CN117196683A - LSTM-based thermal power enterprise intra-month rolling matching transaction bidding system - Google Patents
LSTM-based thermal power enterprise intra-month rolling matching transaction bidding system Download PDFInfo
- Publication number
- CN117196683A CN117196683A CN202311247089.7A CN202311247089A CN117196683A CN 117196683 A CN117196683 A CN 117196683A CN 202311247089 A CN202311247089 A CN 202311247089A CN 117196683 A CN117196683 A CN 117196683A
- Authority
- CN
- China
- Prior art keywords
- thermal power
- data
- period
- model
- representing
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000005096 rolling process Methods 0.000 title claims abstract description 44
- 230000005611 electricity Effects 0.000 claims abstract description 106
- 238000000034 method Methods 0.000 claims abstract description 41
- 239000003245 coal Substances 0.000 claims abstract description 22
- 238000012545 processing Methods 0.000 claims description 16
- 238000004364 calculation method Methods 0.000 claims description 9
- 238000005457 optimization Methods 0.000 claims description 8
- 238000010606 normalization Methods 0.000 claims description 7
- 238000010276 construction Methods 0.000 claims description 6
- 230000004913 activation Effects 0.000 claims description 5
- 238000007781 pre-processing Methods 0.000 claims description 5
- 230000002159 abnormal effect Effects 0.000 claims description 4
- 239000000203 mixture Substances 0.000 claims description 4
- CURLTUGMZLYLDI-UHFFFAOYSA-N Carbon dioxide Chemical group O=C=O CURLTUGMZLYLDI-UHFFFAOYSA-N 0.000 claims description 3
- 238000009826 distribution Methods 0.000 claims description 2
- 238000010248 power generation Methods 0.000 abstract description 4
- 230000006870 function Effects 0.000 description 27
- 230000008901 benefit Effects 0.000 description 6
- 238000013528 artificial neural network Methods 0.000 description 5
- 230000008569 process Effects 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000015654 memory Effects 0.000 description 2
- 230000006403 short-term memory Effects 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 241000512668 Eunectes Species 0.000 description 1
- 229920002430 Fibre-reinforced plastic Polymers 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000002860 competitive effect Effects 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000011151 fibre-reinforced plastic Substances 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 230000007787 long-term memory Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000005303 weighing Methods 0.000 description 1
Abstract
A thermal power enterprise intra-month rolling matching transaction bidding method and system based on LSTM, the method comprises the following steps: step 1, collecting output values of thermal power units, operation data of coal consumption coefficients, cost data of the thermal power units, rated power of the thermal power units, historical electricity price data of each period and real winning electricity price of each period; step 2, constructing an average cost model of the thermal power enterprise; step 3, predicting the electricity price by using an LSTM model to obtain predicted electricity price of each period of historical rolling matching transaction, and comparing the predicted electricity price with the actual winning electricity price to obtain an offset proportion; constructing probabilities thereofThe density model is used for constructing a probability density function f (M) taking the price in the time period as a random variable Real i ) The method comprises the steps of carrying out a first treatment on the surface of the And constructing a thermal power enterprise quotation model containing the thermal power unit power, and solving a quotation strategy containing thermal power unit power data. The application obtains the relation between the power generation cost and each parameter of the generator set, so that the electricity price prediction is more consistent with the actual situation and is more reliable.
Description
Technical Field
The application belongs to the technical field of power systems, and particularly relates to an LSTM-based thermal power enterprise intra-month rolling matching transaction bidding system.
Background
In the electric power market, power generation enterprises, typified by thermal power enterprises, need to offer price according to their unit and parameter response capabilities. More specifically, thermal power enterprises must master the factors in terms of supply and demand relationships, cost levels, and the like at any time and quickly make reasonable quotations in each time period.
The bidding method and system in the prior art often consider the composition factors of the generating cost of the unit insufficiently comprehensively, so that the obtained bidding scheme is relatively inaccurate. Meanwhile, the existing bidding method and system often neglect the influence of the bidding interval on the final bidding scheme in the bidding scheme acquisition process, and the accuracy of the final bidding scheme is reduced by adopting a fixed bidding interval instead of a flexible and variable bidding interval.
The bidding method and system in the prior art are limited by the prediction speed of electricity price prediction in terms of speed, and the timeliness is relatively low. In the rolling matching transaction in the month, the quotation is specifically divided into quotations of each time period, so that a time-period quotation scheme is required. Therefore, in order to enhance the bidding efficiency and bidding precision in the rolling matching transaction, there is a need for a thermal power enterprise intra-month rolling matching transaction bidding system with higher precision and better meeting the practical situation and capable of obtaining a time-division bidding scheme.
The authorized prior art 1 (CN 115545768B) provides a large hydropower trans-provincial trans-regional pre-day random bidding method considering contract decomposition, by constructing a random expected model with maximum total expected benefits of two parts of middle-long-term contract market and pre-day market as objective functions and a trans-provincial trans-regional random coordination optimization model with coordinated peak regulation performance and economic benefits, converting the model into an MILP model by adopting a multi-objective conversion and linearization method, and solving the model by utilizing optimization solving software to obtain a hydropower bidding scheme weighing economic benefits of a power station and peak regulation requirements of a power grid at a receiving end, which has the disadvantage that even though the coordination of the economic benefits and peak regulation performance is considered in the objective functions, the method may not fully consider factors such as characteristics, load requirements and power supply capacity of each power station, so that the overall optimization of the bidding scheme is not accurate and fine enough.
