CN110705806A - A power prediction method based on capacity utilization hours - Google Patents

Abstract

The invention discloses an electric quantity prediction method based on the capacity utilization hours. The predictions are carried out according to different conditions of various industries. If the change trend is close, the industry electric quantity is predicted with the capacity utilization hours as a coefficient. For different trends of GDP, check the average utilization hours of capacity to predict the industry's electricity, and then sum it up to get the electricity forecast result. The method of the invention is more concise and feasible than the general intelligent algorithm for electric quantity prediction, avoids the uncertainty caused by self-learning in the parameter estimation in the intelligent algorithm, and makes the periodic electric quantity prediction result more accurate when it is released as a decision basis. The data changes to stabilize the actual operation trend of the load. This method adopts the linear prediction method, which is more suitable for the characteristics of power changes than the general quadratic or spline curve prediction and other non-intelligent methods, and avoids the appearance of abnormal distortion points. It is used to guide the planning and construction of power grids. More accurate control of investment scale.

Description

A power prediction method based on capacity utilization hours

technical field

The invention belongs to the field of power consumption forecasting of power grids, and particularly relates to a power forecasting method based on capacity utilization hours.

Background technique

In the power grid planning, in order to accurately track the load to determine the power supply capacity and the power supply capacity of the power grid, and realize the power balance of the power grid, power forecasting is generally performed, and direct forecasting methods are often used.

In view of the characteristics of less relevant historical data and more complex influencing factors for medium and long-term electricity forecasting, North China Electric Power University proposed an optimal combined forecasting model based on improved GM(1,1) and support vector machine. The model combines the improved gray prediction model and the support vector machine model, uses the leapfrog optimization algorithm to obtain the weight of each single model in the combined prediction model, and builds a combined prediction model based on the leapfrog optimization.

The optimized combined forecasting model was applied to my country's medium and long-term electricity forecasting, and the electricity from 1991 to 2005 was selected for analysis, and the electricity from 2006 to 2010 was forecasted, and compared with the general combined forecasting model and each single model.

Aiming at the problem that the traditional medium and long-term electricity forecasting method has a single idea and ignores the internal relationship between different levels of electricity forecasting and affects the accuracy of medium and long-term electricity forecasting, South China University of Technology proposed a new medium and long-term electricity forecasting method based on electricity industry classification.

First, the classification principles and methods of electricity consumption industry suitable for electricity forecasting are designed; then, based on 8 forecasting methods with complementary characteristics, an optimal combination forecasting model is established, and the electricity demand of the whole area to be forecasted and the electricity demand of each electricity consumption industry are separately analyzed. Annual and quarterly forecasts; finally, a two-dimensional and two-dimensional coordination model is established by using the multi-level forecast coordination theory, which corrects the electricity forecast value in the previous step, improves the forecast accuracy, and obtains a unified regional overall and the future annual year of each industry. and quarterly electricity forecasts.

According to the characteristics of electricity in different regions, China Agricultural University classifies the development law of electricity and electricity increments, and gives the corresponding prediction models of electricity and electricity increments. Based on the initial value sensitivity of chaotic motion and the analysis of the chaotic optimization search process, a parallel adaptive chaotic optimization method is proposed. On this basis, the parallel adaptive chaotic optimization method is used to determine the parameters of the electricity forecasting model, and the specific implementation steps and main measures are given.

SUMMARY OF THE INVENTION

The purpose of the present invention is to provide an electricity forecasting method based on capacity utilization hours. Assuming that the operating capacity of each industry is the maximum load peak value, the maximum load utilization hours of the capacity are positively obtained and compared with the maximum load utilization hours of the normal industry. , it is found that the deviation value is also large and cannot provide a reference.

Due to the unstable economic situation of various industries, not all industries will have a corresponding relationship between capacity and electricity at the same time. This method is only suitable for some industries to use in a certain period of time, and it needs to be used when the economic situation of the industry is favorable or stable. Below, this method can be used for short-term industry electricity forecasting.

According to the different situations of each industry, make predictions respectively. If the change trend is close, use the capacity utilization hours as a coefficient to predict the industry power. If the change trend is very different, according to the different change trends of the industry GDP, check the average capacity utilization hours to predict the industry. power.

Then sum it up to get the power prediction result.

