In order to detect fire automatically, a forest fire image recognition method based on convolutional neural networks is proposed in this paper. There are two main types of fire recognition algorithms. One is based on traditional image processing technology and the other is based on convolutional neural network technology. The former is easy to lead in false detection because of blindness and randomness in the stage of feature selection, while for the latter the unprocessed convolutional neural network is applied directly, so that the characteristics learned by the network are not accurate enough, and recognition rate may be affected. In view of these problems, conventional image processing techniques and convolutional neural networks are combined, and an adaptive pooling approach is introduced. The fire flame area can be segmented and the characteristics can be learned by this algorithm ahead. At the same time, the blindness in the traditional feature extraction process is avoided, and the learning of invalid features in the convolutional neural network is also avoided. Experiments show that the convolutional neural network method based on adaptive pooling method has better performance and has higher recognition rate.

In thermal state diagnosis of substation equipment, problems such as low precision and slow speed usually exist. Aiming at the problems, an improved diagnostic scheme is proposed in this article. First, an infrared weak target extraction method based on local variance mapping and genetic algorithm threshold calculation is used to segment the region with abnormal thermal state in the equipment. Then the relationship between image gray parameters and temperature parameters of equipment region is established, and the improved relative temperature difference method is implemented to complete the classification and diagnosis of the thermal state of the equipment, and the abnormal area of the thermal state is captured and positioned at the same time. The experiment results show that the extraction accuracy and efficiency of the abnormal thermal area are improved based on the proposed scheme, which improves the fault tolerance of the thermal state diagnosis of the equipment and further ensures the operational stability of the substation and the entire power system.

The dynamic behavior of DC-DC converters is analyzed on the basis of a state-space averaged model including the feed back loop. The low frequency ripple behavior of DC-DC converters is analyzed in s field. A buck DC-DC converter is used as an example. The condition causing oscillation and impulse during starting period and in case of input perturbation is obtained. The output to input voltage transfer function in the feedback loop is deduced. The stability requirement is obtained based on Hurwitz stability criterion. The frequency and amplitude of the low frequency ripple voltage are determined by using zero-pole analysis method. The proposed approach is verified by simulation and experiment results