JPWO2022070251A5 - - Google Patents

Description

The supply rate function indicates a relationship in which a bid rate Rs is input, and if the bid amount at a rate equal to or lower than the bid rate Rs is equal to or higher than the bid price Ps, which is the output of the supply rate function, the item is sold. The demand rate function indicates a relationship in which a bid rate Rd is input, and if the bid amount at a rate equal to or lower than the bid rate Rd is equal to or lower than the bid price Pd, which is the output of the demand rate function, the item is purchased.

The same applies to the case where the function form is determined in advance and the parameters are appropriately given as in Patent Document 2. In addition, since the supply and demand curve is generated by accumulating bids, it becomes a step-like function. If there are more market participants bidding, even power sources of the same generation method will not bid at exactly the same price, so it can be expected that the function will become closer to a smooth function, but there is no clear distinction between price ranges. , there is a problem that it is difficult to link it with the knowledge of trading participants as described above.

According to the present invention, it is possible to provide a technique for predicting the demand curve and the supply curve based on the trading participants' knowledge of the power generation cost and the electricity selling price.

The error calculation and backpropagation means 115 compares the market data consisting of the buy total bid volume, the sell total bid volume, the execution amount and the execution price with the result of the execution price and execution amount calculation means 114 to calculate the estimated error is calculated, and the feedback amount to the bid amount estimation/learning and prediction means 111 is calculated.

The display data generating means 117 generates display data from the demand curve data, the supply curve data, the calculation of the error , and the output of the back propagation means 115 .

The error calculation and backpropagation means 115 uses the instruction data 102 and the output of the contract price and contract quantity calculation means 114 (input 183 to the error calculation and backpropagation means 115 ) to estimate/learn and predict the bid volume. The error feedback amount to each node of the output layer 118 of the means 111 is calculated, and these vectors 184 are fed back to the output layer 118 . In addition, in order to display the estimation result in the learning stage, the display error 185 is also calculated and output to the display data generation means 117 .

Equation (3-1) refers to the case where the contract quantity estimate esbids is determined by the demand curve and the contract price estimate esprice is determined by the supply curve. Described. To explain this condition graphically, it shows the condition where the portion of the demand curve that is horizontal to the price axis and the portion of the supply curve that is perpendicular to the price axis intersect. mathematically, the minimum value of the price range of the k-th supply bid step is between the maximum value of the price range of the i-1th demand bid step and the maximum value of the price range of the i-th demand bid step, When there is k and i such that the bid volume up to the i-th demand bid step is less than the bid volume up to the k-th supply bid step and greater than the bid volume up to the k-1-th supply bid step , indicates that it is sufficient to determine that such a condition holds.

Equation (3-2) refers to the case where the supply curve determines the contract volume estimate esbids and the contract price estimate esprice is determined by the demand curve. It is described by a formula. Graphically, it shows the condition where the part of the supply curve that is horizontal to the price axis and the part of the supply curve that is perpendicular to the price axis intersect.

The difference between the actual value (contract amount excbids ) and the estimated contract amount (contract amount estimated value esbids ) is the minimum bid amount (price range minimum value) lower than the contract price in the supply bidding steps (supply bidding steps 171 to 175). Since it can be considered that there is a problem in estimating the step (supply bidding steps 172 to 175) with which the value of Eq. is the simplest error correction method. Equation (4) expresses the calculation of such an error correction vector.

When a decision tree is used as the market pattern cluster learning means 802 and the market pattern cluster prediction means 902, when a certain weather forecast value is given, it is determined which leaf of the decision tree the data set corresponds to. , will determine the expected clusters. However, since any leaf is usually not 100% cluster A or class B, it is probabilistically predicted that cluster A is 60%, cluster C is 20%, cluster D is 10%, and so on. In the case of neural networks as well, in the class classification problem, a softmax function is used in the output layer to determine one cluster. have similar results. Therefore, it is possible to simultaneously execute a plurality of bid support devices (811, 812, 813) considering that they belong to a plurality of clusters with respective probabilities. In that case, the output from the market pattern cluster prediction means 902 to the display data generation means 917 at the time of prediction (prediction result 931 ) is provided with identification information of clusters assumed to occur from the weather forecast data and information on their occurrence probability. .

As the teaching data 1102, the sell bid price and sell bid volume, the buy bid price and the buy bid volume can be used. Demand curve constructing means using demand bid volume forecast values 1131 to 1135 corresponding to a plurality of demand bid prices and supply bid volume forecast values 1141 to 1145 corresponding to a plurality of supply bid prices included in teaching data 1102 constituting 112 and supply curve constructing means 113 generate demand curve data 181 and supply curve data 182 , which are input to display data generating means 1117 and error calculation and backpropagation means 1115 .