In order to achieve the above object, the technical scheme adopted by the present invention is as follows:

A power prediction method based on capacity utilization hours, comprising the following steps:

(1) The i-th annual power of a certain industry n is Wn_i, the in-transit capacity is Sn_i, and the average capacity utilization hours is Kn, then Kn can be calculated from formula (1):

In the above formula, m represents the number of years of statistics;

n_m，则定义行业n在m年的增长平稳度Hn_m为式(2)所示，(2) The electricity increment in year i relative to the previous year is ΔWn_i, and the average increment in year m is

n_m, then define the growth stability Hn_m of industry n in m years as shown in formula (2),

Setting the limit ε, for an industry with a large Hn_m value (Hn_m>ε), it indicates that the growth of the industry is not stable, otherwise it is relatively stable, and for the case of ΔWi·ΔWi-1<0, it indicates that there is an inflection point in the growth, There has been a sharp change in growth trends;

(3) According to whether there is an inflection point and growth stability,

1) Set condition A as ΔWi·ΔWi-1>0 for all years and overall Hn_m<ε;

2) Set condition B as ΔWi·ΔWi-1>0 for all years and overall Hn_m>ε;

3) Set the condition C as the existence of one or more years ΔWi·ΔWi-1<0 and Hn_m>ε;

4) Set other conditions as condition D;

(4) For industries that meet different conditions, the following forecasting methods are adopted respectively:

1) Satisfy condition A, use formula (3) to calculate,

W _{n_i+1} =K _n S _{n_i+1} =K _n (S _{n_i} +ΔS _{n_i} ) (3)

Among them, ΔSn_i is the capacity of the newly added transformer that has been constructed and put into operation in the ith year;

2) Satisfying the condition B, it reflects that the power growth has changed significantly due to industry policies. For this reason, the change of the main indicator of economic development, GDP, has been predicted in more detail by segmented capacity utilization hours. Here, formula (1) is adjusted to formula (4),

Here p represents the number of segments according to GDP, and Kn-j represents the capacity utilization hours of the jth segment. Since the economic promotion policy has a certain continuity, the period is relatively stable, but there are large fluctuations in the early stage of formulation. for dealing with fluctuations;

3) If condition C is satisfied, then increase the segment of the inflection point,

Assume that the number of occurrences of ΔWi·ΔWi-1<0 in m years is q, the first occurrence is l year, l<=m, the current year's electricity is recorded as Wn_l, and the qth corresponding year's electricity is Wl+Δmq-1 , where Δmq-1 represents the difference between the qth time and the 1st time; the calculation method of condition B is used in each segment;

4) Satisfying the condition D indicates that the overall stability is stable, but there are inflection points locally, which are generally traditional industries. The inflection point is segmented, and the prediction method of condition A is used in each segment.

Compared with the prior art, the beneficial effects achieved by the present invention are as follows:

The method of the invention is more concise and feasible than the general intelligent algorithm for electric quantity prediction, avoids the uncertainty caused by self-learning in the parameter estimation in the intelligent algorithm, and makes the periodic electric quantity prediction result more accurate when it is released as a decision basis. The data changes and the actual operation trend of the load is stable, and the credibility is stronger, which is conducive to social recognition. This method adopts the linear prediction method, which is more suitable for the characteristics of power change than the general quadratic or spline curve prediction and other non-intelligent methods, avoids the occurrence of abnormal distortion points, and can be used to guide the planning and construction of power grids to control the investment scale more accurately.

Description of drawings

Accompanying drawing 1 is the line graph of the electricity data of the information transmission, computer service and software industry in a certain region from 2013 to 2016;

Accompanying drawing 2 is a line graph of electricity data of agriculture, forestry, animal husbandry and fishery in a certain area of the embodiment from 2013 to 2016;

Figure 3 is a line graph of the electricity data of the industry in a certain region from 2013 to 2016 in the embodiment;

FIG. 4 is the prediction result and result error analysis of electric quantity data of various industries in a certain region of the embodiment.

Detailed ways

The present invention will be described in further detail below in conjunction with the accompanying drawings.