Prior art 2 (CN 110276638A) discloses a method and a system for predicting electricity price based on Bi-directional long-short term neural network, by using Bi-LSTM to predict electricity price, forward input and backward input are selected to combine when training a model, and the influence of future data on the present is considered, so as to improve the utilization of existing data. The method has the defects that the prediction speed and the effectiveness are to be enhanced, meanwhile, the electricity price prediction is not actually applied, and the operability and the applicability are to be improved.
Disclosure of Invention
The application provides a thermal power enterprise intra-month rolling matching transaction bidding system based on LSTM, which comprehensively analyzes the power generation cost of a thermal power enterprise unit, including variable cost and fixed cost, and simultaneously predicts the electricity price more quickly and effectively based on a long-short-period neural network LSTM, so as to construct a probability density function taking the price in a period as a random variable, and then construct a bidding model according to the probability density function taking the price in the period as the random variable to obtain a bidding strategy, thereby solving the technical problem that the thermal power enterprise is difficult to rapidly cope with and bid effectively under the background of flexible rolling matching transaction.
The application adopts the following technical scheme. The application provides a thermal power enterprise intra-month rolling matching transaction bidding method based on LSTM, which comprises the following steps:
step 1, collecting output values of thermal power units, operation data of coal consumption coefficients, cost data of the thermal power units, rated power of the thermal power units, historical electricity price data of each period and real winning electricity price of each period;
step 2, constructing an average cost model of the thermal power enterprise according to the data obtained in the step 1;
step 3, predicting the electricity price by using the data obtained in the step 1 and using an LSTM model to obtain the predicted electricity price of each period of the historical rolling matching transaction, and comparing the predicted electricity price with the actual winning electricity price to obtain an offset proportion;
step 4, constructing a probability density model by using the offset ratio obtained in the step 3, and further constructing a probability density function f (M Real i )；
Step 5, combining the average cost model of the thermal power enterprise constructed in the step 2 and the probability density function f (M) taking the price in the time period as a random variable constructed in the step 4 Real i ) And constructing a thermal power enterprise quotation model containing the thermal power unit power, and solving a quotation strategy containing thermal power unit power data.
Preferably, step 2.1, constructing a thermal power unit marginal cost model according to the acquired thermal power unit output value, the operation data of the coal consumption coefficient, the thermal power unit cost data and the rated power of the thermal power unit obtained in step 1;
and 2.2, constructing an average cost model of the thermal power enterprise by using the marginal cost model of the thermal power unit obtained in the step 2.1 and combining the cost composition of the thermal power enterprise.
Preferably, in step 2.1, the thermal power generating unit marginal cost model is expressed by the following formula:
MC＝(2·a·P+b)·Cp+EC+BC (1)
wherein:
MC represents marginal cost of the thermal power generating unit;
a represents a coefficient of 2 times of coal consumption of the thermal power unit;
p represents the output value of the thermal power unit;
b represents a factor of 1 of the coal consumption thermal power unit;
cp represents the trade month coal price;
EC represents carbon dioxide unit electricity discharge cost;
BC represents the blowdown unit power cost;
in step 2.2, the average cost model of the thermal power enterprise is expressed by the following formula:
wherein:
AC represents the average cost of the thermal power unit;
q represents the electric quantity of the thermal power unit in the whole life cycle;
FC represents the fixed cost of the thermal power unit;
LC represents the no-load cost of the thermal power unit;
MC represents marginal cost of the thermal power generating unit.
Preferably, step 3 specifically includes:
step 3.1, preprocessing the data obtained in the step 1, wherein the data are expressed by the following formula:
wherein:
N r power rate data representing no error after pretreatment;
N i the i-th error-free electricity price data;
mean value data representing electricity prices;
n represents a characteristic value of abnormal electricity price data;
step 3.2, normalizing the data obtained in step 3.1, and expressing the normalized data by the following formula:
wherein:
N s the value of the electricity price data sample after normalization processing;
N r,min representing a minimum value in the sample data;
N r,max representing the maximum value in the sample data;
step 3.3, substituting the normalized data in the step 3.2 into an LSTM model to predict the electricity price, and obtaining the predicted electricity price of each period of the historical rolling matching transaction;
step 3.4, outputting R from step 3.3 s Carrying out data inverse normalization calculation on the data to obtain the predicted power price of each period of the historical rolling matching transaction;
and 3.5, calculating an offset ratio by using the predicted electricity price and the actual winning electricity price of each period obtained in the step 3.4, and expressing the offset ratio by the following formula:
wherein:
r i representing the offset ratio of the bid price to the predicted data in each actual period;
M real i Representing the actual bid price per time period.
Preferably, in step 3.3, the input gate I of the LSTM model s Output door O s Forgetting door F s Expressed by the following formula:
wherein:
I si representing data to be processed by the input gate;
F si representing data to be processed in the forget gate;
O si representing data to be processed in the output gate;
sig represents Sigmoid activation function;
q i a weight vector representing valid data in the input gate;
q f a weight vector representing invalid data in the forget gate;
q o a weight vector representing valid data in the output gate;
U s representing input electricity price data;
β i representing the predicted number of electricity pricesA phase vector according to the data;
β f a phase vector representing invalid data in a forget gate of the model;
β o a non-zero vector representing valid data in the model output gate;
tanh represents a hyperbolic tangent function;
q c 、β c a weight vector and a phase vector respectively representing states of the intermediate unit;
R s-1 r represents s The previous output data of this output data value;
wherein the current and last LSTM cell states are expressed as follows:
wherein: c (C) s 、C s-1 Representing the current and last LSTM cell states, respectively.