FIG. 12 shows an embodiment in which, instead of the bid amount learning/predicting means 1111, separate bid amount learning/predicting means 1211a to 1211j are used to learn and predict the bid amounts for each price range. In the embodiments so far, an example of configuring with a single neural network has been shown, but in this embodiment, for example, each can be configured by means of support vector regression, linear regression, or the like. Of course, it is also possible to regress with a neural network. Since learning is performed exclusively for each price range, it is possible to make highly accurate predictions compared to the implementation method of the first embodiment. The calculation of the error and the processing of the backpropagation means 1215 are the same as in the embodiment of FIG. 11, but the backpropagation must be configured to return to the respective bid amount learning/prediction means 1211a to 1211j separately. (1284a-j).

Contract performance data learning/prediction means 1401 calculates and outputs contract performance data prediction results 1402 for learning inputs including weather forecasts and dates. The error calculation and backpropagation means 1415 calculates the error 1484 between the same items in the teaching data 102 and the contract performance data prediction result 1402 and back propagates it to the contract performance data learning/prediction means 1401 .

The correspondence between (1) to (4) above and FIG. (3) corresponds to error computation and backpropagation means 115; (4) corresponds to feeding back vector 184 to output layer 118; That is, in this embodiment, the learning means includes the bid amount estimation/learning and forecasting means 111, the demand curve constructing means 112, the supply curve constructing means 113, and the error calculation and backpropagation means 115.

Further, in the power market trading support platform according to the second embodiment, a classification means (for example, clustering means 801) for classifying the state of the market into a finite number using the area price, and each finite number of market states , a finite number of first learning means (e.g., bidding support device 811, 812, 813), a second learning means (for example, market pattern cluster learning means 802) for learning the relationship between past weather forecast values, the date and time, and the state of division of the market, and a second learning means from the weather forecast values a first prediction means (for example, market pattern cluster weather forecast performance selection means 803) for predicting one or more market conditions corresponding to forecast values using the learning result of the learning means ; Using one or more of the learning results of the first learning means for learning the weather forecast value and the relationship between the date and time and the bid volume in the bid price range for one or more market conditions corresponding to the prediction result of the prediction means 2. The method is characterized by estimating the bidding volume for each price band on the demand curve and the bidding volume for each price band on the supply curve from weather forecast values.

100,500 bid support device 101 learning input 102 teaching data 111 bid amount estimation/learning and prediction means (model, learning means)
112 demand curve construction means (learning means)
113 supply curve construction means (learning means)
114 Calculation means for execution price and execution amount 115 Error calculation and backpropagation means (learning means)
117,517 Display data generation means 118 Output layer 119 Display device 120 Input means 121-125 Price range of demand curve 131-135, 141-145 Output 151-155 Price range of supply curve 161-165 Demand bidding step 171-175 Supply Bidding step 410, 710 Display screen (display means)
411 Supply and demand curve estimation result 412 Price range setting unit 413 Price range confirmation unit 415 User 420,720 Storage resource 421,721 Calculation resource 422,722 Learning model database 423,723 User preference database 424,724 Market execution result database 430,730 Weather forecast service 440,740 Power market system 501 Prediction input 511 Bid amount prediction means (means for predicting bid amount)
512 demand curve constructing means 513 supply curve constructing means 710 display screen (display means for user environment)
711 Prediction result 712 Weather forecast data 713 Setting unit 714 Equipment capacity information 715 User 718 Input unit 725 Generator information database 750 Power generation facility 760 Website 800, 900 Bidding support system 801 Clustering means (classifying means)
802 Market pattern cluster learning means (second learning means)
803 Market pattern cluster weather forecast performance selection means (first prediction means)
804 Input means 811, 812, 813 Bid support device (first learning means)
821 weather forecast performance data 822 clustering data 903 supply and demand curve prediction means selection means 1400, 1500 bidding support system
1401 contract performance data learning/predicting means (third learning means)
1402, 1502 contract performance data prediction result 1484 error 1415 error calculation and backpropagation means 1501 contract performance data prediction means 1515 user

Claims (1)

The power market trading support platform according to any one of claims 3 to 6,
Classification means for classifying market conditions into a finite number using area prices;
For each of the finite number of market conditions, a finite number of first market conditions are input, and the relationship between the weather forecast value or actual value and the bid amount in the bid price range is learned. a means of learning;
a second learning means for learning the relationship between the past weather forecast values, the date and time, and the segmented state of the market;
a first prediction means for predicting one or more market conditions corresponding to the weather forecast values using the learning results of the second learning means from the weather forecast values ;
one or more of the learning results of the first learning means for learning the relationship between the weather forecast value, the date and time, and the bid volume in the bid price range for one or more market conditions corresponding to the prediction result of the first prediction means; An electric power market transaction support platform characterized by estimating the bidding volume for each price band on the demand curve and the bidding volume for each price band on the supply curve from weather forecast values, using a plurality of data.

Images