3 results about "Business forecasting" patented technology

The invention relates to a dynamic adjust gyroscope life forecasting method based on gray model. By data collection of vibration effective value, random drift and environmental temperature parameter which are preprocessed using radial neural networks, influence of environmental temperature on vibration effective value and random drift is eliminated and random drift and effective value just related to time are obtained by subtracting drift constant value term, then trend term of vibration effective value and random drift are extracted by using wavelet transformation and gray model are built separately for their trend term. The smaller data in two values of life predicted of dynamic adjust gyroscope unless two predicted values exceeding performance parameter limitation when dynamic adjust gyroscope is considered losing effect. The invention uses performance parameter of life probative period of product to predict its life, showing discipline of performance parameter and life of dynamic adjust gyroscope. It is easy and convenient economical and reliable.

This invention provides a base station energy-saving method and apparatus based on service forecasting. The method includes: inputting real-time service load forecast data, real-time network status data, and real-time environmental data of the target base station into a decision model to obtain an energy-saving adjustment strategy output by the decision model; the decision model is obtained through reinforcement learning training based on a state space, action space, and reward function; the state space is constructed based on historical service load forecast data, historical network status data, and historical environmental data; the action space is constructed based on adjustable antenna parameters, adjustable carrier parameters, and adjustable computing power parameters; the reward function is constructed based on energy-saving benefit information and user experience information; and the energy consumption of the target base station is adjusted based on the energy-saving adjustment strategy. This method deeply integrates forecasting, multi-dimensional collaboration, and intelligent decision-making, enabling the generation of refined adjustment strategies based on dynamically changing service characteristics, thus improving the refined energy-saving control process of the base station.

The invention discloses a medium and long term prediction method and system for regional thunder and lightning climate, and the method enables the interannual change of thunder and lightning activity to be objectively predicted in the future 2-12 months through building the quantitative mapping relation between the thunder and lightning occurrence frequency distance level and the multi-source climate factor and the large-scale climate index. Therefore, thunder and lightning approaching early warning which is originally limited to a minute-to-hour level is expanded to seasonal and annual climate scales, and a thunder and lightning-oriented climate service prediction link is formed by uniformly utilizing historical thunder and lightning observation, air temperature, rainfall, an El Nino index and other data. The large-scale air-sea abnormal signals can be converted into quantitative judgment of too much and too little regional thunder and lightning, and the defect that in the prior art, only empirical correlation analysis stays is overcome. And meanwhile, by performing standardized grading on the lightning frequency interval, the grading results of'more, more, normal, less and less' and the like are directly output, so that the prediction product can be directly called and applied by the lightning protection and disaster reduction department, the planning and design department and the like.