Example 1

As shown in Figures 1-4, a method for predicting electricity based on capacity utilization hours includes the following steps:

Step 1: Calculate the average capacity utilization hours of a certain industry n;

Step 2: Calculate the growth stability of industry n in a specific year m years in step 1;

Step 3: Judging whether there is an inflection point and growth stability according to the growth value of industry n in a specific year m years in step 2;

Step 4: Predict electricity for industries that meet different conditions.

The calculation method of the average capacity utilization hours of a certain industry n is as follows:

The i-th year electricity of a certain industry n is Wn_i, the in-transit capacity is Sn_i, and the average utilization hours of the capacity are Kn, then Kn can be calculated from formula (1):

In the above formula, m represents the number of years of statistics.

The calculation method of the growth stability of industry n in a specific year m years is as follows:

则定义行业n在m年的增长平稳度Hn_m为式(2)所示，The electricity increment in year i relative to the previous year is ΔWn_i, and the average increment in year m is

Then define the growth stability Hn_m of industry n in m years as shown in formula (2),

Setting the limit ε, for an industry with a large Hn_m value (Hn_m>ε), it indicates that the growth of the industry is not stable, otherwise it is relatively stable, and for the case of ΔWi·ΔWi-1<0, it indicates that there is an inflection point in the growth, Growth trends have changed dramatically.

Determine whether there is an inflection point and growth stability in the growth value of industry n in a specific year m years;

Set condition A as ΔWi·ΔWi-1>0 in all years and overall Hn_m<ε;

The electricity forecast method of the industry that satisfies the condition A is calculated by formula (3),

W _{n_i+1} =K _n S _{n_i+1} =K _n (S _{n_i} +ΔS _{n_i} ) (3)

Among them, ΔSn_i is the newly added transformer capacity that is completed and put into operation in the ith year.

Embodiment 2

A power prediction method based on capacity utilization hours, comprising the following steps:

Step 1: Calculate the average capacity utilization hours of a certain industry n;

Step 2: Calculate the growth stability of industry n in a specific year m years in step 1;

Step 3: Judging whether there is an inflection point and growth stability according to the growth value of industry n in a specific year m years in step 2;

Step 4: Predict electricity for industries that meet different conditions.

The calculation method of the average capacity utilization hours of a certain industry n is as follows:

The i-th year electricity of a certain industry n is Wn_i, the in-transit capacity is Sn_i, and the average utilization hours of the capacity are Kn, then Kn can be calculated from formula (1):

In the above formula, m represents the number of years of statistics.

The calculation method of the growth stability of industry n in a specific year m years is as follows:

则定义行业n在m年的增长平稳度Hn_m为式(2)所示，The electricity increment in year i relative to the previous year is ΔWn_i, and the average increment in year m is

Then define the growth stability Hn_m of industry n in m years as shown in formula (2),

Setting the limit ε, for an industry with a large Hn_m value (Hn_m>ε), it indicates that the growth of the industry is not stable, otherwise it is relatively stable, and for the case of ΔWi·ΔWi-1<0, it indicates that there is an inflection point in the growth, Growth trends have changed dramatically.

Determine whether there is an inflection point and growth stability score in the growth value of industry n in a specific year m years;

Set the condition B as ΔWi·ΔWi-1>0 for all years and overall Hn_m>ε;

The electricity forecast method of the industry that satisfies the condition B reflects the significant changes in electricity growth due to industry policies. For this reason, the change of GDP, the main indicator of economic development, has been forecasted in more detail by segmented capacity utilization hours. Here, formula (1) Adjusted to formula (4),

Here p represents the number of segments according to GDP, and Kn-j represents the capacity utilization hours of the jth segment. Since the economic promotion policy has a certain continuity, the period is relatively stable, but there are large fluctuations in the early stage of formulation. Used to deal with fluctuations.

Embodiment 3

A power prediction method based on capacity utilization hours, comprising the following steps:

Step 1: Calculate the average capacity utilization hours of a certain industry n;

Step 2: Calculate the growth stability of industry n in a specific year m years in step 1;

Step 3: Judging whether there is an inflection point and growth stability according to the growth value of industry n in a specific year m years in step 2;

Step 4: Predict electricity for industries that meet different conditions.

The calculation method of the average capacity utilization hours of a certain industry n is as follows:

The i-th year electricity of a certain industry n is Wn_i, the in-transit capacity is Sn_i, and the average utilization hours of the capacity are Kn, then Kn can be calculated from formula (1):

In the above formula, m represents the number of years of statistics.