Preferably, step 4 specifically includes:
step 4.1, calculating the mean and standard deviation of the offset ratio, using the mean and standard deviation, expressing the inexpensive ratio as a random variable from a normal distribution, and further constructing a probability density model f (r) i )；
Step 4.2, using the probability density model f (r) with offset ratio as random variable obtained in step 4.1 i ) Constructing a probability density function f (M Real i )。
Preferably, the probability density model f (r) with offset ratio as random variable in step 4.1 i ) Expressed by the following formula:
wherein:
mean value representing offset ratio;
the standard deviation of the offset ratio is shown.
In step 4.2, the probability density function f (M) is marked as a random variable in the time period Real i ) Expressed by the following formula:
wherein:
X i and represents predicted electricity rate data for each period.
Preferably, step 5 specifically includes:
an objective function of a thermal power enterprise quotation model is constructed and expressed by the following formula:
the constraint conditions of the thermal power enterprise quotation model are constructed and expressed by the following formulas:
wherein:
R k indicating expected revenue for the kth period;
θ represents the proportion of the thermal power unit rated power in the period;
P c representing the rated power of the thermal power generating unit;
Y k an optimal bid representing a kth period;
AC k representing the average cost of the kth period;
Δt represents the duration of the period;
Y m representing the maximum limit.
Preferably, in step 5, the width of the electricity price quotation interval is adjustedTo solve for multiple sets of optimization results to ultimately compare to generate a bid strategy, where the formula for the bid width can be written as:
wherein:
representing the subinterval width of the offer;
m represents the subinterval number thereof;
m represents the number of electron intervals.
The second aspect of the application provides an LSTM-based thermal power enterprise intra-month rolling matching transaction bidding system, which operates the LSTM-based thermal power enterprise intra-month rolling matching transaction bidding method, comprising the following steps: the system comprises a data acquisition module, a data calculation module, a model construction module and an output module;
the data acquisition module is used for acquiring the output value of the thermal power generating unit, the operation data of the coal consumption coefficient, the cost data of the thermal power generating unit, the rated power of the thermal power generating unit, the historical electricity price data of each period and the real winning electricity price of each period;
the data processing module is used for calculating average cost of the thermal power enterprise according to the output value of the thermal power enterprise unit, the operation data of the coal consumption coefficient and the cost data of the thermal power enterprise unit, calculating predicted historical rolling matching transaction electricity prices of each period according to the historical electricity price data of each period, which are acquired by the data acquisition module, based on the LSTM model, and calculating offset proportion of the bid price relative to predicted data in each period according to the predicted historical rolling matching transaction electricity prices of each period and the real bid price of each period, which is acquired by the data acquisition module;
the model construction module is used for constructing an offset proportion probability density function according to the offset proportion of the price in each actual time period relative to the predicted data obtained in the data processing module, and further constructing a probability density function taking the price in the time period as a random variable based on the rated power of the thermal power unit obtained by the data obtaining module; further constructing a thermal power enterprise quotation model containing the power of the thermal power unit;
the output module is used for outputting a quotation strategy containing thermal power unit power data obtained through solving.
Compared with the prior art, the application has the beneficial effects that at least:
1. the application comprehensively analyzes the generating cost of the unit machine set, calculates the marginal cost and the average cost of the power generation respectively from the angles of variable cost and fixed cost, and obtains the relation between the generating cost and each parameter of the generating machine set, so that the electricity price prediction is more in line with the actual situation and is more reliable.
2. The method for predicting the electricity price based on the long-short-period neural network predicts the electricity price in time periods, has higher prediction speed and better timeliness, and can provide more accurate basis for rapid and effective quotation in time periods of subsequent thermal power enterprises according to the prediction results of each time period.
3. The method and the system perform flexible quotation in time intervals, output an optimal quotation strategy in time intervals, increase quotation effectiveness of thermal power enterprises, and stabilize the competitive environment of the electric power market.
Drawings
FIG. 1 is a flow chart of a thermal power enterprise intra-month rolling matching transaction bidding method based on LSTM provided by the application;
FIG. 2 is a schematic diagram of the logic processing unit of the long and short term memory neural network of the present application;
FIG. 3 is a graph showing electricity price prediction and price quotation comparison at 24 hours in an embodiment of the application.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the technical solutions of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application. The described embodiments of the application are only some, but not all, embodiments of the application. All other embodiments, which can be made by those skilled in the art without inventive faculty, are within the scope of the application, based on the spirit of the application.
As shown in fig. 1, embodiment 1 of the present application provides a thermal power enterprise intra-month rolling matching transaction bidding method based on LSTM, which includes the following steps:
step 1, collecting output values of thermal power units, operation data of coal consumption coefficients, cost data of the thermal power units, rated power of the thermal power units, historical electricity price data of each period and real winning electricity price of each period.
In a preferred but non-limiting embodiment of the present application, the historical electricity rate data is 24 hour period historical electricity rate data.
And 2, constructing an average cost model of the thermal power enterprise according to the data obtained in the step 1. In a preferred but non-limiting embodiment of the present application, step 2 specifically comprises:
step 2.1, constructing a marginal cost model of the thermal power unit according to the acquired output value of the thermal power unit, the operation data of the coal consumption coefficient, the cost data of the thermal power unit and the rated power of the thermal power unit, which are obtained in the step 1, and specifically, the marginal cost model is expressed by the following formula:
MC＝(2•a•P+b)•Cp+EC+BC (1)
wherein:
MC represents marginal cost of the thermal power generating unit;
a represents a coefficient of 2 times of coal consumption of the thermal power unit;
p represents the output value of the thermal power unit;
b represents a factor of 1 of the coal consumption thermal power unit;
cp represents the trade month coal price;
EC represents carbon dioxide unit electricity discharge cost;
BC represents the blowdown unit power cost.
