Electric charge income prediction method based on historical data periodic description list

Abstract

The invention provides an electric charge income prediction method based on a historical data periodic description list, and the method comprises the steps: building the historical data periodic description list, building a matching mechanism, finding out the approximate regularity of the time period corresponding to the electric charge income on the historical data, and achieving the prediction of the electric charge income. Through the method, the regularity of the electric charge income can be discovered in different time periods, and the purpose of predicting the electric charge income byusing less attribute data is achieved.

Description

technical field [0001] The invention relates to a method for forecasting income data, in particular to a method for forecasting electricity fee income based on a periodic description list of historical data, and belongs to the technical field of power grid electricity fee management. Background technique [0002] For the cash flow of power grid management companies, if the cash inflow generated by electricity fee income within a certain financial period can be known more accurately, it is very important to maintain a good financial monitoring status of a power grid management company and determine the pace of operation and decision-making. Therefore, it is of great significance to forecast electricity revenue. [0003] For the prediction of cash flow and income, the current technology mainly adopts the following methods: 1. Directly rely on management and operating experience to formulate a list of cash inflows to predict cash flow; this method is more effective for enterpri...

AI Technical Summary

Problems solved by technology

[0003] For the prediction of cash flow and income, the current technology mainly adopts the following methods: 1. Directly rely on management and operating experience to formulate a list of cash inflows to predict cash flow; this method is more effective for enterprises with stable income sources and strong regularity (For example, you will get payment from other companies or the government at a fixed time period), and for customers who need to pay electricity bills, due to their large number and diverse characteristics and behaviors, it is difficult to form a strong regularity, making this method difficult to implement

2. Use traditional time series models (such as Markov models) to predict changes in time series. This method can be successful in relatively simple scenarios; however, this method assumes that the reasons behind the generation of time series can be obtained through existing attribute data to predict or have a strong association (such as: predicting the possibility of rainfall through changes in humidity and light), and for many customers who pay electricity bills, on the one hand, different customers may be associated with different attributes, and a large number of examples All possible attributes are difficult for grid companies to realize (many attributes such as marital status, health status, etc. are difficult for grid companies to collect); daily payment), and the payment cycle of customers is quite different (for example, the payment cycle of ordinary local residents, industrial and mining enterprises, and general merchants is different), and it is difficult to make effective predictions by using a unified model. to analyze

Images