The calculation method of the growth stability of industry n in a specific year m years is as follows:

The electricity increment in year i relative to the previous year is ΔWn_i, and the average increment in year m is Then define the growth stability Hn_m of industry n in m years as shown in formula (2),

Setting the limit ε, for an industry with a large Hn_m value (Hn_m>ε), it indicates that the growth of the industry is not stable, otherwise it is relatively stable, and for the case of ΔWi·ΔWi-1<0, it indicates that there is an inflection point in the growth, Growth trends have changed dramatically.

Determine whether there is an inflection point and growth stability score in the growth value of industry n in a specific year m years;

The set condition C is that there are one or more years ΔWi·ΔWi-1<0 and Hn_m>ε;

If the electricity forecast method of the industry meets the condition C, the segment of the inflection point is added.

For the calculation of the segment of the inflection point, it is assumed that the number of occurrences of ΔWi·ΔWi-1<0 in m years is q, the first occurrence is l year, l<=m, the current electricity is recorded as Wn_l, and the qth time corresponds to The annual electricity is Wl+Δmq-1, where Δmq-1 represents the difference between the qth time and the 1st time; the electricity calculation method of condition B is used in each section.

Embodiment 4

A power prediction method based on capacity utilization hours, comprising the following steps:

Step 1: Calculate the average capacity utilization hours of a certain industry n;

Step 2: Calculate the growth stability of industry n in a specific year m years in step 1;

Step 3: Judging whether there is an inflection point and growth stability according to the growth value of industry n in a specific year m years in step 2;

Step 4: Predict electricity for industries that meet different conditions.

The calculation method of the average capacity utilization hours of a certain industry n is as follows:

The i-th year electricity of a certain industry n is Wn_i, the in-transit capacity is Sn_i, and the average utilization hours of the capacity are Kn, then Kn can be calculated from formula (1):

In the above formula, m represents the number of years of statistics.

The calculation method of the growth stability of industry n in a specific year m years is as follows:

则定义行业n在m年的增长平稳度Hn_m为式(2)所示，The electricity increment in year i relative to the previous year is ΔWn_i, and the average increment in year m is

Then define the growth stability Hn_m of industry n in m years as shown in formula (2),

Setting the limit ε, for an industry with a large Hn_m value (Hn_m>ε), it indicates that the growth of the industry is not stable, otherwise it is relatively stable, and for the case of ΔWi·ΔWi-1<0, it indicates that there is an inflection point in the growth, Growth trends have changed dramatically.

Determine whether there is an inflection point and growth stability score in the growth value of industry n in a specific year m years;

Conditions other than conditions A, B, and C in the first to third embodiments above are set as condition D.

Satisfying condition D indicates that the power consumption of the industry is generally stable, but there is a local inflection point, which is generally a traditional industry. In the inflection point segment, the electric quantity prediction method of condition A is adopted in each segment.

The method of the invention is more concise and feasible than the general intelligent algorithm for electric quantity prediction, avoids the uncertainty caused by self-learning in the parameter estimation in the intelligent algorithm, and makes the periodic electric quantity prediction result more accurate when it is released as a decision basis. The data changes and the actual operation trend of the load is stable, and the credibility is stronger, which is conducive to social recognition. This method adopts the linear prediction method, which is more suitable for the characteristics of power change than the general quadratic or spline curve prediction and other non-intelligent methods, avoids the occurrence of abnormal distortion points, and can be used to guide the planning and construction of power grids to control the investment scale more accurately.

Taking the power grid of a certain area as an example to illustrate the working process of the present invention as follows:

(1) Obtain the GDP statistics of various industries in the economic development of the region;

(2) Obtain the annual electricity statistics data of various industries in the region;

(3) Calculate the growth stability index Hn_m of each industry;

(4) Determine the industry adaptation conditions and make electricity forecast;

The corresponding curve with relatively stable change is shown in Figure 1.

Using formula (3), the forecast data for 2017 can be calculated.

If the growth trend is not stable, it is considered to meet the condition B, as shown in Figure 2, first use the formula (4) to calculate the capacity utilization hours in sections, and then use the formula (3) to calculate the predicted value of electricity consumption in 2017.

