Load characteristic statistics and hourly load curve generation method

Abstract

The invention mainly relates to the technical field of power loads, and provides a load characteristic statistics and hourly load curve generation method, which is characterized by comprising the following steps of: collecting current whole-network hourly load data of a power system; based on the current situation hourly load data, current situation whole-network annual load characteristics, current situation whole-network monthly / weekly load characteristics and current situation whole-network daily load characteristics are counted according to monthly and weekly division; based on the current situation whole-network annual load characteristic, the whole-network monthly / weekly load characteristic and the whole-network daily load characteristic, future whole-network electrical load prediction is carried out; on the basis of a future whole-network electrical load prediction result, supplementing annual future whole-network load characteristics by referring to current whole-network load characteristics; performing reasonable distribution of year, month and day scales on the partition load characteristics according to the load characteristics of the whole network, and supplementing the partition load characteristics; on the basis of load characteristics of the whole network and the partitions, hourly load curves of the whole network and the partitions are generated so as to support researches of safety and stability calculation, balance simulation, partition calculation and the like of a novel power system.

Description

Technical Field

[0001] The invention mainly relates to the technical field of electric load, in particular to a method for designing load characteristic statistics and hourly load curve generation. Background Art

[0002] The new power system is the most important component of the new energy system and the main carrier for realizing efficient and reliable consumption of new energy. The integration of water, wind, solar and storage and the integration of source, grid, load and storage are important connotations of the new power system. The power load of the power system is a very critical "load" element among the four major categories of source, grid, load and storage in the new power system. The statistical analysis of power load characteristics, the prediction of power load, the generation of power load curves, etc. will become a hot topic in the future research of new power systems.

[0003] Large-scale development and consumption of new energy is a key measure to accelerate the construction of a new power system. Due to the volatility, randomness and instability of the output of new energy such as wind power and photovoltaics, the large-scale grid connection of new energy has brought challenges to the safe and stable operation of the power system. At the same time, the demand for electricity load also shows the characteristics of volatility, which increases the difficulty of the power system to maintain source-load matching and supply-demand balance. The existing power system balance simulation technology will refine the simulation calculation period to the hourly level or even the 15-minute level to scientifically reflect the uncertainty of new energy output and cope with the volatility of electricity load demand. Therefore, it is necessary to reasonably generate the future load curve hour by hour based on the statistical characteristics of the current load, and generate the partition load curve based on the load curve of the whole network to support the safe and stable calculation, balance simulation, partition calculation and other related research of the new power system. Summary of the invention

[0004] Technical Problems to be Solved by the Invention

[0005] A method for load characteristic statistics and hourly load curve generation is provided, aiming to obtain the characteristics of power load, predict power load and generate power load curve based on the power load characteristics, so as to support related research such as safe and stable calculation, balance simulation and partition calculation of new power system.

[0006] The technical solution adopted by the present invention to solve the above technical problems

[0007] A method for load characteristic statistics and hourly load curve generation, comprising:

[0008] Step 1: Collect the current hourly load data of the entire power system, and divide the current hourly load data into monthly and weekly data to calculate the current annual load characteristics of the entire power system, the current monthly / weekly load characteristics of the entire power system, and the current daily load characteristics of the entire power system;

[0009] Step 2: Based on the current annual load characteristics, monthly / weekly load characteristics and daily load characteristics of the entire network, forecast the future power load of the entire network;

[0010] Step 3: Based on the prediction results of the future grid-wide power load characteristics and referring to the current grid-wide load characteristics, complete the future grid-wide load characteristics for the whole year;

[0011] Step 4: Generate the future hourly load curve of the entire network based on the forecast results of the future maximum load and annual power consumption of the entire network and the future load characteristics of the entire network after the full year is supplemented.

[0012] Furthermore, step 1 specifically includes:

[0013] Step 11, collecting hourly load data of the current status of the power system;

[0014] Step 12: Calculate the current annual maximum load FM of the entire grid based on the collected hourly load data of the power system max , divided by month, the current statistical status of the annual load characteristics of the entire network, monthly distribution RM m , obtain the seasonal imbalance coefficient ρ, divide the statistics by week, and calculate the weekly distribution of the annual load characteristics of the entire network RW w ;

[0015] Step 13: Statistical status of the daily distribution of the monthly load characteristics of the entire network RD m,d , get the monthly imbalance rate σ m ;

[0016] Step 14: Statistical status of the daily distribution of weekly load characteristics of the entire network RD w,t , and calculate the daily distribution of the weekly load characteristics of the entire network Considering holidays and shift adjustments, renumber the days of the week from 1 to 8 and calculate the equalization status after renumbering. Daily distribution of weekly load characteristics of the entire network

[0017] Step 15: Divide the current status of the daily load characteristics of the entire network by month and calculate the distribution RH m,d,h , get the first status quo daily average load rate of the entire network γ m,d , calculate the current daily load characteristics of the entire network and the monthly average hourly load Current monthly average daily load rate of the entire network And the upward fluctuation value of hourly load within the month ΔRH m,h and downward fluctuation value

[0018] Step 16: Statistical analysis of the current status of the daily load characteristics of the entire network by week: RH distribution w,t,h , and obtain the second status quo daily average load rate of the entire network γ w,t , calculate the current weekly average hourly load of the entire network Current weekly average daily load rate of the entire network And the upward fluctuation value of hourly load during the week ΔRH w,h and downward fluctuation value

[0019] Step 17: Statistical status by month: Monthly average load PM of the entire network m , the cumulative current hourly load of the entire network throughout the year FH m,d,h Get the current annual power consumption PM of the entire network sum , calculate the current load hours of the entire network H fh , analyze the current status of the total network load hours H fh Monthly distribution of annual load characteristics of the entire network RM m , Current status of the monthly load characteristics of the entire network Daily distribution RD m,d The current status of the daily load characteristics of the entire network divided by month is RH m,d,h The current status of the daily load characteristics of the entire network divided by week RH w,t,h The law of.

[0020] Furthermore, step 2 includes:

[0021] Step 21: Based on the current annual maximum network load FM max The current annual power consumption of the entire network PM sum Predict the maximum annual load FM′ of the entire network in the future max 、Future annual power consumption of the entire network PM′ sum and the future annual load utilization hours H′ of the entire network fh ;

[0022] Step 22: Obtain the future seasonal imbalance rate ρ' of the entire network, and adjust the monthly distribution of the annual load characteristics of the current entire network RM on a monthly basis based on the change in the seasonal imbalance coefficient (ρ'-ρ). m , and obtain the monthly distribution of the annual load characteristics of the entire network in the future RM′ m ;

[0023] Step 23: Based on the current status of the monthly average daily load rate of the entire network Predict the average daily load rate of the entire network in the future According to the change of daily average load rate Adjustment status by month Daily load characteristics of the entire network Monthly average hourly load Get the future network daily load characteristics and hourly load monthly average Combined with the forecast of the changes in the peak daily load of the entire network in the future, the monthly average hourly load of the entire network in the future is calculated. make reasonable amendments;

[0024] Step 24: Assuming that the load on each day of the month remains unchanged, the monthly imbalance rate σ of the current monthly load characteristics of the entire network mAs the initial value, combined with the monthly distribution of the annual load characteristics of the entire network in the future, RM′ m and the average daily load factor for the future month Calculate the first future load utilization hours Based on the assumed first future load utilization hours And the predicted future annual load utilization hours H′ fh The relative error of the monthly load characteristics is corrected to obtain the monthly imbalance rate σ m Get the monthly unbalanced rate σ′ of the future monthly load characteristics of the entire network m .

[0025] Furthermore, step 3 includes:

[0026] Step 31: Complete the monthly load characteristic daily distribution RD′ of the entire network in the future month by month m,d ;

[0027] Step 32: Completing the daily load characteristic distribution RH′ of the entire network in the future day by day m,d,h ;

[0028] Step 33: Perform demand-side management peak shifting. If the peak shifting rate is η, then some RM′ m RD′ m,d >1-η, and conduct peak shifting for demand-side management, and adjust the distribution of the future daily load characteristics of the entire network to RH′ m,d,h Adjust the maximum value to no more than Calculate the total value of the reduction Δinc in the hours with large load, calculate the total value of the increase Δdec in the hours with small remaining load, and calculate the daily load characteristic time distribution RH′ of the date to be adjusted m,d,h Hour by hour The proportion of the reduction is obtained to obtain the distribution of the daily load characteristics of the entire network after peak shifting.

