What Is a Load Curve?

The time-varying curve of various power loads in the power system. It is the basis for dispatching the power of the power system and planning the power system. The load of the power system involves various types of users in a wide range of areas. The power consumption of each user is very different, and it is impossible to know in advance when, where, and which type of load to increase. Therefore, the load variation of the power system is random. People use the load curve to describe the change of load with time, and study the regularity of load change accordingly.

The ratio of the average load to the maximum load in a day (month, year) is called the daily (month, year) load factor. The average load is the average value of the load power in a certain period (day, month, year). The load factor is a value less than 1. The load factors for different types of loads are also different. Non-ferrous metal smelting, steel, chemical, papermaking and other continuous production industries have load factors above 90%, and textile, machinery manufacturing and other industries have load factors of about 60%. The load coefficient also varies in different regions in different seasons, about 70 to 80%.

Forecast of the power load for the next 24 hours. It is the basis on which the power system dispatcher arranges daily dispatch plans, decides on start-up and shutdown plans, economically distributes loads, and arranges rotating reserve capacity. The accuracy of daily load curve prediction directly affects the economic benefits of power system operation. The daily load curve changes regularly. For example, the characteristics of the curves of each day in the same month and the same month are similar, and the typical daily load curve shapes of the same month in different years are similar. The daily load rate V} and the minimum load rate U can reflect the characteristics of the curve and Shape, and all have a close relationship with the structure of social power consumption and the division of electricity in various sectors. For example, the secondary industry in the system has a large proportion, the VU value is high, otherwise the VU value is low. Based on this feature, and considering the strength of our country The department's accumulation of historical data, this paper proposes a new daily load curve prediction method. This method decomposes the prediction process into two steps. The first step is to predict the characteristic parameters based on the analysis of the electrical structure. The second step is to use the characteristic parameters and The curve is predicted based on the reference load curve. Based on this, a mathematical model with clear physical meaning and simple expression is established. Based on the characteristics of the problem, a fast and effective algorithm is proposed. This method has been applied to the Northeast Power Grid load forecasting software and achieved good results.

The change law of the load curve of the power system is a non-stationary random process. If it is discretely measured at 1-hour intervals, a random time series can be obtained. Due to social factors such as people's production and living arrangements and the influence of natural seasonality, the changes in the load curve appear to be cyclical. From different time observations, it can be considered that the change of the load curve has a change cycle of one day, one week, one month, or one year. The forecast of daily load curve should make full use of the cyclical characteristics of this change.

Daily load curve prediction methods include multiple correlation algorithms, time series methods, and harmonic decomposition methods. However, these methods do not take into account the impact of meteorological conditions, and the load is closely related to meteorological conditions. For more accurate load forecasting, meteorological factors must be taken into account, and a meteorological load model or a necessary modification of the load model based on meteorological conditions must be taken to obtain a more realistic daily load curve forecast.

Load curve multiple correlation algorithm

From the load sample data (that is, the historical data of the load curve) to find the correlation of the power system load in each cycle, construct multiple prediction models, generally a first-order linear model. The predicted values and their variances obtained from each model are optimally combined to obtain a weighted average. According to the linear estimation theory, the weights should be inversely proportional to their respective variances, and the reciprocal of the variance of the weighted average is equal to the sum of the respective reciprocals of the variances. Holidays need special consideration, and the predicted value of a corresponding model is discarded. ^[3]

Load curve time series method

The load sample data is sequenced in chronological order. According to the truncation performance of the autocorrelation function and partial autocorrelation function of this sequence, an autoregressive model, a moving average model, or an autoregressive moving average model is established. In the prediction method, conditional expectation prediction, balanced linear minimum variance prediction or innovation method adaptive prediction can be used.