Here, 5 segments are taken according to GDP, and the access to each industry segment is as follows:

1) Coal mining and washing industry

When 0≤GDP≤5%, its K value is 1397. -5%≤GDP≤0%, its K value is 1125.

When 5%≤GDP≤10%, its K value is 1507. -10%≤GDP≤-5%, its K value is 1032

When 10%≤GDP≤20%, its K value is 1764. -20%≤GDP≤-10%, its K value is 1032, and its K value is 985.

2) Ferrous metal mining and dressing industry

When 0≤GDP≤5%, its K value is 1218. -5%≤GDP≤0%, its K value is 1125.

When 5%≤GDP≤10%, its K value is 1588. -10%≤GDP≤-5%, its K value is 1137

When 10%≤GDP≤20%, its K value is 1957. -20%≤GDP≤-10%, its K value is 1042.

3) Non-metallic mineral products industry

When 0≤GDP≤5%, its K value is 2527. -5%≤GDP≤0%, its K value is 2423.

When 5%≤GDP≤10%, its K value is 2631. -10%≤GDP≤-5%, its K value is 2162

When 10%≤GDP≤20%, its K value is 2936. -20%≤GDP≤-10%, its K value is 1790.

4) Ferrous metal smelting and rolling processing industry

When 0≤GDP≤5%, its K value is 2211. -5%≤GDP≤0%, its K value is 2148.

When 5%≤GDP≤10%, its K value is 2419. -10%≤GDP≤-5%, its K value is 2086

When 10%≤GDP≤20%, its K value is 2728. -20%≤GDP≤-10%, its K value is 2052.

5) Nonferrous metal smelting and rolling processing industry

When 0≤GDP≤10%, its K value is 1532. -10%≤GDP≤0%, its K value is 1336.

When 10%≤GDP≤20%, its K value is 1607. -20%≤GDP≤-10%, its K value is 1257

When 20%≤GDP≤30%, its K value is 2815. -30%≤GDP≤-20%, its K value is 1058.

For the inflection point that satisfies C, as shown in Figure 3, according to the inflection point segmentation processing of condition 4,

Here, 5 segments are taken according to GDP, and the access to each industry segment is as follows:

1) Coal mining and washing industry

When 0≤GDP≤5%, its K value is 1397. -5%≤GDP≤0%, its K value is 1125.

When 5%≤GDP≤10%, its K value is 1507. -10%≤GDP≤-5%, its K value is 1032

When 10%≤GDP≤20%, its K value is 1764. -20%≤GDP≤-10%, its K value is 1032, and its K value is 985.

2) Ferrous metal mining and dressing industry

When 0≤GDP≤5%, its K value is 1218. -5%≤GDP≤0%, its K value is 1125.

When 5%≤GDP≤10%, its K value is 1588. -10%≤GDP≤-5%, its K value is 1137

When 10%≤GDP≤20%, its K value is 1957. -20%≤GDP≤-10%, its K value is 1042.

3) Non-metallic mineral products industry

When 0≤GDP≤5%, its K value is 2527. -5%≤GDP≤0%, its K value is 2423.

When 5%≤GDP≤10%, its K value is 2631. -10%≤GDP≤-5%, its K value is 2162

When 10%≤GDP≤20%, its K value is 2936. -20%≤GDP≤-10%, its K value is 1790.

4) Ferrous metal smelting and rolling processing industry

When 0≤GDP≤5%, its K value is 2211. -5%≤GDP≤0%, its K value is 2148.

When 5%≤GDP≤10%, its K value is 2419. -10%≤GDP≤-5%, its K value is 2086

When 10%≤GDP≤20%, its K value is 2728. -20%≤GDP≤-10%, its K value is 2052.

5) Nonferrous metal smelting and rolling processing industry

When 0≤GDP≤10%, its K value is 1532. -10%≤GDP≤0%, its K value is 1336.

When 10%≤GDP≤20%, its K value is 1607. -20%≤GDP≤-10%, its K value is 1257

When 20%≤GDP≤30%, its K value is 2815. -30%≤GDP≤-20%, its K value is 1058.

See Figure 4 for the power forecast results and result error analysis of various industries.