[0029] Step 34: Adjust the daily distribution RD′ of the future monthly load characteristics of the entire network m,d :Based on the monthly distribution of the annual load characteristics of the entire network in the future RM′ m , Monthly load characteristics daily distribution RD′ m,d and daily load characteristics distribution Calculate the second future load utilization hours Based on the second future load utilization hours and the predicted future load utilization hours H′ fh The relative error of the monthly load characteristic imbalance rate σ′ in the future m Get the final future monthly load characteristic monthly imbalance rate Based on the relative change in the monthly imbalance rate before and after the correction Adjust the daily distribution RD′ of the load characteristics of the future month on a daily basism,d , and obtain the predicted value of the daily distribution of the final monthly load characteristics

[0030] Furthermore, in step 31, the daily distribution RD′ of the future monthly load characteristics of the entire network is supplemented month by month. m,d The method is any one of the following 4:

[0031] (1) Changes in monthly imbalance rate based on current and future conditions (σ′) m -σ m ) for the current daily distribution of the monthly load characteristics of the entire network RD m,d Adjust daily to get the daily distribution RD′ of the future monthly load characteristics of the entire network m,d ;

[0032] (2) Based on the current weekly distribution of the annual load characteristics of the entire network RW w and daily distribution of average weekly load characteristics Calculate the daily distribution of load characteristics for each week And converted into the initial value of the daily distribution of monthly load characteristics Find the monthly maximum value for the initial value of the daily distribution of monthly load characteristics Normalize to get calculate The average value of the monthly imbalance rate Based on the monthly imbalance rate change Adjust the normalized initial value of the daily distribution of monthly load characteristics on a daily basis Get the future monthly load characteristic daily distribution RD′ m,d ;

[0033] (3) Based on the predicted monthly distribution of future annual load characteristics RM′ m , determine the control week within the month The load characteristics of the control week are equal to the predicted value of the monthly load characteristics RM′ m , and the load characteristics of other weeks except the control week are obtained by interpolation, so as to obtain the initial value of the annual load characteristic weekly distribution Initial value of weekly distribution based on annual load characteristics Calculate the daily distribution of weekly load characteristics And converted into the initial value of the daily distribution of monthly load characteristics Normalized to get calculate The average value of the monthly imbalance rate Based on the monthly imbalance rate change Adjust the normalized initial value of the daily distribution of monthly load characteristics on a daily basis Get the future monthly load characteristic daily distribution RD′ m,d; Wherein, the control week is determined in the following manner: if a month involves 5 weeks, the third week is the control week of the corresponding month; if a month involves 6 weeks, if the maximum load RM′ of the next month m+1 Less than the maximum monthly load RM' m , then the third week is the control week, otherwise, the fourth week is the control week;

[0034] (4) Based on the predicted monthly distribution of future annual load characteristics RM′ m , determine the load characteristics of the middle, first and last ten days through interpolation, and superimpose the daily distribution of the average weekly load characteristics Get the initial value of the daily distribution of monthly load characteristics Converted to normalized initial value of daily distribution of monthly load characteristics calculate The average value of the monthly imbalance rate Based on the monthly imbalance rate change Adjust the normalized initial value of the daily distribution of monthly load characteristics on a daily basis Get the future monthly load characteristic daily distribution RD′ m,d .

[0035] Furthermore, in step 32, the daily load characteristic time distribution RH′ of the entire grid in the future horizontal year is supplemented day by day. m,d,h The method uses any of the following 3 methods:

[0036] (1) According to the principle of the same maximum load control hours per day within a month, check the current daily load characteristics and distribution RH every day m,d,h , and the predicted monthly average hourly load of the entire network in the future Control the dates with different hours, exchange and sequence the peak hour load, and obtain the time distribution of the daily load characteristics after sequencing And calculate the monthly average value to get the mean value of the daily load characteristic distribution after sequencing The change in the monthly average of future hourly load and the monthly average of current hourly load after adjustment The time distribution of daily load characteristics after sequence adjustment Adjust hourly to obtain the daily load characteristic time distribution RH′ of each day in the future month m,d,h ;

[0037] (2) Refer to the weekly average of the current hourly load divided by week As the initial daily load characteristic time distribution of the week for future prediction, the random fluctuation of the load is increased every day of the week, and a random fluctuation control parameter λ is set, which represents no fluctuation, overall fluctuation, peak fluctuation, valley fluctuation, peak valley fluctuation or random combination floating strategy. The load characteristics of 24 hours are sorted from large to small as h', and the hours to be fluctuated are grouped into a set Hset according to different floating strategies. When h'∈Hset, hourly fluctuation is required, and a random number rand(-1,1) between -1 and 1 is taken. According to the positive and negative values ​​of the random number and the downward fluctuation value of the hourly load in the week, the hourly fluctuation is calculated. and the upward fluctuation value ΔRH w,h , calculate the hourly load floating value δ(h), update the daily load characteristic time distribution of each day in the week, and perform peak hour load interchange and sequencing on the updated daily load time distribution of each day in the week according to the principle of the same daily maximum load control hour in the month, and obtain the daily load characteristic time distribution after sequencing And calculate the monthly average value to get the mean value of the daily load characteristic distribution after sequencing Changes based on the monthly average of hourly load The time distribution of daily load characteristics after sequence adjustment Adjust hourly to obtain the daily load characteristic time distribution RH′ of each day in the future month m,d,h ;

[0038] (3) Based on the monthly forecast of the average monthly hourly load of the entire network As the initial time distribution of January in the future forecast, the random fluctuation of load is increased every day in the month, and the random fluctuation control parameter λ is set to select the floating strategy and determine the hours to be floated. According to the positive and negative values ​​of the random number rand (-1,1) and the downward fluctuation value of the hourly load in the month and the upward fluctuation value ΔRH m,h , calculate the hourly load fluctuation value δ(h), and update the load characteristic distribution of each day in the month to obtain And calculate the monthly average value to get the updated hourly load monthly average value Changes based on the monthly average of hourly load Updated daily load characteristics and time distribution Adjust hourly to obtain the daily load characteristic time distribution RH′ of each day in the future month m,d,h .

[0039] Furthermore, step 4 is specifically as follows: the predicted future maximum network load FM′ max , multiplied by the monthly distribution of the future annual load characteristics of the entire network RM′ m , Modified monthly load characteristics daily distribution and daily load characteristics distribution Get the future hourly load data of the entire network FH′ m,d,h , based on the future hourly load data of the entire network FH′ m,d,h Generate the future hourly load curve of the entire network.

[0040] Furthermore, the method for generating load characteristic statistics and hourly load curves according to the present invention further includes step 5: generating future partition hourly load curves, which specifically includes the following steps:

[0041] Step 51: Based on the temporal correlation of the current future growth rate forecast and the spatial correlation of the network partition characteristics, the maximum load of the partition in the future design level year is predicted. Annual electricity consumption by region and zone load utilization hours

[0042] Step 52: Obtain the future predicted partition load characteristics, including the monthly distribution of the future partition annual load characteristics. The average daily load factor of the future zone in the month Future zone daily load characteristics Hourly load monthly average Initial value of monthly imbalance rate of monthly load characteristics of future zones

[0043] Step 53, similar to step 24, combined with the monthly distribution of the annual load characteristics of the partition and the average daily load factor of the zone for the month Assuming that the load on each day of the month remains unchanged, calculate the assumed load utilization hours of the partition Assumed value based on load utilization hours of the zone With the predicted value The relative error of the initial value of the imbalance rate of the future partition Adjust to get the predicted value of monthly imbalance rate of monthly load characteristics of the partition

[0044] Step 54: Perform the daily distribution of the monthly load characteristics of each zone on a monthly basis similar to step 31 Completion, similar to step 32, daily load characteristics time distribution Similar to step 33, the daily load characteristic time distribution is adjusted for some dates that exceed the demand side management critical value based on the peak demand of the power consumption side of the partition, and the daily load characteristic time distribution of the partition is obtained. Similar to step 34, use the load hour constraint to adjust and update the daily distribution of monthly load characteristics

[0045] Step 55: Use the predicted future partition maximum load Multiply by the monthly distribution of the annual load characteristics of the future partitions Daily distribution of monthly load characteristics and daily load characteristics distribution Get the future partition hourly load data Generate future zone hourly load curves.

[0046] Furthermore, the method of obtaining the future predicted partition load characteristics in step 52 adopts any one of the following two methods:

[0047] (1) If complete and abundant data on the current zoning load characteristics are available, the prediction of future zoning load characteristics includes:

[0048] Step 521: Obtain the future partition seasonal imbalance rate ρ′ 0 , based on the change in the seasonal imbalance coefficient between the future and current regions ρ′ 0 -ρ 0 ) Adjust the monthly distribution of the annual load characteristics of the current zone Get the monthly distribution of annual load characteristics of future partitions

[0049] Step 522: Based on the current status of the monthly average daily load rate of the partition Predict the average daily load rate of the future sub-region According to the change of the average daily load rate between the future and the current situation Adjust the current daily load characteristics of the zone and the monthly average hourly load Get the future partition daily load characteristics and hourly load monthly average Combined with the prediction of the changes in the peak daily power load of different regions in the future, make corrections;

[0050] Step 523: Calculate the monthly imbalance rate of the monthly load characteristics of the current partition As the initial value of the monthly imbalance rate of the future partition

[0051] (2) If the data on the current regional load characteristics is insufficient, the regional load characteristics are reasonably allocated on an annual, monthly and daily basis based on the load characteristics of the entire network, and the predicted values ​​of the future regional load characteristics are adjusted, including:

[0052] Step 521: Based on the predicted future partition load hours The total network load hours H′ fh The relative change of the change is divided into three equal parts and distributed to the seasonal imbalance coefficient ρ′ 0 , Annual average of monthly imbalance rate Annual average daily load factor Including the calculation of the relative change rate of load hours ε by taking the square root of the cube 0 , calculate the change of the seasonal imbalance coefficient of the partition Δρ′0 , Annual average change of monthly imbalance rate and the annual average daily load factor change

