimates shows that architecture of PE5 (fig. 6) is the most rational. The quantitative estimates presented in the table 3 allow selecting PE5 as a base calculation unit of SINS FOP.

The next stage of development is constructing of rational FOP architecture. It begins with transforming of initial algorithms.

The approach to transforming the algorithms is illustrated with the Savage algorithm (fig. 7) as an example. The initial graph contains a set of tops related to each other and tops (operators) do not contain practically base relations. The resulting graph is weakly related and its tops are maximally oriented to base operation (fig. 8, 10). Then the problem of "covering" the transformed algorithm graph by FOP architecture is to be solved. Efficiency of data processing on the architecture of multi-processor vector FOP for Savage algorithm depends essentially on a number of PE. On the class of vector architectures the FOP architecture containing three PE like PE5 (table 3) is maximally efficient.

The results of the theoretical study and simulating of FOP architectures are applied to two practical systems. One of them is vector FOP based on 9 PE. The type of architecture is SIMD. Mass is 1.5 kg. Equivalent performance on the class of SINS algorithms is 8x10⁶ operations per second. It was developed in 1992.

The second of them is PE with RISC architecture. Mass is 0.5 kg. Equivalent performance is 10⁷ operations per second. It was developed in 1995.

The results of the study may be used in implementation FOP as a fragment of VLSI.

Two groups of algorithms are discriminated (algorithms of ideal operation and processing algorithms for the case of redundancy) and peculiarities are revealed for setting certification problem and certification execution for both these groups. Factors are mentioned that should be considered while certification of the data-processing program, particularly delay factor.

Some types of damage shock models are considered. Based on the distribution of damage, caused by a shock effecting a system, the intervals with small, intermediate and large damage are introduced. The initial homogeneous Poisson shock process is split into 3 homogeneous Poisson processes and studied independently. Several criteria of failure are considered, based on assumption that shocks with small level of damage are harmless for system, shocks with large level of damage results in system's failure and shocks with intermediate level of damage can result in system's failure only with some probability. The second model is based on an assumption that shocks with intermediate level of damage are harmless to a system, if they are not too close to each other. The probability of system's failure-free performance in [0,t) is derived explicitly. Simple asymptotic exponential approximations are obtained. The accuracy of this method is analyzed. Possible generalizations are discussed.

Considered is an operation check and calibration method for accelerometers with surge capability less than 1g in land conditions consisting in short-term arrangement of quasi-zero-gravity conditions due to the device free fall and fixation of its output signal values during the fall. The results are presented for the method experimental working-off.

In the abstract we examine the operation principle and mathematical model of the controlled stand made on the basis of the above-stated way for creation of the test means.

The controlled stand works in two modes. In the first mode - low-speed, from 2x10^-4 up to 0.5 rad/sec. - as inertial sensitive elements there are used the angular velocity sensors (AVS) with electrical feedback and three precision quartz accelerometers, which are measuring tangential acceleration of points of their fastening to the stand platform located at the angles 120^o from each other at distance R from the center of the platform. In the second mode - high-speed, from 0.5 up to 20 rad/sec. - as inertial sensitive elements there are used three precision quartz accelerometers, which are measuring acceleration, and three precision quartz accelerometers, which are measuring centripetal acceleration of points of fastening to the stand platform, also located at the angles 120? from each other at distance R1 from the center of the platform. Sensitive elements of stand measurement system are the same elements, which form control effect for each operation mode.

The tested device is fixed on the platform, which has a vertical rotation axis. Sensitive elements of stand control system, which are AVS and accelerometers, are fixed on the platform. Each of accelerometers has a feedback amplifier. The signals from feedback circuits of amplifiers come to the device that represents two-channel (channel 1- for centripetal acceleration meters, channel 2 - for tangential acceleration meters) summarizing amplifier.

he signals from outputs 1 and 2 of the amplifier come through DAC in PC to stand measurement system, as well as to engine control system through the adder. Thus in the range of small angular velocity (up to 0.5 rad/sec.) there is given out the command from PC providing the connection of adder input with output of feedback amplifier of AVS. During operation in the range of large angular velocity (more than 0.5 rad/sec.) similarly there is given out the command from PC providing the connection of adder input with output of summarizing amplifier channel, which ensures signal transmission of accelerometers measuring centripetal acceleration. The output of summarizing amplifier channel, which ensures signal transmission of accelerometers measuring tangential acceleration, is connected to the input of the adder in all operation modes. It means, that in the range of angular velocity values up to 0.5 rad/sec. stand control is carried out according signals of AVS and accelerometers measuring tangential acceleration. During operation in the range angular velocity values more than 0.5 rad/sec. stand control is carried out according signals of accelerometers measuring tangential and centripetal acceleration. The device of resistance moment minimization along rotation axis of the stand is entered into the controlled stand under the examined scheme, that's why a high stability of angular velocity specification is provided, especially in the range of small values. The friction moment in ball-bearing supports is minimized because the supports are completely unloosened from masses of the platform and tested device, this is achieved by "parrying" the weight of the platform with tested device by effort of torsion tension. Minimization of power supply devices moments is achieved at the expense of using of momentless face current-supplier and tracing mechanism, providing that the upper limit of angle of mismatching of current-supplier shoes positions is no more than ±3^o. Differential equations system of the controlled stand is received on the basis of the theorem about change of the movement quantity moment. For each operation mode of the controlled stand it is reasonable to make its own differential equations system.

Results of frequency characteristics calculation and mathematical modeling proves the stable work of precision stand with parameters of the selected correcting parts, thus in the first mode: reserve in amplitude - not less than 25 dB; reserve in phase - not less than 43^o; cut-off frequency - not less than 13 Hz; scale factor (constant) - 3^o/sec V. In the second mode: reserve in amplitude - 22-25 dB; reserve in phase - not less than 35-67^o; cut-off frequency - 10-12 Hz; scale factor (alternating) - not less 20^o/sec V.

The instability of specified angular velocity during stand operation in modes 1 and 2 was investigated by integration of differential equations (1) and (2) at i=3 by a method of Runge - Kutt. The results of integration have shown that examined interferences create instability of specified angular velocity in mode 1 - 1.6x10^-7 rad/sec., in mode 2 - 1.5x10^-6 rad/sec.

The article considers the role of B.I.Kudrevich as a founder of applied theory of gyroscopes, theory and practice of gyroscopic compasses and other gyroscopic devices in our country. The description of his scientific, engineering and educational work is presented, as well as the review of the principal articles that he had published.