And 2.2, constructing an average cost model of the thermal power plant by using the marginal cost model of the thermal power plant obtained in the step 2.1 and combining the cost composition of the thermal power plant, and specifically, expressing the average cost model by the following formula:
wherein:
AC represents the average cost of the thermal power unit;
q represents the electric quantity of the thermal power unit in the whole life cycle;
FC represents the fixed cost of the thermal power unit;
LC represents the no-load cost of the thermal power unit;
MC represents marginal cost of the thermal power generating unit.
It should be noted that, a person skilled in the art may use any other data types to calculate the marginal cost of the thermal power generating unit and the average cost of the thermal power generating enterprise, and the model given in the steps 2.1 and 2.2 is a preferred but non-limiting example, and compared with other models, the model has small calculation amount and low model complexity under the condition of guaranteeing the accuracy, and improves the calculation efficiency.
And 3, predicting the electricity price by using the data obtained in the step 1 and using an LSTM (Long Short-Term Memory neural network) model to obtain the predicted electricity price of each period of the historical rolling matching transaction, and comparing the predicted electricity price with the actual winning electricity price to obtain the offset proportion. In a preferred but non-limiting embodiment of the present application, step 3 specifically comprises:
step 3.1, preprocessing the data obtained in the step 1, wherein the data are expressed by the following formula:
wherein:
N r indicating after pretreatmentError-free electricity price data;
N i the i-th error-free electricity price data;
mean value data representing electricity prices;
n represents a characteristic value of the abnormal electricity rate data.
Step 3.2, normalizing the data obtained in step 3.1, and expressing the normalized data by the following formula:
wherein:
N s the value of the electricity price data sample after normalization processing;
N r,min representing a minimum value in the sample data;
N r,max representing the maximum value in the sample data.
Notably, the actual electricity price is fluctuating and there are often outliers that negatively impact the forecast of the electricity price. Therefore, after collecting the electricity price data, the data processing module of the application preprocesses the random errors in the abnormal electricity price data; the preprocessed electricity price data can enable the collected time-period electricity price data of the rolling match of the electric power market to be more real and effective, and enable the following electricity price prediction to be more real and accurate.
And 3.3, substituting the normalized data in the step 3.2 into an LSTM model to predict the electricity price, and obtaining the predicted electricity price of each period of the historical rolling matching transaction.
Specifically, the logic processing unit of the LSTM model is shown in fig. 2, in which: u (U) s Representing the electricity price data value input after the normalization in the step 3.2; i s Representing the input gate, T s Representing memory cell, O s Representing the output gate, F s Indicating forgetful door R s Representing the inputAnd outputting a data value, wherein I represents an input data preprocessing vector, and T represents a memory data processing vector.
Input value U s During the data processing process, the input gate I is accessed s Preprocessing vector I, output gate O s Forgetting door F s Inputting in equal area, extracting effective data, forgetting ineffective data, and finally obtaining output data value R s Is the most efficient data.
Wherein the input gate I s Output door O s Forgetting door F s The model formulas of (a) are respectively as follows:
wherein:
I si representing data to be processed by the input gate;
F si representing data to be processed in the forget gate;
O si representing data to be processed in the output gate;
sig represents Sigmoid activation function;
q i a weight vector representing valid data in the input gate;
q f a weight vector representing invalid data in the forget gate;
q o a weight vector representing valid data in the output gate;
U s representing input electricity price data;
β i a phase vector representing electricity price prediction processing data;
β f a phase vector representing invalid data in a forget gate of the model;
β o a non-zero vector representing valid data in the model output gate;
tanh represents a hyperbolic tangent function;
q c 、β c a weight vector and a phase vector representing the states of the intermediate cells, respectively.
Wherein the current and last LSTM cell states are expressed as follows:
wherein: c (C) s 、C s-1 Representing the current and last LSTM cell states, respectively.
Step 3.4, outputting R from step 3.3 s The data is subjected to data inverse normalization calculation to obtain the predicted power price of each period of the historical rolling matching transaction, and the predicted power price is expressed by the following formula:
X i ＝R s ·(R s,max -R s,min )+R s,min (7)
wherein:
X i power rate data representing predicted time periods;
R s,min representing the minimum value of the output value, R s,max Representing the maximum value of the output value.
And 3.5, calculating an offset ratio by using the predicted electricity price and the actual winning electricity price of each period obtained in the step 3.4, and expressing the offset ratio by the following formula:
wherein:
r i representing the offset ratio of the bid price to the predicted data in each actual period;
M real i Representing the actual bid price per time period.
Step 4, constructing a probability density model by using the offset ratio obtained in the step 3, and further constructing a probability density function f (M Real world i) A. The application relates to a method for producing a fibre-reinforced plastic composite In a preferred but non-limiting embodiment of the present application, step 4 specifically comprises:
step 4.1, calculating the mean and standard deviation of the offset ratio, using the mean and standard deviation, representing the inexpensive ratio as a function of the normal distributionThe machine variable, and then a probability density model f (r) taking the offset proportion as a random variable is constructed i ) Specifically, the expression is as follows:
wherein:
mean value representing offset ratio;
the standard deviation of the offset ratio is shown.
