Les méthodes d'induction électromagnétique permettent de caractériser la conductivité électrique des matériaux, dont les corps planétaires telluriques, depuis les couches superficielles de la croûte jusqu'aux zones les plus internes, dans le manteau inférieur. Pour une source de champ électromagnétique donnée, des courants sont induits dans les matériaux qui y sont soumis. Avec l'essor des données magnétiques satellitaires, de nouvelles méthodes d'analyse des données magnétiques permettent d'obtenir des images unidimensionnelles de la structure électrique de ces corps car la structure spatio-temporelle de la source électromagnétique en est bien connue. Les travaux de mon doctorat ont eu pour but de mettre en place une nouvelle méthode d'analyse permettant de déterminer des modèles de structure interne globaux pour n'importe quel corps du système solaire pour lequel on dispose de longues séries temporelles magnétiques satellitaires. Après avoir testé cette méthode sur des modèles synthétiques et l'avoir appliqué au cas de données réelles terrestre pour lesquelles des études d'induction électromagnétiques antérieurs permettent d'avoir un a priori sur le modèle de conductivité électrique attendu, nous avons obtenu les premiers modèles de conductivité électrique martien en utilisant les données magnétiques du satellite Mars Global Surveyor. Ces résultats nous ont permis de valider des modèles de structure interne antérieurs établis à partir d'analyses géochimiques et minéralogiques des météorites martiennes. Cette méthode innovante est aujourd'hui la seule capable d'obtenir une image électrique des manteaux telluriques à partir de données magnétiques satellitaires pour des corps autres que la Terre ou la Lune et pour lesquels aucun a priori sur la structure spatio-temporelle du champ électromagnétique inducteur externe n'est nécessaire. / My Ph.D. work consists in the investigation of satellite magnetic data to infer the deep internal conductivity distribution. I developed a new global electromagnetic induction method applied to planetary magnetic datasets without strong a priori hypothesis on the external inducing source field. My method is based on a spectral correction of gapped data magnetic time series to restore the time spectral content of the source field. This external source depends on the planetary environment and is therefore different for each planetary bodies. The method aims at recovering with a maximum accuracy internal and external spherical harmonic coefficients of transients fields, whose ratio is used as a transfer function to retrieve the internal distribution of electrical conductivity. While for the Earth, a good proxy of the source field activity is the Dst index, no such proxy exists for other planets. Hence, for our study of Mars transient magnetic field from MGS, one of the major part of my work is the determination of an appropriate continuous proxy for the external variability. On Earth the external electromagnetic source is well known, and may be described by a spherical harmonic geometry dominated by the dipole term. This source field may be characterized using a magnetic activity index named the Dst index. The method has been tested on synthetic data generated within the framework of SWARM mission. This mission consists of a 3 satellites constellation. One of the main objectives is to infer the 3D electrical distribution in the deep Earth. SWARM synthetic data consist in a time series of spherical harmonic (SH) coefficients, external and internal, generated from a simple non-realistic 3D model. In this model, several regional and local conductors, in a radially symmetric 3 layers model have been embedded. Using this dataset, our method give satisfactory results. We have been able to obtain the external and internal SH coefficients - for the first SH degree, which is known to be the most energetic degree of the external source - using only one of the 3 synthetic time series. Then, the method has been used on real data from Ørsted. In this case, we had to pre-process the data to correct from ionospheric and aligned currents contributions. We developed a statistical analysis to remove the ionospheric field using 2 geomagnetic indices : AL and Kp. Hence, we have enlarged data toward higher and lower latitudinal zones than what has been done in previous works. Finally, we have been able to obtain 1D conductivity models, which fits reasonably with existing conductivity data in the deep Earth. Finally, we worked on Mars Global Surveyor (MGS) data. One of the most time consuming parts of this work was the determination of an appropriate continuous proxy for the external variability in the vicinity of Mars. Without any measurements of the IMF (Interplanetary Magnetic Field) during MGS sciences acquisition, we have used ACE (Advanced Composition Explorer) data. This satellite orbits around the L1 point of the Sun-Earth system, measuring solar wind magnetic characteristics. We have time-shifted ACE data to Mars position for 4 temporal windows where Mars and Earth were closed to the same Parker's spiral's arm, and finally determined a proxy explaining the major part of the variability observed in Mars data. Despite numerous gaps in MGS data, we have been able to establish the 1D conductivity distribution, fitting reasonably existing geochemical models. Although the method may be unstable for some cases, we obtained satisfactory results for in depth conductivity of the planet.