The above-mentioned embodiments are only preferred embodiments of the present invention, rather than an exhaustive list of feasible implementations of the present invention. For those skilled in the art, any obvious changes made to it without departing from the principle and spirit of the present invention should be considered to be included in the protection scope of the claims of the present invention.

Claims (10)

1. A method for predicting electric quantity based on capacity utilization hours is characterized by comprising the following steps:

the method comprises the following steps: calculating to obtain the average utilization hours of the capacity of a certain business n;

step two: calculating to obtain the increase smoothness of the industry n in the specific year in the number of m years in the step one;

step three: judging whether an inflection point and growth stability exist according to the growth value of the n industry in the specific year, the number of the m years;

step four: and predicting the electric quantity of the industries meeting different conditions.

2. The method of claim 1, wherein the capacity utilization hours based power prediction method comprises: the average number of hours of capacity utilization of a certain business n is calculated as follows:

the i-th annual electric quantity of a certain business n is W_{n_i}At a transport capacity of S_{n_i}The number of capacity average utilization hours is K_nThen K is_nCan be calculated by the formula (1):

in the above formula, m represents the number of years counted.

3. The method of claim 2, wherein the capacity utilization hours based power prediction method comprises: the method for calculating the growth smoothness of the industry n in a specific year in parts of m years comprises the following steps:

the increment of the electric quantity of the ith year relative to the previous year is delta W_{n_i}Average increment of m years isDefining the growth smoothness H of the industry n in m years_{n_m}Is shown in a formula (2),

setting a limit value epsilon for H_{n_m}High value industry (H)_{n_m}>Epsilon) indicates that the industry is not growing smoothly, and vice versa, for Δ W_i·ΔW_i-1<The case of 0 indicates that an inflection point appears in the growth, and the growth trend changes sharply.

4. The method of claim 3, wherein the capacity utilization hours based power prediction method comprises: judging whether the inflection point and the growth stability exist in the growth value of the industry n in a specific year, namely m years or not is divided into four conditions:

1) setting the condition A as Δ W for all years_i·ΔW_i-1>0 and overall H_{n_m}<ε；

2) Setting the condition B as DeltaW for all years_i·ΔW_i-1>0 and overall H_{n_m}>ε；

3) Setting condition C as the presence of one or more years Δ W_i·ΔW_i-1<0 and H_{n_m}>ε；

4) The other case is set as condition D.

5. The method of claim 4, wherein the capacity utilization hours based power prediction method comprises:

the electric quantity prediction method meeting the condition A industry adopts the formula (3) to calculate,

W_{n_i+1}＝K_nS_{n_i+1}＝K_n(S_{n_i}+ΔS_{n_i}) (3)

wherein, Delta S_{n_i}And the capacity of the newly added and put into operation transformer is built for the ith year.

6. The method of claim 4, wherein the capacity utilization hours based power prediction method comprises:

the electric quantity prediction method meeting the condition B industry reflects that the electric quantity growth is obviously changed due to the industry policy, so that the change of the main economic development index GDP is predicted in more detail by using the small utilization of the segment capacity, wherein the formula (1) is adjusted to the formula (4),

where p represents the number of segments according to GDP, K_n-jThe number of hours of capacity utilization representing the jth segment is stable due to certain continuity of economic promotion policies, but the segment is mainly used for processing fluctuation at the beginning of the establishment.

7. The method of claim 4, wherein the capacity utilization hours based power prediction method comprises: and if the electric quantity prediction method in the industry meeting the condition C is met, increasing sections of inflection points.

8. The method of claim 7, wherein the capacity utilization hours based power prediction method comprises: the calculation of the segments of the inflection points, assuming that Δ W occurs in m years_i·ΔW_i-1<The number of 0 is q, the first occurrence is l years, l<M, the current year electricity is marked as W_{n_l}The electric quantity of the q-th corresponding year is W_l+Δmq-1Wherein Δ mq-1 represents the difference between the qth and 1 st times; and adopting an electric quantity calculation method of the condition B in each section.

9. The method of claim 4, wherein the capacity utilization hours based power prediction method comprises: the satisfaction of the condition D shows that the power consumption of the industry is generally stable, but inflection points occur locally.

10. The method of claim 9, wherein the capacity utilization hours based power prediction method comprises: and (4) segmenting at the inflection point, and adopting an electric quantity prediction method of the condition A in each segment.

Images