[0053] Step 522: Future partition seasonal imbalance coefficient ρ′ 0 =ρ′+Δρ′ 0 , according to the monthly distribution of the annual load characteristics of the entire network in the future RM' m , based on the change in the seasonal imbalance rate between the sub-region and the whole region Δρ' 0 , and adjust monthly to obtain the monthly distribution of annual load characteristics of future partitions

[0054] Step 523: The average daily load rate of the future partition in the next year Based on the average daily load rate of the entire network in the future Change in daily average load factor based on the annual average of the sub-region and the whole region Adjust monthly to get the average daily load rate of the future partition

[0055] Step 524: Calculate the monthly average hourly load based on the future daily load characteristics of the entire network Based on the change of the daily average load rate of the sub-area and the whole area Adjust monthly to obtain the daily load characteristics of the future partition and the monthly average hourly load

[0056] Step 525: Annual average monthly imbalance rate of future partitions According to the monthly imbalance rate σ' of the entire network in the future m , based on the change in the annual average of the monthly imbalance rate between the sub-region and the whole region Adjust monthly to get the monthly imbalance rate of future partitions

[0057] Furthermore, in step 21 and step 51, the annual average growth rate method, elasticity coefficient method or unit power consumption method is used to predict the future annual maximum load FM of the entire network. max Or the maximum annual load of the future partition Future annual power consumption of the entire network Or future annual electricity consumption of each region Future network load utilization hours or future zone load utilization hours

[0058] Beneficial Effects of the Invention

[0059] (1) The method of load characteristic statistics and hourly load curve generation described in the present invention includes: statistically analyzing the current load characteristics based on the current hourly load data, predicting the future power load of the entire network based on the current load characteristics, supplementing the future power load characteristics of the entire network for a whole year based on the prediction results of the future power load of the entire network and referring to the current load characteristics, and generating the future hourly load curve of the entire network to support the safe and stable calculation of the new power system, the balance simulation calculation and other related research.

[0060] (2) Based on the current hourly load data, the present invention has statistically analyzed the current annual load characteristics, monthly / weekly load characteristics, and daily load characteristics according to the monthly and weekly divisions. The system has completely extracted and statistically analyzed the characteristics of the current hourly load data, revealing the regular characteristics of the current power load. This provides important basic data for the prediction of future power load characteristics, and is also an important reference for completing the future annual load characteristics.

[0061] (3) The present invention provides four methods, namely, a gradual change method in the upper, middle and lower tenth of the month, a gradual change method on a weekly basis, a reference weekly characteristic method and a reference daily characteristic method, to complete the daily distribution of the monthly load characteristics of the whole year in the future. It provides three methods, namely, a monthly reference method, a weekly reference method and a daily reference method, to complete the temporal distribution of the daily load characteristics of the whole year in the future. Different completion methods are respectively suitable for the reference degree of different current load data, thereby improving the success rate of completing the daily distribution of the monthly load characteristics of the whole year in the future and the temporal distribution of the daily load characteristics of the whole year in the future. In addition, six random floating strategies of daily load characteristics, namely, no floating, overall floating, peak floating, valley floating, peak valley floating and random combination floating, are provided for the balance under different situations to reflect the volatility of the power load. A method for staggered processing of daily load characteristics considering demand-side management is also provided. Finally, the load characteristics are corrected by the constraint of the annual load hours, and finally a future hourly load curve that strictly meets the laws of maximum load, annual power consumption and load characteristic prediction values ​​is generated.

[0062] (4) Based on the method for generating the future full-grid load curve, the present invention also proposes a method for generating the future partition hourly load curve, which reasonably distributes the change of the partition load hours relative to the full-grid load hours to three parts: the annual load characteristic monthly distribution, the monthly load characteristic monthly distribution and the daily load characteristic monthly distribution. Similarly, the partition monthly load characteristic daily distribution is adjusted, supplemented and corrected, and the daily load characteristic hourly distribution throughout the year is adjusted, supplemented, randomly floated and demand-side management is performed to stagger the peak, and finally the future partition hourly load curve that meets the maximum load and annual electricity consumption forecast value is generated, laying a foundation for the simulation and calculation of new power system partitions. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 is a flow chart of the method of the present invention;

[0064] Figure 2 The figure is a detailed flow chart of the method of the present invention. DETAILED DESCRIPTION

[0065] like Figure 1 As shown, the method for generating load characteristic statistics and hourly load curves according to the present invention includes steps 1 to 5.

[0066] Step 1: Collect the current hourly load data of the power system, such as Figure 2 As shown, based on the current hourly load data of the power system, the current annual load characteristics of the entire network, the current monthly / weekly load characteristics of the entire network and the current daily load characteristics of the entire network are divided into monthly and weekly statistics.

[0067] (1) Current annual load characteristics statistics

[0068] When divided by month, the annual load characteristics need to count the monthly maximum loads of 12 months as the annual maximum load, and normalize them to obtain the monthly distribution RM of the annual load characteristics m , the average value of the monthly distribution of annual load characteristics is the seasonal imbalance coefficient ρ.

[0069]

[0070] In the formula, RM m Indicates the monthly distribution of annual load characteristics, m represents the month, FM m Indicates the monthly maximum load (10,000 kW), FM max Indicates the annual maximum load (10,000 kW), FD m,d represents the daily maximum load (10,000 kW), and d represents the day of the month (some months have less than 31 days, which are simplified to 31 days, the same below).

[0071] When divided by week, the annual load characteristics need to count the maximum loads of 53 weeks and normalize them to obtain the weekly distribution of the annual load characteristics.

[0072]

[0073] Where RW m Indicates the weekly distribution of annual load characteristics, w indicates the number of weeks, FW w Indicates the maximum weekly load (10,000 kW), FW max Indicates the annual maximum load (10,000 kW, with FM max equal), FD w,t It represents the maximum daily load (10,000 kW), and t represents the week number.

[0074] (2) Monthly load characteristics statistics

[0075] When divided by month, the monthly load characteristics need to count the daily maximum loads of 31 days each month and normalize them to obtain the daily distribution RD of the monthly load characteristicsm,d The average value of the monthly load characteristics is the monthly imbalance rate σ m .

[0076]

[0077] Where, RD m,d Indicates the daily distribution of monthly load characteristics, FH m,d,h Indicates the hourly load (10,000 kW) divided by month and day, where h represents the number of hours. m Indicates the monthly maximum load

[0078] (3) Weekly load characteristics statistics

[0079] When divided by week, the weekly load characteristics need to count the daily maximum load of 7 days each week and normalize it to get the daily distribution RD of the weekly load characteristics w,t At the same time, since the power load changes regularly in a week, the average value of 53 weeks can be calculated to obtain the averaged weekly load characteristics.

[0080]

[0081] Where, RD w,t Indicates the daily distribution of weekly load characteristics, FH w,t,h Indicates the hourly load divided by weekday (10,000 kW), It represents the daily distribution of averaged weekly load characteristics (not every Monday to Sunday has 53 daily loads in a year, it is just a simplified expression).

[0082] In addition, since the power load of a week is closely related to holidays and shift work, the power load on holidays is lower than that on weekdays, and the weekly load characteristics need to take into account the correction of holidays and shift work. Add number 8 to represent holidays on the basis of the 7-day numbering from Monday to Sunday, adjust the weekend numbers to weekday numbers according to the shift work arrangement, and recalculate the averaged daily distribution of weekly load characteristics.

[0083]

[0084] Where t' is the number of the week after considering holidays, Nw t' To take into account the number of weeks corresponding to each Sunday after holidays and shifting (can be less than 53), is the averaged daily distribution of weekly load characteristics after renumbering the days, RD w,t' The daily distribution of weekly load characteristics after renumbering the days of the week. The actual data statistical results show that the averaged weekly load characteristics show the daily distribution characteristics of Monday to Friday>Saturday and Sunday>holidays. This rule can be used as a reference for the generation of hourly load curves in the future.

[0085] (4) Daily load characteristics statistics

[0086] Regardless of whether it is divided by month or week, the maximum load of 24 hours must be counted daily and normalized to obtain the daily load characteristic time distribution RH m,d,h The average value of the daily load characteristics is the first status quo daily average load rate of the entire network γ m,d or γ w,t .

[0087] When divided by month, the hourly load average values ​​of different days in the month can be counted Maximum and minimum values, so as to count the monthly fluctuations of hourly load ΔRH m,h .

[0088]

[0089] In the formula, RH m,d,h It represents the distribution of daily load characteristics divided by month, It represents the monthly average of hourly loads on different days of the month (needs to be renormalized, with the maximum load control hour set to 1). Indicates the monthly average daily load rate, ΔRH m,h Indicates the upward fluctuation value of hourly load within a month. Indicates the downward fluctuation value of hourly load within a month. m,d Indicates the maximum daily load, FH m,d,h Indicates hourly load divided by month and day

[0090] Similarly, when divided by week, the average, maximum and minimum values ​​of the hourly load on different days of the week can be counted, thereby counting the ups and downs of the hourly load during the week.

