The invention discloses a streaming type statistical definition and operation method based on a configuration file. The method comprises the following steps: creating a configuration file, wherein a content of the configuration file includes task information and a computing model; editing the task information, wherein the task information includes an attribute, a task name, a task type and a task description; editing the computing model, wherein constituent elements required by a computing model descriptive statistical process includes a data source definition, a mode definition, a statistical operation definition and an action definition; verifying the validity of the configuration file; carrying out validity management on the configuration file, wherein all streaming type statistical configuration files are uniformly managed by separate task configuration files and can take effect only being increased to the valid streaming type statistical configuration files in the task configuration file; and operating streaming type statistical tasks, wherein start parameters of a statistical service start time bandwidth include a task number, and in a statistical task execution process, all data are of a JSON format and are accessed via an attribute name. The method disclosed by the invention avoids the code maintenance, can be defined flexibly and is applied to scheduling an automation system.