Step 4.2, using the probability density model f (r) with offset ratio as random variable obtained in step 4.1 i ) Constructing a probability density function f (M Real i ) Specifically, the expression is as follows:
wherein:
X i and represents predicted electricity rate data for each period.
Step 5, combining the average cost model of the thermal power enterprise constructed in the step 2 and the probability density function f (M) taking the price in the time period as a random variable constructed in the step 4 Real i ) And constructing a thermal power enterprise quotation model containing the thermal power unit power, and solving a quotation strategy containing thermal power unit power data. In a preferred but non-limiting embodiment of the present application, step 5 specifically comprises:
an objective function of a thermal power enterprise quotation model is constructed and expressed by the following formula:
the constraint conditions of the thermal power enterprise quotation model are constructed and expressed by the following formulas:
wherein:
R k indicating expected revenue for the kth period;
θ represents the proportion of the thermal power unit rated power in the period;
P c representing the rated power of the thermal power generating unit;
Y k an optimal bid representing a kth period;
AC k representing the average cost of the kth period;
Δt represents the duration of the period;
Y m representing the maximum limit.
In a further preferred but non-limiting embodiment of the present application, step 5 further comprises: adjusting the width of the electricity price quotation intervalAnd solving a plurality of groups of optimization results.
Specifically, solving the above quotation model can obtain a set of optimized results of quotations in each period, but the accuracy of a set of results is insufficient to meet the requirements, so the application adjusts the width of the electricity price quotation intervalTo solve for multiple sets of optimization results to ultimately compare to generate a bid strategy, where the formula for the bid width can be written as:
wherein:
representing the subinterval width of the offer;
m represents the subinterval number thereof;
m represents the number of electron intervals.
The value of l is increased continuously, so thatWith a consequent increase, thus having L nested price intervals +.> Continuously adjusting the lower limit in the same upper limit interval, and recording the quotation result of each adjustment +.>And simultaneously recording the final benefit points under different quotations, and generating the quotation strategies of the component time periods.
The embodiment 2 of the application provides an LSTM-based thermal power enterprise intra-month rolling matching transaction bidding system, which comprises the following steps: the system comprises a data acquisition module, a data calculation module, a model construction module and an output module.
The data acquisition module is used for acquiring the output value of the thermal power enterprise unit, the operation data of the coal consumption coefficient, the cost data of the thermal power enterprise unit, the rated power of the thermal power unit, the historical electricity price data of each period and the real winning electricity price of each period.
The data processing module is used for calculating average cost of the thermal power enterprise according to the output value of the thermal power enterprise unit, the operation data of the coal consumption coefficient and the cost data of the thermal power enterprise unit, calculating predicted historical rolling matching transaction electricity prices of each period according to the historical electricity price data of each period, which are acquired by the data acquisition module, based on the LSTM model, and calculating the offset proportion of the bid price relative to the predicted data in each period according to the predicted historical rolling matching transaction electricity prices of each period and the real bid price of each period, which is acquired by the data acquisition module.
The model construction module is used for constructing an offset proportion probability density function according to the offset proportion of the price in each actual time period relative to the predicted data obtained in the data processing module, and further constructing a probability density function taking the price in the time period as a random variable based on the rated power of the thermal power unit obtained by the data obtaining module; and further constructing a thermal power enterprise quotation model containing the power of the thermal power unit.
The output module is used for outputting a quotation strategy containing thermal power unit power data obtained through solving.
In order to more clearly describe the outstanding essential features and the beneficial technical effects that can be obtained, the present application is further described below with reference to specific examples:
the application is intended to predict in a single step of step size 24, wherein 24 electricity prices in a time series of electricity prices originate from analysis of historical data over the same period of time in the past, and based on 24 observations, the electricity prices in the future 24 times are predicted, and in a specific time, the electricity prices are predicted in synchronization with the historical data.
The test used was software in PyCharm, a language of Python, with Intel (R) Core (TM) i7-4870HQCPU server, and installation of the Python package by Anaconda, with the LSTM mode using the Keras architecture in TensorFlow.
In terms of parameter setting, the activation function adopts a default sigmoid activation function, the Timesteps of LSTM is selected to be 24, the MSE is taken as a loss function, the Adam optimization algorithm is taken as network training, the initial learning rate is set to be 0.05, and finally a prediction result is obtained as shown in figure 3.
Through the information of fig. 3, the LSTM model can reasonably predict the trend of variation of the electricity prices in 24 periods.
TABLE 1 LSTM model predictive value
/>
As shown in Table 1, of the 24 predictions, there were 2 sample points with errors greater than 3%, at 6:00 and 15:00, respectively. The statistics of 22 relatively stable electric power market sample points show that the absolute error percentage is within 3%, and the accuracy degree meets the requirement of 83.3%, so that the LSTM model is considered to have higher prediction accuracy.
And according to the data result predicted by the prediction model, quoting the thermal power enterprise according to the calculation flow of the step four, wherein the specific time-of-day quotation electricity price data are shown in the table 2.
Table 2 model bid values
/>
Compared with the prior art document 1, the method can be widely applied to the quotation decision process of thermal power enterprises, can meet the requirements of thermal power units with different scales and characteristics, and expands the application scene and operability. Meanwhile, through the flexibility of time-division quotation, the method can meet the requirement change under different time intervals and conditions, improves the market adaptability, and better adapts to the dynamic power market requirement compared with the total expected maximum profit model based on medium-long-term contract market and daily market in the authorized technology.