[0091]

[0092] In the formula, RH w,t,h It represents the distribution of daily load characteristics divided by week, represents the weekly average of hourly loads on different days of the week (needs to be renormalized, with the maximum load control hour set to 1), γ w,t The average daily load rate of the entire network in the second situation is γ w,t , Indicates the weekly average daily load rate, ΔRH w,h Indicates the upward fluctuation value of hourly load within a week. Indicates the downward fluctuation value of hourly load within a week.

[0093] (5) Annual electricity consumption and load hours statistics

[0094] Generally, the monthly average load PM m Conduct statistics and then calculate the annual electricity consumption PMsum and load hours H fh .

[0095]

[0096] In the formula, PM m Indicates the monthly average load (10,000 kW), PM sum Indicates annual electricity consumption (10,000 kWh), H fh Indicates load hours (h).

[0097] Based on the above, the annual electricity consumption is actually the hourly load FH throughout the year. m,d,h The cumulative value of load hours is the annual electricity consumption divided by the annual maximum load, which actually normalizes and de-unitizes the hourly load throughout the year. Load hours are only related to the monthly distribution of annual load characteristics RM m , Monthly load characteristics daily distribution RD m,d and daily load characteristics when the distribution RH m,d,h This law is an important constraint for the generation of future hourly load curves.

[0098] Step 2: Forecast the future power load of the entire network based on the current annual load characteristics of the entire network, the monthly / weekly load characteristics of the entire network, and the daily load characteristics of the entire network

[0099] The power load forecast of the whole network is as follows: Figure 2 As shown, it mainly includes two parts: one is the maximum load FM′ of the whole network. max and annual electricity consumption PM' sum The second is the prediction of the load characteristics of the entire network, including the monthly distribution of annual load characteristics RM' m , Monthly average of daily load characteristics distribution Monthly imbalance rate σ' m , homogenized daily distribution of weekly load characteristics wait.

[0100] (1) Forecast of the maximum load of the entire network and annual power load

[0101] Before forecasting, we first determine the future design level year. Based on the construction period and commissioning time of the research object, we consider the level year in which it can fully play its role in the power system and determine it as the design level year for power system analysis and calculation, which is generally consistent with the country's five-year plan.

[0102] For the maximum load FM' in the future design level year max The prediction of power consumption can generally be based on the maximum load and power consumption in the current year, and the annual average growth rate method can be used to predict the future horizontal year. The elasticity coefficient method, unit power consumption method, etc. can also be used for prediction. Taking the annual average growth rate method as an example, the calculation formula is as follows:

[0103]

[0104] Where FM' max is the predicted maximum load of the entire network in the future horizontal year (10,000 kW), R f is the predicted annual growth rate of the maximum load of the entire network, PM' sum is the predicted annual power generation of the entire grid in the future horizontal year (10,000 kWh), R p is the predicted annual average growth rate of the annual power load of the entire network, Δn represents the difference between the current level year and the future design level year, H' fh It represents the load hours (h) in the future horizontal year.

[0105] (2) Forecast of annual load characteristics of the entire network

[0106] The annual load characteristic prediction is mainly based on the change of the annual average seasonal imbalance rate and is adjusted based on the monthly distribution of the annual load characteristics in the current level year. The following is a specific method for predicting the monthly distribution of the annual load characteristics.

[0107] First, we determine whether the changing trend of load characteristics in future years is increasing or decreasing, and predict the future seasonal imbalance rate ρ' based on the current seasonal imbalance rate ρ.

[0108] Second, according to the current annual load characteristics, the monthly distribution RM m Based on the seasonal imbalance rate change (ρ'-ρ), the monthly load characteristic monthly distribution RM' of the future horizontal year is obtained by adjusting it month by month m , the specific calculation formula is as follows:

[0109]

[0110] The above formula calculation results can satisfy and ensures the maximum load control month (RM m = January) will remain unchanged, and reasonable increase or decrease will be made in other months.

[0111] (3) Forecast of monthly average daily load characteristics of the entire network

[0112] The time distribution of daily load characteristics needs to predict the results of 365 days throughout the year. However, due to the hourly fluctuations, daily fluctuations, and monthly changes of electricity load, it is not suitable to predict directly on a daily basis. The general simplified approach is to predict the monthly average value of the time distribution of daily load characteristics. First, the regular characteristics of daily load characteristics in different months are clarified. The time distribution of daily load characteristics on different dates in the month can be supplemented by some calculation methods (see step 3 for details). The following is a specific method for predicting the monthly average value of the time distribution of daily load characteristics on a monthly basis.

[0113] First, we will determine whether the trend of the daily load characteristics in the future will increase or decrease, based on the current monthly average daily load rate. Predict the average daily load rate in the future month

[0114] Second, according to the current daily load characteristics, the monthly average hourly load (normalized), based on the change in daily average load factor Adjust monthly to get the daily load characteristics of the future horizontal year and the monthly average hourly load The specific calculation formula is as follows:

[0115]

[0116] The above formula calculation results can satisfy and ensures the maximum load control hours (i.e. The remaining hours will be adjusted accordingly with reasonable increase or decrease.

[0117] Third, combined with the forecast of the changes in the peak value of the daily power load in the future, the monthly average value of the hourly load of the daily load characteristics is calculated. For example, further considering the increase in residential cooling load in summer, slightly increase the hourly load at noon.

[0118] (4) Forecast of monthly imbalance rate of monthly load characteristics of the entire network

[0119] The daily distribution of monthly load characteristics requires monthly prediction of the results of the entire month for 31 days. However, due to the daily fluctuation characteristics of the electricity load, it is rather cumbersome to directly predict day by day. The general simplified approach is to predict the monthly average value of the monthly load characteristics (i.e., the monthly imbalance rate). First, the regular characteristics of the monthly load characteristics in different months are clarified, and the daily distribution within the month can be supplemented by some calculation methods (see step 3 for details). The following is a specific method for predicting the monthly imbalance rate of the monthly load characteristics.

[0120] First, the monthly imbalance rate σ of the current monthly load characteristics m As the initial value, combined with the predicted monthly distribution of future annual load characteristics RM' m and monthly average daily load factor Assuming that the load on each day of the month remains unchanged, calculate the assumed future load utilization hours

[0121] Second, based on the assumed value of future load utilization hours and the predicted value H' fh The relative error of the monthly imbalance rate is corrected to obtain the future forecast value σ' m .

[0122]

[0123] Step 3: Based on the prediction results of the future grid-wide power load characteristics, refer to the current grid-wide load characteristics, and complete the future grid-wide load characteristics for the whole year.

[0124] like Figure 2 As shown, to generate the future hourly load curve of the entire network, the monthly distribution of annual load characteristics RM' obtained in step 2 is required. m , Daily load characteristics Hourly load monthly average Monthly load characteristics monthly imbalance rate σ' m On this basis, the future load characteristics are further estimated and supplemented. The main work of supplementing the future load characteristics of the entire network is the daily distribution RD' of the monthly load characteristics. m,d The second is the distribution of daily load characteristics on different days of the month RH' m,d,h To make up for the two deficiencies, both deficiencies must be made on a monthly basis and cover the entire year.

[0125] There are two main bases for completing future load characteristics: one is to assume that some future load characteristics are similar to the current load characteristics and use the current load characteristics as a reference; the other is to make judgments based on the laws of future load characteristics and predict and propose daily distribution characteristics. In special cases, the current load characteristic data may not be complete, or the laws of future loads may change significantly and the reference is insufficient. In this case, different methods need to be adopted according to the reference degree of the current load characteristic data.

[0126] (1) Completing the daily distribution of monthly load characteristics month by month

[0127] According to the reference degree of the current load data, four methods can be adopted, from most to least, namely, reference daily characteristics method, reference weekly characteristics method, weekly gradual change method, and upper, middle and lower ten-day gradual change method to complete the daily distribution of the monthly load characteristics of the entire network.

[0128] The first is the reference daily characteristic method. Refer to the current monthly load characteristic daily distribution RD m,d As the initial daily distribution of future predictions, and then based on the future predicted and initialized monthly imbalance rate changes (σ' m -σ m ), and adjust daily to obtain the daily distribution of monthly load characteristics in the future horizontal year RD' m,d This method is based on the assumption that the distribution of load characteristics in the future days will remain consistent with the current situation, which is relatively direct and simple.

[0129]

[0130] The above formula calculation results can satisfy and ensures the maximum load control date (RD m,d=1 moment) remains unchanged, and reasonable increase or decrease fine-tuning is uniformly carried out in the remaining hours.

[0131] The second is the reference weekly characteristics method. First, refer to the weekly distribution of the current annual load characteristics RW w , and the daily distribution of average weekly load characteristics after considering holidays and shift breaks Calculate the load characteristics for each day of the week on a weekly basis Convert to initial value of daily distribution of monthly load characteristics

[0132]

[0133] Secondly, find the maximum value within the month and normalize it to get Statistical average to get the intermediate result monthly imbalance rate

[0134]

[0135] Reference daily characteristic method, based on monthly imbalance rate change Adjust the normalized initial value of the daily distribution of monthly load characteristics on a daily basis Get the future monthly load characteristic daily distribution RD' m,d .

