The digital transformation establishes the development of intelligent technical systems, for which networks and inherent partial intelligence are mandatory features to enlarge their functionality. Self-optimizing systems as representatives of intelligent systems, are characterized by the autonomous objective-conform adaption of system behavior. Mechatronic systems provide the basic principles of this system class by sensing environmental and operating conditions as well as system states. These capabilities allow for precise manipulation of dynamic system behavior. The wide functionality of intelligent technical systems leads to an increasing system complexity, which is a serious threat regarding their dependability. In contrast to this threat, self-optimizing systems offer means to increase dependability by exploiting potentials of advanced behavior adaption, which are laid in the sophisticated support of the development process. In order to govern the increasing system complexity, the existing models of the development process need to be exploited to analyze dependability from early development stages to subsequent life cycle phases. These existing models are not yet sufficiently used in current approaches. A method for the integrated modeling of reliability, the central attribute of dependability, and dynamic system behavior is developed. Three application examples are investigated to show, that the method is capable of supporting the development of means to increase dependability in self-optimizing systems, the application on complex mechatronic systems and the setup of a digital twin in order to analyze dependability during operation.