Compared with the prior art document 2, the application constructs the probability density function taking the price in the time period as a random variable, and constructs the quotation model based on the probability density function, so as to solve the problem that thermal power enterprises are difficult to rapidly deal with and effectively quote under flexible rolling matching transaction, and avoid using stronger applicability and applicability. .
Finally, it should be noted that the above embodiments are only for illustrating the technical solution of the present application and not for limiting the same, and although the present application has been described in detail with reference to the above embodiments, it should be understood by those skilled in the art that: modifications and equivalents may be made to the specific embodiments of the application without departing from the spirit and scope of the application, which is intended to be covered by the claims.
Claims (10)
1. The LSTM-based thermal power enterprise intra-month rolling matching transaction bidding method is characterized by comprising the following steps of:
step 1, collecting output values of thermal power units, operation data of coal consumption coefficients, cost data of the thermal power units, rated power of the thermal power units, historical electricity price data of each period and real winning electricity price of each period;
step 2, constructing an average cost model of the thermal power enterprise according to the data obtained in the step 1;
step 3, predicting the electricity price by using the data obtained in the step 1 and using an LSTM model to obtain the predicted electricity price of each period of the historical rolling matching transaction, and comparing the predicted electricity price with the actual winning electricity price to obtain an offset proportion;
step 4, constructing a probability density model by using the offset ratio obtained in the step 3, and further constructing a probability density function f (M Real i )；
Step 5, combining the average cost model of the thermal power enterprise constructed in the step 2 and the probability density function f (M) taking the price in the time period as a random variable constructed in the step 4 Real i ) And constructing a thermal power enterprise quotation model containing the thermal power unit power, and solving a quotation strategy containing thermal power unit power data.
2. The LSTM-based intra-month rolling matching transaction bidding method for thermal power enterprises of claim 1, wherein the method comprises the following steps:
step 2.1, constructing a thermal power unit marginal cost model according to the acquired thermal power unit output value, the operation data of the coal consumption coefficient, the thermal power unit cost data and the rated power of the thermal power unit obtained in the step 1;
and 2.2, constructing an average cost model of the thermal power enterprise by using the marginal cost model of the thermal power unit obtained in the step 2.1 and combining the cost composition of the thermal power enterprise.
3. The LSTM-based intra-month rolling matching transaction bidding method for thermal power enterprises of claim 2, wherein the method comprises the following steps:
in step 2.1, the marginal cost model of the thermal power generating unit is expressed by the following formula:
MC＝(2·a·P+b)·Cp+EC+BC (1)
wherein:
MC represents marginal cost of the thermal power generating unit;
a represents a coefficient of 2 times of coal consumption of the thermal power unit;
p represents the output value of the thermal power unit;
b represents a factor of 1 of the coal consumption thermal power unit;
cp represents the trade month coal price;
EC represents carbon dioxide unit electricity discharge cost;
BC represents the blowdown unit power cost;
in step 2.2, the average cost model of the thermal power enterprise is expressed by the following formula:
wherein:
AC represents the average cost of the thermal power unit;
q represents the electric quantity of the thermal power unit in the whole life cycle;
FC represents the fixed cost of the thermal power unit;
LC represents the no-load cost of the thermal power unit;
MC represents marginal cost of the thermal power generating unit.
4. The LSTM-based intra-month rolling matching transaction bidding method for thermal power enterprises of claim 1, wherein the method comprises the following steps:
the step 3 specifically comprises the following steps:
step 3.1, preprocessing the data obtained in the step 1, wherein the data are expressed by the following formula:
wherein:
N r power rate data representing no error after pretreatment;
N i the i-th error-free electricity price data;
mean value data representing electricity prices;
n represents a characteristic value of abnormal electricity price data;
step 3.2, normalizing the data obtained in step 3.1, and expressing the normalized data by the following formula:
wherein:
N s the value of the electricity price data sample after normalization processing;
N r,min representing a minimum value in the sample data;
N r,max representing the maximum value in the sample data;
step 3.3, substituting the normalized data in the step 3.2 into an LSTM model to predict the electricity price, and obtaining the predicted electricity price of each period of the historical rolling matching transaction;
step 3.4, step3.3 output R s Carrying out data inverse normalization calculation on the data to obtain the predicted power price of each period of the historical rolling matching transaction;
and 3.5, calculating an offset ratio by using the predicted electricity price and the actual winning electricity price of each period obtained in the step 3.4, and expressing the offset ratio by the following formula:
wherein:
r i representing the offset ratio of the bid price to the predicted data in each actual period;
M real i Representing the actual bid price per time period.
5. The LSTM-based intra-month rolling matching transaction bidding method for thermal power enterprises of claim 4, wherein the method comprises the following steps of:
in step 3.3, the input gate I of the LSTM model s Output door O s Forgetting door F s Expressed by the following formula:
wherein:
I si representing data to be processed by the input gate;
F si representing data to be processed in the forget gate;
O si representing data to be processed in the output gate;
sig represents Sigmoid activation function;
q i a weight vector representing valid data in the input gate;
q f a weight vector representing invalid data in the forget gate;
q o a weight vector representing valid data in the output gate;
U s representing input electricity price data;
β i a phase vector representing electricity price prediction processing data;
β f a phase vector representing invalid data in a forget gate of the model;
β o a non-zero vector representing valid data in the model output gate;
tanh represents a hyperbolic tangent function;
q c 、β c a weight vector and a phase vector respectively representing states of the intermediate unit;
R s-1 r represents s The previous output data of this output data value;
wherein the current and last LSTM cell states are expressed as follows:
wherein: c (C) s 、C s-1 Representing the current and last LSTM cell states, respectively.