[0136] This method assumes that the weekly distribution of load characteristics in the future years will remain unchanged, and uses the averaged daily distribution of weekly load characteristics to represent the daily load changes of all days in the week. It is an initialization alternative method in the absence of daily distribution of monthly load characteristics.

[0137] The third is the weekly gradual change method. This method is similar to the reference week characteristic method. The difference is that the first step does not have the reference week division of the current annual load characteristic weekly distribution RW w , but based on the predicted monthly distribution of future annual load characteristics RM' m , the annual load characteristic weekly distribution is initialized by interpolation method Initial value of weekly distribution based on annual load characteristics Calculate the daily distribution of weekly load characteristics And converted into the initial value of the daily distribution of monthly load characteristics Normalized to get calculate The average value of the monthly imbalance rate Based on the monthly imbalance rate change Adjust the normalized initial value of the daily distribution of monthly load characteristics on a daily basis Get the future monthly load characteristic daily distribution RD′ m,d ;

[0138] Weekly distribution of annual load characteristics The interpolation method first determines the control week within the month If one month involves 5 weeks It is assumed that the third week is the control week of the month; if a month involves 6 weeks According to the monthly maximum load (RM' m and RM' m+1 ) is determined by the increase or decrease. If the load decreases next month, it is assumed to be the 3rd week. If the load increases next month, it is assumed to be the 4th week.

[0139]

[0140] Secondly, determine the load characteristics of each week within the month The load characteristic of the control week is equal to the predicted value of the monthly load characteristic RM' m , between the control week of this month and the control week of last month and next month ( and Corresponding to RM' m-1 and RM' m+1 ) to obtain the load characteristics of other weeks except the control week. The load characteristics of all weeks show a gradual change.

[0141]

[0142] The fourth is the gradual change method of the first, middle and last week. This method no longer directly refers to or interpolates the annual load characteristics RW divided by week. w , but the interpolation initialization of the daily distribution of the monthly load characteristics is divided into ten days. The interpolation method is similar to the weekly gradual change method, but it does not interpolate with a week as the step size, but distinguishes between the upper, middle and lower ten days for interpolation, which is relatively simple. Get the initial value of the interpolation of the daily distribution of the monthly load characteristics After that, it is converted into the normalized initial value of the daily distribution of monthly load characteristics calculate The average value of the monthly imbalance rate Based on the monthly imbalance rate change Adjust the normalized initial value of the daily distribution of monthly load characteristics on a daily basis Get the future monthly load characteristic daily distribution RD' m,d .

[0143] Daily distribution of monthly load characteristics The interpolation method first determines the gradual change law of the load characteristics in the middle, first and last ten days. The load characteristics of the middle month are 1, and the load characteristics of the first ten days are RM' m and last month's load characteristic RM' m-1 The daily linear interpolation, the second half of the month is the load characteristics of this month and the load characteristics of the next month RM' m+1 The daily linear interpolation of .

[0144] Secondly, the daily distribution of average weekly load characteristics is superimposed every day within the month. (t' represents the weekday number after taking into account holidays and adjusted holidays) to reflect the changes and fluctuations within the week.

[0145]

[0146] The above four filling methods refer to the law of current load characteristics to a certain extent, and are also based on the predicted future load characteristics. The remaining unknown information is formed into a trunk through linear interpolation, and the intra-week volatility is filled with the daily distribution of the average weekly load characteristics. You can choose a reasonable method according to the situation during use.

[0147] (2) Completing the time distribution of daily load characteristics on different days of the month on a daily basis

[0148] According to the reference degree of the current load data, three methods, namely daily reference method, weekly reference method and monthly reference method, can be adopted from most to least to complete the distribution of daily load characteristics on different dates of the month for the entire network.

[0149] The first is the daily reference method. This method unifies and adjusts the control hours based on the assumption that the load characteristics distribution in the future days is similar to the current situation. It is relatively direct and simple.

[0150] First, refer to the current daily load characteristics of each day of the month. m,d,h , as the initial daily time distribution for future predictions; secondly, unify the maximum load control hours. Since the current daily load characteristic time distribution may have inconsistent maximum load control hours on different days, it is necessary to check the current daily load characteristic time distribution RH on a daily basis according to the principle of the same maximum load control hours per day within the month. m,d,h , and the predicted monthly average hourly load of the entire network in the future Control the dates with different hours, exchange and sequence the peak hour load, and obtain the time distribution of the daily load characteristics after sequencing And calculate the monthly average value to get the mean value of the daily load characteristic distribution after sequencing The hourly load corresponding to the control hours of all dates after the adjustment is equal to 1; again based on the change in the future forecast value and the adjusted hourly load monthly average Adjust hourly to get the daily load characteristic time distribution RH' of each day in the future horizontal year m,d,h .

[0151]

[0152] The above formula calculation results can satisfy and ensures the maximum load control date (i.e. The remaining hours will be adjusted accordingly with reasonable increase or decrease.

[0153] The second is the weekly reference method. This method refers to the weekly average of the current hourly load divided by week. Initial daily load characteristics for a week as future forecast It is an alternative initialization method when there is no daily load characteristic distribution. However, since the load distribution is the same every day of the week, the fluctuation characteristics of the power load are not reflected. An optional approach is to increase the random fluctuation of the load every day of the week.

[0154] For the random floating method of hourly load, a random fluctuation control parameter λ is first set, which represents several strategies including no floating, overall floating, peak floating, valley floating, peak-valley floating and random combined floating. λ=0 means no random floating, λ=1 means overall floating (increase or decrease in the other 23 hours except the control hour), λ=2 means floating during peak hours (increase or decrease in the 2nd to 9th hours except the control hour), λ=3 means floating during valley hours (increase or decrease in the 17th to 24th hours), λ=4 means floating during peak and valley hours (increase or decrease in the 2nd to 9th hours and the 17th to 24th hours), λ=5 means random combined floating, that is, a value is randomly selected from 1 to 4 as the parameter of this randomization, and overall floating, peak floating, valley floating and peak-valley floating all have probability of occurring.

[0155] Secondly, the load characteristics of 24 hours are sorted from large to small (hours after sorting are represented by h'), and the hours that need to float are grouped into a set Hset according to different floating strategies. When λ=1, Hset=[2,24], when λ=2, Hset=[2,9], when λ=3, Hset=[17,24], and when λ=4, Hset=[2,9]∪[17,24]. Therefore, when h'∈Hset, hourly floating is required.

[0156] The next step is to calculate the hourly load fluctuation value δ(h) and update the load distribution of each day of the week. Take a random number between -1 and 1, rand(-1,1). If it is less than 0, it means the hourly floating reduction. The reduction amount is the random number multiplied by the downward fluctuation value of the hourly load within the week. If it is greater than 0, it means the hourly floating adjustment is increased. The adjustment amount is the random number multiplied by the upward fluctuation value of the hourly load within the week ΔRH w,h .

[0157]

[0158] Finally, the maximum load control hours are unified (similar to the daily reference method). Since the weekly average of the current hourly load is divided by week Monthly average forecast hourly load divided by month If the maximum load control hours are not uniform, it is necessary to check the predicted monthly average of the future network hourly load on a daily basis according to the principle that the maximum load control hours are the same every day within the month. Whether the control hours are the same, the peak hour loads of different dates are interchanged and reordered. After the reordering, the hourly loads corresponding to all control hours are equal to 1.

[0159] Get the load distribution of each day after considering hourly fluctuation and adjusting the sequence And calculate the monthly average hourly load The change in the monthly average of hourly load based on future forecast values ​​and after sequencing Adjust hourly to get the daily load characteristic time distribution RH' of each day in the future horizontal year m,d,h , the calculation formula is consistent with formula 3-7.

[0160] The third is the monthly reference method. This method is based on the monthly average of the future forecast hourly load of the entire network. Distribution of the initial time of January as a future forecast It is also an alternative initialization method when there is no daily and weekly load characteristic time distribution. And because the load time distribution is the same every day in a month, it is more difficult to reflect the fluctuation characteristics of the power load than the weekly reference method. It is recommended to increase the random fluctuation of the load every day in the month.

[0161] The method for random fluctuation of hourly load within a month is similar to the weekly reference method. A random fluctuation control parameter λ is also set. The load characteristics of 24 hours are sorted from large to small (represented by h'). The hours to be fluctuated are grouped into a set Hset. When h'∈Hset, hourly fluctuation is required. The hourly load fluctuation value δ(h) is calculated and the load time distribution of each day in the month is updated. The difference from the weekly reference method is that the monthly average hourly load is used in formula 3-8. Upward fluctuation value of hourly load within a month ΔRH m,h and downward fluctuation value Substitute for weekly average hourly load Upward fluctuation value of hourly load during the week ΔRH w,h and downward fluctuation value In addition, the monthly reference method does not require the unification of the maximum load control hours because they are already unified.

[0162] Similarly, the load distribution of each day considering hourly fluctuation is obtained Then, use formula 3-7 to adjust hourly to get the daily load characteristic time distribution RH' of each day in the month of the future horizontal year m,d,h .

[0163] (3) Demand-side management peak shifting

[0164] When forecasting future loads, it may be necessary to set demand-side management measures based on the needs of power system balance and stability, that is, when the load reaches a certain percentage of the annual maximum load (expressed by the peak shift rate η), orderly power consumption control measures are taken to distribute the power load to non-peak hours to control the maximum load from growing. At this time, it is necessary to adjust some load characteristics that exceed the management threshold, mainly the daily load characteristic time distribution RH' of the corresponding date. m,d,h Make adjustments.

[0165] When RM' m RD' m,d >1-η, demand-side management peak shifting is required on that day, and the maximum value of the hourly distribution of daily load characteristics needs to be adjusted from 1 to no more than Calculate the total value of the reduction Δinc when the load is large, and the total value of the increase Δdec when the remaining load is small, and calculate it hour by hour. The proportion of the reduction is finally obtained, and the daily load characteristic time distribution after peak shifting is obtained.

[0166]

[0167] (4) Adjust the daily distribution of monthly load characteristics according to the number of load hours

[0168] The future monthly imbalance rate σ' predicted in step 2 m It is assumed that the load on each day of the month remains unchanged and is corrected according to the load hour constraint. After completing the daily distribution of the monthly load characteristics and the daily distribution of the load characteristics in step 3, the monthly load characteristics and daily load characteristics on different dates in the month have fluctuated differently, which is inconsistent with the assumption that the load on each day of the month remains unchanged. It is necessary to further adjust and update the daily distribution of the monthly load characteristics RD' according to the load hour constraint. m,d .

[0169] First, according to the above-mentioned monthly distribution of future annual load characteristics RM' m , Daily distribution of monthly load characteristics RD' m,d and daily load characteristics distribution The cumulative sum is used to obtain the current future load utilization hours.

[0170] Second, based on the current value of future load utilization hours The predicted value H' in step 2 fh The relative error of the monthly imbalance rate is corrected to obtain the final future forecast value

[0171]

[0172] The third is based on the relative change in the monthly imbalance rate. Further adjust the monthly load characteristic daily distribution RD' m,d , and get the final future prediction value

[0173]

[0174] The above formula calculation results can satisfy and ensures the maximum load control date (RD m,d =1 moment) remains unchanged, and reasonable increase or decrease fine-tuning is performed uniformly in the remaining hours. The final calculation result satisfies the load utilization hours constraint

[0175] Step 4: Generate the future hourly load curve of the entire network based on the future maximum load and annual power consumption forecast results of the entire network and the future load characteristics of the entire network after the full year is completed.

[0176] Using the predicted future maximum network load FM' max , multiplied by the monthly distribution of the future annual load characteristics of the entire network RM' m , Daily distribution of monthly load characteristics and daily load characteristics distribution You can get the future hourly load data of the entire network FH' m,d,h , thereby generating the future hourly load curve of the entire network.

[0177]

[0178] Step 5: Generate the hourly load curve of the future partition by referring to the method for generating the hourly load curve of the future whole network

[0179] (1) Forecast of maximum load and annual electricity consumption in different regions

[0180] The prediction is mainly based on two factors: one is time correlation, starting from the current maximum load and annual power consumption of the sub-region, using the average annual growth rate method, elasticity coefficient method, unit power consumption method, etc. to make sub-regional predictions for the future horizontal year; the other is spatial correlation, based on the future maximum load predicted by the entire network. Based on the annual electricity consumption, regional forecasting is carried out in combination with the regional electricity consumption characteristics and its correlation with the overall network electricity load.

[0181]

[0182] In the formula, Indicates the future maximum load of the partition (10,000 kW), Indicates the current maximum load of the partition (10,000 kW). Indicates load utilization hours, It represents the predicted value of the average annual growth rate of the maximum load of the partition. ω1 and ω2 are the maximum load prediction coefficients, which respectively represent the time correlation of the maximum load of the partition based on the current growth rate prediction and the spatial correlation of the prediction based on the characteristics of the partition of the whole network. Indicates the future annual electricity consumption of the zone (10,000 kWh), Indicates the current annual electricity consumption of the partition (10,000 kWh), It represents the predicted value of the average annual growth rate of the maximum load of the partition. ω3 and ω4 are the annual electricity consumption prediction coefficients, which respectively represent the time correlation of the annual electricity consumption of the partition based on the current growth rate prediction and the spatial correlation of the prediction based on the characteristics of the partition of the whole network.

[0183] (2) Obtaining load characteristics of different regions

[0184] Obtaining load characteristics of each zone, including the monthly distribution of annual load characteristics of each zone in the future The average daily load factor of the future zone in the month Future zone daily load characteristics Hourly load monthly average Initial value of monthly imbalance rate of monthly load characteristics of future zones One of the following two methods can be used according to the richness of the current load characteristic data of the partition.

[0185] First, when there is abundant data on the current regional load characteristics, the future regional load characteristics prediction is carried out by referring to the method of predicting the future network load characteristics, including:

[0186] Similar to formula 2-2, obtain the future partition seasonal imbalance rate ρ' 0 , based on the change in the seasonal imbalance coefficient between the future and current regions (ρ' 0 -ρ 0 ) Adjust the monthly distribution of the annual load characteristics of the current zone Get the monthly distribution of annual load characteristics of future partitions

[0187] Similar to formula 2-3, based on the current monthly average daily load rate of the partition Predict the average daily load rate of the future sub-region According to the change of the average daily load rate between the future and the current situation Adjust the current daily load characteristics of the zone and the monthly average hourly load Get the future partition daily load characteristics and hourly load monthly average Combined with the prediction of the changes in the peak daily power load of different regions in the future, make corrections;

[0188] Monthly imbalance rate of monthly load characteristics based on the current situation As the initial value of the monthly imbalance rate of the future partition

[0189] Second, when the data on the current load characteristics of the sub-region is insufficient (due to the complex relationship between the sub-region and the entire grid, the difficulty in unifying the statistical caliber, the difficulty in obtaining detailed data, etc., it is difficult to directly predict), such as Figure 2 As shown in the figure, the load characteristics of each partition are reasonably allocated on a yearly, monthly and daily basis according to the load characteristics of the entire network, and the predicted values ​​of the load characteristics of each partition in the future are adjusted, including:

[0190] First, the distribution of load hour changes in different regions

[0191] Based on the predicted future zone load hours The total network load hours H' fh The relative change of the seasonal imbalance coefficient ρ' is divided into three equal parts. 0 , Annual average of monthly imbalance rate Annual average daily load factor

[0192] Calculate the relative change rate of load hours in three equal parts ε 0 The annual load characteristics, monthly load characteristics, daily load characteristics and load utilization hours are in a multiplication relationship, so the relative change rate is calculated by taking the square root of the cube.

[0193] Calculate the change of the seasonal imbalance coefficient of the partition Δρ' 0 , Annual average change of monthly imbalance rate Annual average daily load factor change

[0194]

[0195] Secondly, adjust the monthly distribution of the annual load characteristics of the future sub-region

[0196] Future partition-season imbalance coefficient ρ' 0 =ρ'+Δρ' 0 , according to the monthly distribution of the annual load characteristics of the entire network in the future RM' m , based on the change in the seasonal imbalance rate between the sub-region and the whole region Δρ' 0 , and adjust monthly to obtain the monthly distribution of annual load characteristics of future partitions

[0197]

[0198] Again, the adjustment is made to obtain the monthly average distribution of the future partition daily load characteristics

[0199] The average daily load factor of the future district in the year Based on the average daily load rate of the entire network in the future Change in daily average load factor based on the annual average of the sub-region and the whole region Adjust monthly to get the average daily load rate of the future partition

[0200]

[0201] According to the daily load characteristics of the entire network in the future, the monthly average hourly load Based on the change of the daily average load rate of the sub-area and the whole area Adjust monthly to obtain the daily load characteristics of the future partition and the monthly average hourly load

[0202]

[0203] Finally, the initial value of the monthly imbalance rate of the future partition monthly load characteristics is obtained by adjustment.

[0204] Annual average monthly imbalance rate of future regions According to the monthly imbalance rate σ′ of the entire network in the future m , based on the change in the annual average of the monthly imbalance rate between the sub-region and the whole region Adjust monthly to get the initial value of the monthly imbalance rate of the future partition

[0205]

[0206] (3) Correction based on the load utilization hours of the partition

[0207] Whether it is directly referring to the method of predicting the load characteristics of the entire network (changing during prediction), or using the method of allocating the change in load hours by region (relative change rate ε 0 After the square root of the cube is calculated), the monthly distribution of annual load characteristics Monthly average daily load factor Monthly imbalance rate After the allocation process within 12 months, the annual utilization hours of the partition load have a certain error, and the monthly imbalance rate needs to be adjusted. Correction is performed to obtain the monthly imbalance rate that satisfies the load hour constraint

[0208] First, the monthly imbalance rate of the region is calculated As the initial value, combined with the monthly distribution of the annual load characteristics of the partition and the average daily load factor of the zone for the month Assuming that the load on each day of the month remains unchanged, calculate the assumed load utilization hours of the partition

[0209] Second, based on the assumed value of the utilization hours of the partition load With the predicted value The relative error of the monthly imbalance rate is corrected to obtain the partition prediction value

[0210]

[0211] (4) Similar to step 3, perform future partition load characteristics supplementation, demand side management and load hour correction

[0212] Similar to the need to complete the load characteristics of the entire network throughout the year, the future regional load characteristics will also need to be completed throughout the year. The four methods of reference daily characteristics method, reference weekly characteristics method, weekly gradual change method, and upper, middle and lower ten-day gradual change method can be similarly selected to complete the daily distribution of regional monthly load characteristics. Similarly, the three methods of daily reference method, weekly reference method, and monthly reference method can be used to complete the time distribution of daily load characteristics on different dates within the regional month. Similarly, the time distribution of daily load characteristics is adjusted for some dates that exceed the demand-side management critical value. Similarly, the load hour constraint is used to adjust and update the daily distribution of monthly load characteristics. The corresponding calculation formulas are relatively similar and will not be repeated here.

[0213] (5) Generate future partition hourly load characteristic curve

[0214] Utilize predicted future partition maximum loads Multiply by the monthly distribution of the annual load characteristics of the future partitions Daily distribution of monthly load characteristics and daily load characteristics distribution You can get the future partition hourly load data Thus, the future partition hourly load curve is generated.

[0215]

Claims

1. A method for load characteristic statistics and hourly load curve generation, characterized in that: include: Step 1: Collect the current hourly load data of the entire power system, and divide the current hourly load data into monthly and weekly data to calculate the current annual load characteristics of the entire power system, the current monthly / weekly load characteristics of the entire power system, and the current daily load characteristics of the entire power system; Step 2: Based on the current annual load characteristics, monthly / weekly load characteristics and daily load characteristics of the entire network, forecast the future power load of the entire network; Step 3: Based on the prediction results of the future grid-wide power load characteristics and referring to the current grid-wide load characteristics, complete the future grid-wide load characteristics for the whole year; Step 4: Generate the future hourly load curve of the entire network based on the forecast results of the future maximum load and annual power consumption of the entire network and the future load characteristics of the entire network after the full year is supplemented.

2. The method for generating load characteristic statistics and hourly load curve according to claim 1, characterized in that: Step 1 specifically includes: Step 11, collecting hourly load data of the current status of the power system; Step 12: Calculate the current annual maximum load FM of the entire grid based on the collected hourly load data of the power system max , divided by month, the current statistical status of the annual load characteristics of the entire network, monthly distribution RM m , obtain the seasonal imbalance coefficient ρ, divide the statistics by week, and calculate the weekly distribution of the annual load characteristics of the entire network RW w ; Step 13: Statistical status of the daily distribution of the monthly load characteristics of the entire network RD m,d , get the monthly imbalance rate σ m ; Step 14: Statistical status of the daily distribution of weekly load characteristics of the entire network RD w,t , and calculate the daily distribution of the weekly load characteristics of the entire network Considering holidays and shift adjustments, renumber the days of the week from 1 to 8 and calculate the equalization status after renumbering. Daily distribution of weekly load characteristics of the entire network Step 15: Divide the current status of the daily load characteristics of the entire network by month and calculate the distribution RH m,d,h , get the first status quo daily average load rate of the entire network γ m,d , calculate the current daily load characteristics of the entire network and the monthly average hourly load Current monthly average daily load rate of the entire network And the upward fluctuation value of hourly load within the month ΔRH m,h and downward fluctuation value Step 16: Statistical analysis of the current status of the daily load characteristics of the entire network by week: RH distribution w,t,h , and obtain the second status quo daily average load rate of the entire network γ w,t , calculate the current weekly average hourly load of the entire network Current weekly average daily load rate of the entire network And the upward fluctuation value of hourly load during the week ΔRH w,h and downward fluctuation value Step 17: Statistical status by month: Monthly average load PM of the entire network m , the cumulative current hourly load of the entire network throughout the year FH m,d,h Get the current annual power consumption PM of the entire network sum , calculate the current network load hours H fh , analyze the current status of the total network load hours H fh Monthly distribution of annual load characteristics of the entire network RM m , Current status of the monthly load characteristics of the entire network Daily distribution RD m,d The current status of the daily load characteristics of the entire network divided by month is RH m,d,h The current status of the daily load characteristics of the entire network divided by week RH w,t,h The law of.

3. The method for generating load characteristic statistics and hourly load curve according to claim 2, characterized in that: Step 2 includes: Step 21: Based on the current annual maximum network load FM max The current annual power consumption of the entire network PM sum Predict the maximum annual load FM′ of the entire network in the future max 、Future annual power consumption of the entire network PM′ sum and the future annual load utilization hours H′ of the entire network fh ; Step 22: Obtain the future network-wide seasonal imbalance rate ρ′, and adjust the monthly distribution of the current network-wide annual load characteristics RM based on the change in the seasonal imbalance coefficient (ρ′-ρ) on a monthly basis m , and obtain the monthly distribution of the annual load characteristics of the entire network in the future RM′ m ; Step 23: Based on the current status of the monthly average daily load rate of the entire network Predict the average daily load rate of the entire network in the future According to the change of daily average load rate Adjustment status by month Daily load characteristics of the entire network Monthly average hourly load Get the future network daily load characteristics and hourly load monthly average Combined with the forecast of the changes in the peak daily load of the entire network in the future, the monthly average hourly load of the entire network in the future is calculated. make reasonable amendments; Step 24: Assuming that the load on each day of the month remains unchanged, the monthly imbalance rate σ of the current monthly load characteristics of the entire network m As the initial value, combined with the monthly distribution of the annual load characteristics of the entire network in the future, RM′ m and the average daily load factor for the future month Calculate the first future load utilization hours Based on the assumed first future load utilization hours And the predicted future annual load utilization hours H′ fh The relative error of the monthly load characteristics is corrected to obtain the monthly imbalance rate σ m Get the monthly unbalanced rate σ′ of the future monthly load characteristics of the entire network m .

4. The method for generating load characteristic statistics and hourly load curve according to claim 3, characterized in that: Step 3 includes: Step 31: Complete the monthly load characteristic daily distribution RD′ of the entire network in the future month by month m,d ; Step 32: Completing the daily load characteristic distribution RH′ of the entire network in the future day by day m,d,h ; Step 33: Perform demand-side management peak shifting. If the peak shifting rate is η, then some RM′ m RD′ m,d >1-η, and conduct peak shifting for demand-side management, and adjust the distribution of the future daily load characteristics of the entire network to RH′ m,d,h Adjust the maximum value to no more than Calculate the total value of the reduction Δinc in the hours with large load, calculate the total value of the increase Δdec in the hours with small remaining load, and calculate the daily load characteristic time distribution RH′ of the date to be adjusted m,d,h Hour by hour The proportion of the reduction is obtained to obtain the distribution of the daily load characteristics of the entire network after peak shifting. Step 34: Adjust the daily distribution RD′ of the future monthly load characteristics of the entire network m,d :Based on the monthly distribution of the annual load characteristics of the entire network in the future RM′ m , Monthly load characteristics daily distribution RD′ m,d and daily load characteristics distribution Calculate the second future load utilization hours Based on the second future load utilization hours and the predicted future load utilization hours H′ fh The relative error of the monthly load characteristic imbalance rate σ′ in the future m Get the final future monthly load characteristic monthly imbalance rate Based on the relative change in the monthly imbalance rate before and after the correction Adjust the daily distribution RD′ of the load characteristics of the future month on a daily basis m,d , and obtain the predicted value of the daily distribution of the final monthly load characteristics 5. The method for generating load characteristic statistics and hourly load curve according to claim 4, characterized in that: In step 31, the daily distribution RD′ of the future monthly load characteristics of the entire network is supplemented month by month m,d The method is any one of the following 4: (1) Changes in monthly imbalance rate based on current and future conditions (σ′) m -σ m ) for the current daily distribution of the monthly load characteristics of the entire network RD m,d Adjust daily to get the daily distribution of the future monthly load characteristics of the entire network RD' m,d ; (2) Based on the current weekly distribution of the annual load characteristics of the entire network RW w and daily distribution of average weekly load characteristics Calculate the daily distribution of load characteristics for each week And converted into the initial value of the daily distribution of monthly load characteristics Find the monthly maximum value for the initial value of the daily distribution of monthly load characteristics Normalize to get calculate The average value of the monthly imbalance rate Based on the monthly imbalance rate change Adjust the normalized initial value of the daily distribution of monthly load characteristics on a daily basis Get the future monthly load characteristic daily distribution RD' m,d ; (3) Based on the predicted monthly distribution of future annual load characteristics RM' m , determine the control week within the month The load characteristic of the control week is equal to the predicted value of the monthly load characteristic RM' m , and the load characteristics of other weeks except the control week are obtained by interpolation, so as to obtain the initial value of the annual load characteristic weekly distribution Initial value of weekly distribution based on annual load characteristics Calculate the daily distribution of weekly load characteristics And converted into the initial value of the daily distribution of monthly load characteristics Normalized to get calculate The average value of the monthly imbalance rate Based on the monthly imbalance rate change Adjust the normalized initial value of the daily distribution of monthly load characteristics on a daily basis Get the future monthly load characteristic daily distribution RD' m,d Wherein, the control week is determined as follows: if a month involves 5 weeks, the third week is the control week of the corresponding month; if a month involves 6 weeks, if the maximum load RM of the next month is m+1 Less than the maximum monthly load RM' m , then the third week is the control week, otherwise, the fourth week is the control week; (4) Based on the predicted monthly distribution of future annual load characteristics RM' m , determine the load characteristics of the middle, first and last ten days through interpolation, and superimpose the daily distribution of the average weekly load characteristics Get the initial value of the daily distribution of monthly load characteristics Converted to normalized initial value of daily distribution of monthly load characteristics calculate The average value of the monthly imbalance rate Based on the monthly imbalance rate change Adjust the normalized initial value of the daily distribution of monthly load characteristics on a daily basis Get the future monthly load characteristic daily distribution RD' m,d .

6. The method for generating load characteristic statistics and hourly load curve according to claim 4, characterized in that: In step 32, the daily load characteristic time distribution RH' of the entire grid in the future horizontal year is supplemented day by day m,d,h The method uses any of the following 3 methods: (1) According to the principle of the same maximum load control hours per day within a month, check the current daily load characteristics and distribution RH every day m,d,h , and the predicted monthly average hourly load of the entire network in the future Control the dates with different hours, exchange and sequence the peak hour load, and obtain the time distribution of the daily load characteristics after sequencing And calculate the monthly average value to get the mean value of the daily load characteristic distribution after sequencing The change in the monthly average of future hourly load and the monthly average of current hourly load after adjustment The time distribution of daily load characteristics after sequence adjustment Adjust hourly to obtain the daily load characteristic time distribution RH' of each day in the future month m,d,h ; (2) Refer to the weekly average of the current hourly load divided by week As the initial daily load characteristic time distribution of the week for future prediction, the random fluctuation of the load is increased every day of the week, and a random fluctuation control parameter λ is set, which represents no fluctuation, overall fluctuation, peak fluctuation, valley fluctuation, peak valley fluctuation or random combination floating strategy. The load characteristics of 24 hours are sorted from large to small as h', and the hours to be fluctuated are grouped into a set Hset according to different floating strategies. When h'∈Hset, hourly fluctuation is required, and a random number rand(-1,1) between -1 and 1 is taken. According to the positive and negative values ​​of the random number and the downward fluctuation value of the hourly load in the week, the hourly fluctuation is calculated. and the upward fluctuation value ΔRH w,h , calculate the hourly load floating value δ(h), update the daily load characteristic time distribution of each day in the week, and perform peak hour load interchange and sequencing on the updated daily load time distribution of each day in the week according to the principle of the same daily maximum load control hour in the month, and obtain the daily load characteristic time distribution after sequencing And calculate the monthly average value to get the mean value of the daily load characteristic distribution after sequencing Changes based on the monthly average of hourly load The time distribution of daily load characteristics after sequence adjustment Adjust hourly to obtain the daily load characteristic time distribution RH' of each day in the future month m,d,h ; (3) Based on the monthly forecast of the average monthly hourly load of the entire network As the initial time distribution of January in the future forecast, the random fluctuation of load is increased every day in the month, and the random fluctuation control parameter λ is set to select the floating strategy and determine the hours to be floated. According to the positive and negative values ​​of the random number rand (-1,1) and the downward fluctuation value of the hourly load in the month and the upward fluctuation value ΔRH m,h , calculate the hourly load fluctuation value δ(h), and update the load characteristic distribution of each day in the month to obtain And calculate the monthly average value to get the updated hourly load monthly average value Changes based on the monthly average of hourly load Updated daily load characteristics and time distribution Adjust hourly to obtain the daily load characteristic time distribution RH' of each day in the future month m,d,h .

7. A method for generating load characteristic statistics and hourly load curves according to any one of claims 4 to 6, characterized in that: Step 4 is as follows: the predicted future maximum network load FM' max , multiplied by the monthly distribution of the future annual load characteristics of the entire network RM′ m , Modified monthly load characteristics daily distribution and daily load characteristics distribution Get the future hourly load data of the entire network FH′ m,d,h , based on the future hourly load data of the entire network FH′ m,d,h Generate the future hourly load curve of the entire network.

8. The method for generating load characteristic statistics and hourly load curve according to claim 1, characterized in that: The step 5 is also included: generating a future zone hourly load curve, which specifically includes the following steps: Step 51: Based on the temporal correlation of the current future growth rate forecast and the spatial correlation of the network partition characteristics, the maximum load of the partition in the future design level year is predicted. Annual electricity consumption by region and zone load utilization hours Step 52: Obtain the future predicted partition load characteristics, including the monthly distribution of the future partition annual load characteristics. The average daily load factor of the future zone in the next month Future zone daily load characteristics Hourly load monthly average Initial value of monthly imbalance rate of monthly load characteristics of future zones Step 53, similar to step 24, combined with the monthly distribution of the annual load characteristics of the partition and the monthly average daily load factor of the zone Assuming that the load on each day of the month remains unchanged, calculate the assumed load utilization hours of the partition Assumed value based on load utilization hours of the zone With the predicted value The relative error of the initial value of the imbalance rate of the future partition Adjust to get the predicted value of monthly imbalance rate of monthly load characteristics of the partition Step 54: Perform the daily distribution of the monthly load characteristics of each zone on a monthly basis similar to step 31 Completion, similar to step 32, daily load characteristics time distribution Similar to step 33, the daily load characteristic time distribution is adjusted for some dates that exceed the demand side management critical value based on the peak demand of the power consumption side of the partition, and the daily load characteristic time distribution of the partition is obtained. Similar to step 34, use the load hour constraint to adjust and update the daily distribution of monthly load characteristics Step 55: Use the predicted future partition maximum load Multiply by the monthly distribution of the annual load characteristics of the future partitions Daily distribution of monthly load characteristics and daily load characteristics distribution Get the future partition hourly load data Generate future zone hourly load curves.

9. The method for generating load characteristic statistics and hourly load curve according to claim 8, characterized in that: Step 52 is to obtain the predicted partition load characteristics in the future by using any one of the following two methods: (1) If complete and abundant data on the current zoning load characteristics are available, the prediction of future zoning load characteristics includes: Step 521: Obtain the future partition seasonal imbalance rate ρ′ 0 , based on the change in the seasonal imbalance coefficient between the future and current regions (ρ′ 0 -ρ 0 ) Adjust the monthly distribution of the annual load characteristics of the current zone Get the monthly distribution of annual load characteristics of future partitions Step 522: Based on the current status of the monthly average daily load rate of the partition Predict the average daily load rate of the future sub-region According to the change of the average daily load rate between the future and the current situation Adjust the current daily load characteristics of the zone and the monthly average hourly load Get the future partition daily load characteristics and hourly load monthly average Combined with the prediction of the changes in the peak daily power load of different regions in the future, make corrections; Step 523: Calculate the monthly imbalance rate of the monthly load characteristics of the current partition As the initial value of the monthly imbalance rate of the future partition (2) If the data on the current regional load characteristics is insufficient, the regional load characteristics are reasonably allocated on an annual, monthly and daily basis based on the load characteristics of the entire network, and the predicted values ​​of the future regional load characteristics are adjusted, including: Step 521: Based on the predicted future partition load hours The total network load hours H′ fh The relative change is divided into three equal parts and distributed to the seasonal imbalance coefficient ρ' 0 , Annual average of monthly imbalance rate Annual average daily load factor Including the calculation of the relative change rate of load hours ε by taking the square root of the cube 0 , calculate the change of the seasonal imbalance coefficient of the partition Δρ' 0 , Annual average change of monthly imbalance rate and the annual average daily load factor change Step 522: Future partition seasonal imbalance coefficient ρ′ 0 =ρ′+Δρ′ 0 , according to the monthly distribution of the annual load characteristics of the entire network in the future RM′ m , based on the change in the seasonal imbalance rate between the sub-region and the whole region Δρ′ 0 , and adjust monthly to obtain the monthly distribution of annual load characteristics of future partitions Step 523: The average daily load rate of the future partition in the next year Based on the average daily load rate of the entire network in the future Change in daily average load factor based on the annual average of the sub-region and the whole region Adjust monthly to get the average daily load rate of the future partition Step 524: Calculate the monthly average hourly load based on the future daily load characteristics of the entire network Based on the change of the daily average load rate of the sub-area and the whole area Adjust monthly to obtain the daily load characteristics of the future partition and the monthly average hourly load Step 525: Annual average monthly imbalance rate of future partitions According to the monthly imbalance rate σ′ of the entire network in the future m , based on the change in the annual average of the monthly imbalance rate between the sub-region and the whole region Adjust monthly to get the monthly imbalance rate of future partitions 10. A method for generating load characteristic statistics and hourly load curves according to claim 3 or 8, characterized in that: Use the annual average growth rate method, elasticity coefficient method or unit power consumption method to predict the future annual maximum load FM of the entire network max Or the maximum annual load of the future partition Future annual power consumption of the entire network PM′ sum Or future annual electricity consumption of each region Future network load utilization hours H′ fh or future zone load utilization hours

Images