6. The LSTM-based intra-month rolling matching transaction bidding method for thermal power enterprises of claim 4, wherein the method comprises the following steps of:
the step 4 specifically comprises the following steps:
step 4.1, calculating the mean and standard deviation of the offset ratio, using the mean and standard deviation, expressing the inexpensive ratio as a random variable from a normal distribution, and further constructing a probability density model f (r) i )；
Step 4.2, using the probability density model f (r) with offset ratio as random variable obtained in step 4.1 i ) Constructing a probability density function f (M Real i )。
7. The LSTM-based intra-month rolling matching transaction bidding method for thermal power enterprises of claim 4, wherein the method comprises the following steps of:
step 4.1In the probability density model f (r) with offset ratio as random variable i ) Expressed by the following formula:
wherein:
mean value representing offset ratio;
the standard deviation of the offset ratio is shown.
In step 4.2, the probability density function f (M) is marked as a random variable in the time period Real i ) Expressed by the following formula:
wherein:
X i and represents predicted electricity rate data for each period.
8. The LSTM-based intra-month rolling matching transaction bidding method for thermal power enterprises of claim 7, wherein the method comprises the following steps:
the step 5 specifically comprises the following steps:
an objective function of a thermal power enterprise quotation model is constructed and expressed by the following formula:
the constraint conditions of the thermal power enterprise quotation model are constructed and expressed by the following formulas:
wherein:
R k indicating expected revenue for the kth period;
θ represents the proportion of the thermal power unit rated power in the period;
P c representing the rated power of the thermal power generating unit;
Y k an optimal bid representing a kth period;
AC k representing the average cost of the kth period;
Δt represents the duration of the period;
Y m representing the maximum limit.
9. The LSTM-based intra-month rolling matching transaction bidding method for thermal power enterprises of claim 8, wherein the method comprises the following steps:
in step 5, the width of the electricity price quotation interval is adjustedTo solve for multiple sets of optimization results to ultimately compare to generate a bid strategy, where the formula for the bid width can be written as:
wherein:
representing the subinterval width of the offer;
m represents the subinterval number thereof;
m represents the number of electron intervals.
10. An LSTM-based intra-month rolling match transaction bidding system for thermal power enterprises, operating an intra-month rolling match transaction bidding method for thermal power enterprises according to any one of claims 1 to 9, comprising: the system comprises a data acquisition module, a data calculation module, a model construction module and an output module; the method is characterized in that:
the data acquisition module is used for acquiring the output value of the thermal power generating unit, the operation data of the coal consumption coefficient, the cost data of the thermal power generating unit, the rated power of the thermal power generating unit, the historical electricity price data of each period and the real winning electricity price of each period;
the data processing module is used for calculating average cost of the thermal power enterprise according to the output value of the thermal power enterprise unit, the operation data of the coal consumption coefficient and the cost data of the thermal power enterprise unit, calculating predicted historical rolling matching transaction electricity prices of each period according to the historical electricity price data of each period, which are acquired by the data acquisition module, based on the LSTM model, and calculating offset proportion of the bid price relative to predicted data in each period according to the predicted historical rolling matching transaction electricity prices of each period and the real bid price of each period, which is acquired by the data acquisition module;
the model construction module is used for constructing an offset proportion probability density function according to the offset proportion of the price in each actual time period relative to the predicted data obtained in the data processing module, and further constructing a probability density function taking the price in the time period as a random variable based on the rated power of the thermal power unit obtained by the data obtaining module; further constructing a thermal power enterprise quotation model containing the power of the thermal power unit;
the output module is used for outputting a quotation strategy containing thermal power unit power data obtained through solving.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311247089.7A CN117196683A (en) | 2023-09-26 | 2023-09-26 | LSTM-based thermal power enterprise intra-month rolling matching transaction bidding system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311247089.7A CN117196683A (en) | 2023-09-26 | 2023-09-26 | LSTM-based thermal power enterprise intra-month rolling matching transaction bidding system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117196683A true CN117196683A (en) | 2023-12-08 |
Family
ID=88999705
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311247089.7A Pending CN117196683A (en) | 2023-09-26 | 2023-09-26 | LSTM-based thermal power enterprise intra-month rolling matching transaction bidding system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117196683A (en) |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060276938A1 (en) * | 2005-06-06 | 2006-12-07 | Equinox Energy Solutions, Inc. | Optimized energy management system |
| JP2008146105A (en) * | 2006-12-05 | 2008-06-26 | Toshiba Corp | System, method and program for evaluating power transaction |
| CN108898282A (en) * | 2018-06-06 | 2018-11-27 | 华北电力大学 | Data center resource Optimization Scheduling and computer storage medium |
| CN109829742A (en) * | 2018-11-08 | 2019-05-31 | 中国电力科学研究院有限公司 | A kind of fired power generating unit optimum price quotation calculation method and system based on Shen Pay as bid |
| CN111882452A (en) * | 2020-06-18 | 2020-11-03 | 国网天津市电力公司电力科学研究院 | Marginal cost calculation method for integrated energy system participating in demand side response |
| CN112183967A (en) * | 2020-09-14 | 2021-01-05 | 南方电网能源发展研究院有限责任公司 | Gas turbine unit control method and device based on operation model |
| US20210044117A1 (en) * | 2019-08-09 | 2021-02-11 | Changsha University Of Science & Technology | Method for dynamically and economically dispatching power system based on optimal load transfer ratio and optimal grid connection ratio of wind power and photovoltaic power |
| CN112862170A (en) * | 2021-01-29 | 2021-05-28 | 西安峰频能源科技有限公司 | Marginal cost-based auxiliary unit multi-section quotation model |
| CN115511541A (en) * | 2022-10-28 | 2022-12-23 | 湖北电力交易中心有限公司 | Time-period electricity price calculation method based on supply and demand situations and system cost |
| CN115829620A (en) * | 2022-09-23 | 2023-03-21 | 国能粤电台山发电有限公司 | Coal-fired unit dynamic power generation cost prediction method considering carbon emission cost |
| CN116227734A (en) * | 2023-03-30 | 2023-06-06 | 东南大学 | Main and auxiliary market joint optimization method considering interruptible load |
-
2023
- 2023-09-26 CN CN202311247089.7A patent/CN117196683A/en active Pending
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060276938A1 (en) * | 2005-06-06 | 2006-12-07 | Equinox Energy Solutions, Inc. | Optimized energy management system |
| JP2008146105A (en) * | 2006-12-05 | 2008-06-26 | Toshiba Corp | System, method and program for evaluating power transaction |
| CN108898282A (en) * | 2018-06-06 | 2018-11-27 | 华北电力大学 | Data center resource Optimization Scheduling and computer storage medium |
| CN109829742A (en) * | 2018-11-08 | 2019-05-31 | 中国电力科学研究院有限公司 | A kind of fired power generating unit optimum price quotation calculation method and system based on Shen Pay as bid |
| US20210044117A1 (en) * | 2019-08-09 | 2021-02-11 | Changsha University Of Science & Technology | Method for dynamically and economically dispatching power system based on optimal load transfer ratio and optimal grid connection ratio of wind power and photovoltaic power |
| CN111882452A (en) * | 2020-06-18 | 2020-11-03 | 国网天津市电力公司电力科学研究院 | Marginal cost calculation method for integrated energy system participating in demand side response |
| CN112183967A (en) * | 2020-09-14 | 2021-01-05 | 南方电网能源发展研究院有限责任公司 | Gas turbine unit control method and device based on operation model |
| CN112862170A (en) * | 2021-01-29 | 2021-05-28 | 西安峰频能源科技有限公司 | Marginal cost-based auxiliary unit multi-section quotation model |
| CN115829620A (en) * | 2022-09-23 | 2023-03-21 | 国能粤电台山发电有限公司 | Coal-fired unit dynamic power generation cost prediction method considering carbon emission cost |
| CN115511541A (en) * | 2022-10-28 | 2022-12-23 | 湖北电力交易中心有限公司 | Time-period electricity price calculation method based on supply and demand situations and system cost |
| CN116227734A (en) * | 2023-03-30 | 2023-06-06 | 东南大学 | Main and auxiliary market joint optimization method considering interruptible load |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Raza et al. | An ensemble framework for day-ahead forecast of PV output power in smart grids | |
| CN111709672A (en) | Virtual power plant economic dispatching method based on scene and deep reinforcement learning | |
| Lu et al. | A hybrid deep learning-based online energy management scheme for industrial microgrid | |
| CN110380444B (en) | Capacity planning method for distributed wind power orderly access to power grid under multiple scenes based on variable structure Copula | |
| Toubeau et al. | Capturing spatio-temporal dependencies in the probabilistic forecasting of distribution locational marginal prices | |
| Huang et al. | A control strategy based on deep reinforcement learning under the combined wind-solar storage system | |
| Yuan et al. | Conditional style-based generative adversarial networks for renewable scenario generation | |
| CN112950098A (en) | Energy planning method and device based on comprehensive energy system and terminal equipment | |
| CN115688970B (en) | Micro-grid two-stage self-adaptive robust optimal scheduling method based on interval probability uncertainty set | |
| Liu et al. | Forecasting the occurrence of extreme electricity prices using a multivariate logistic regression model | |
| Xiao et al. | Short-term power load interval forecasting based on nonparametric Bootstrap errors sampling | |
| Bagheri et al. | Stochastic optimization and scenario generation for peak load shaving in Smart District microgrid: sizing and operation | |
| CN112348287A (en) | Electric power system short-term load probability density prediction method based on LSTM quantile regression | |
| CN115313361A (en) | Joint optimization control method and device for large-scale adjustable resources | |
| Dumas et al. | Deep learning-based multi-output quantile forecasting of PV generation | |
| Bakhtvar et al. | A vision of flexible dispatchable hybrid solar‐wind‐energy storage power plant | |
| Budin et al. | Day-ahead multiple households load forecasting using deep learning and unsupervised clustering | |
| Yang et al. | Short-term demand forecasting method in power markets based on the KSVM–TCN–GBRT | |
| Huang et al. | Portfolio management for a wind‐storage system based on distributionally robust optimisation considering a flexible ramping product | |
| Zhang | Short-term power load forecasting based on SAPSO-CNN-LSTM model considering autocorrelated errors | |
| CN115545768B (en) | Large hydropower trans-province trans-regional day-ahead random bidding method considering contract decomposition | |
| Liu et al. | Embedded Scenario Clustering for Wind and Photovoltaic Power, and Load Based on Multi-Head Self-Attention | |
| CN116362136A (en) | Self-dispatching optimization method and system for independent energy storage system | |
| Feng et al. | A novel hybrid model for short-term prediction of PV power based on KS-CEEMDAN-SE-LSTM | |
| CN117196683A (en) | LSTM-based thermal power enterprise intra-month rolling matching transaction bidding system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |