The power industry is undergoing an unprecedented industry restructuring around the world, the essence of which is to introduce competition into the power industry. The first to introduce competition into the power industry is California. In the California daily bidding power market, the electricity supplier submits the bidding curve of the electricity company's total or each generator set hourly electricity price to the power trading intermediary. PX accumulates the electricity price bid curve of all power suppliers. Get the bidding curve of the system's total supply electricity price. On the other hand, the user entrusts the electric service company to submit the PX hourly demand electricity price bidding curve, which is also accumulated by PX to form the system total demand electricity price bidding curve, and then PX is recorded as λ above. After the bidding is over, each bidder also needs to use the bidding power generation as the system demand, and carry out the unit combination or the combined heat and power economic dispatch in the traditional sense to minimize the cost of power generation.
In the electricity market, for every power supplier participating in the bidding, the ideal bidding strategy should require the maximum profit to be used for bidding purposes. The use of game theory is a more natural method. The game theory method is more suitable for the case where the bidding strategy of a single power supplier has no effect on λ. Because there are many random factors in the power market, bidding has certain risks, so it may be more realistic to find a satisfactory solution in the sense of probability than to find the optimal solution. In this regard, we have adopted a new intelligent computing method--sequence optimization method to study the selection of an independent power supplier bidding strategy. Simulation results show that order optimization is an effective method.
According to the rules, there is no special compensation for the start-stop costs of the units in the bidding profit. Therefore, each power supplier should take into account the starting and stopping costs of the unit while giving the bidding curve. At this time, the formula (1) becomes a problem of how to select the bidding curve to maximize the profit under the constraints of the minimum start-up time and the slope of the climbing slope. Since the PX bidding market rules stipulate that the bidding curve should be a monotonically increasing piecewise linear function, choosing the bidding strategy is actually how to select the two endpoints of each segmented line on the bidding curve.
This simple model describes how the bidder's bidding strategy choice affects the MCP in the case where other bidders' bidding strategies have been determined. In general, the closer the point of λ is, the greater the influence is, ie the corresponding θ value is also Bigger. If all θ are zero, E will become pricetaker, ie no matter how E's bidding strategy is chosen, it will not affect λMCP. Note that the above model is only a simple model of the impact of bidding strategy. In order to reflect the error perturbation, an ii d (independently distributed) error perturbation term k(t) is added to equation (5), and the model is rewritten as follows: Note that the above model does not consider the game theory between bidders. Although it is possible to use a regression analysis or a game theory to obtain a more accurate model in order to better reveal the essential relationship between the two, it does not hinder the use of sequential optimization methods to obtain better results, because the essence of sequential optimization That is to use a rough model to evaluate the advantages and disadvantages between the various solutions.
2 Bidding optimization strategy based on the order optimization method 2.1 Given the basic bidding curve to give the basic bidding curve has two purposes: 1 This is the basis for determining the relationship between the bidding curve and λ using equation (5). A bidding curve for a single power supplier's bidding strategy that has no effect on λ.
In order to obtain the basic bidding curve, the number of endpoints that should be given for each time period in the K group electricity market price is given. Using equation (7), the corresponding power generation for the maximum profit for each unit can be obtained, or the physical characteristics of a single unit can be met, where k is the serial number of the endpoint on the bid curve. When given a set of market prices bk = 1, the problem (8) is actually a parameter optimization problem for integrated resource scheduling, and it is also a problem when using the Lagrangian relaxation method to solve the unit combination problem or the joint economic dispatch of water and thermal power. The problem, where λ(t) can be thought of as the Lagrangian multiplier obtained by the higher-level iteration. The dynamic bidding method can be used to effectively solve the result. The basic bidding curve of the single unit and the accumulating unit of the generating capacity p of each unit per period can be obtained by formula (9), Electricity market column? Guan Xiaohong and other power market bidding strategies for power suppliers are selected (10) respectively: 2. 2 perturbation of the basic bidding curve. By composing the basic bidding curve, N bidding curves can be obtained: Δb is a simple implementation. The method of Δb( , t) is to keep the power generation amount g(t ) unchanged, and randomly select from the following intervals to obtain the monotonically increasing piecewise linear function bid curve that b can always maintain the mechanism PX. Based on the bidding disturbance curve b to obtain the corresponding λ2.3. Selecting a sufficiently good bidding strategy The bidding curve obtained from Section 3.2 can be roughly evaluated and sorted by the method of order optimization, and the specific evaluation is completed by the formula (13). : Note that (13) is only a rough estimate of its profit under given λ. Using the formulas (2) to (4), the real profit J and the profit obtained by the formula (13) may have a large error, that is, where X is the error term.
The advantage of the order optimization method is that a relatively coarse model can be used to distinguish between better and worse solutions, that is, due to the error, the size order between the solutions is relatively unaffected. Even if the N bid curves are evaluated using a very rough model, the probability that the bid strategy subset contains some good enough solutions is still large.
The main purpose of order optimization is to find a subset of bidding strategies S, with a high probability of containing a good enough solution, and at the same time determine the size of s = S. The degree of good enough is determined by the column probability, where G is a sufficiently good subset of the bidding strategy, defined as topn a of the N bidding strategies defined as the alignment level, visually speaking the column probability is a subset of the bidding strategy At least one bidding strategy in S is also a probability of being sufficiently good in the subset of bidding strategies G.
Using formula (11) to select s bidding strategies from N bidding strategies, the resulting profits can be obtained by equation (13) and sorted in order. The best bidding strategy is selected as the bidding strategy selection subset S, where the size of s can be obtained by the calculation equation in [4]: ​​where g = G is the size of the subset G Z o, d, V, Z is a coefficient or a parameter whose values ​​depend on the magnitude of the column line probability P and the corresponding order characteristic curve.
It takes very little time to evaluate N bidding strategies by using (13), and the method of order optimization can ensure that there are enough good solutions in s bidding strategies.
2. 4 Select the final bidding strategy Use formula (2) and formula (4) to accurately evaluate s bidding strategies, and select the best bidding strategy based on forecast λ. Since s = S is much smaller than N, the time is greatly shortened compared to the exact solution of N bidding strategies.
3 Simulation and Results Simulation The selection strategy strategy was chosen by an independent power supplier with 10 thermal power units. The physical characteristics of the 10 units are listed in the literature [7]. Four cases are considered, that is, the λ of the two sets of basic market prices obtained by equations (5) and (6) are tested separately, and the force market PX day bidding scheduling The λM CP of May 1, 1998 and January 4, 1999, see Figure 1. The parameters required in the optimization method are as follows: bidding strategy selection space: N = 1000 column line probability: = 0. 95 is good enough Subset: G is the experimental basis in each bidding strategy. There is no noise. If you want to use equation (15) to find s, you need to guide the bidding strategy to solve the problem. OPC is Bell. We used the scientific computational simulation language MA TLAB 5. 0 to calculate the exact O PC using the scheduling algorithm, with an average time of 11 h, which is difficult to apply in practice. In order to calculate s, it is also necessary to know the variance of the noise distribution. The X variances of Experiments 1 and 2 are 0. 089 and 0. 0343, respectively. Correspondingly, the parameters for s can be calculated for W. 0. It is worth noting that the experiments show that the error is basically a normal distribution, which is different from the uniform distribution assumed in [4]. So here is just an approximate calculation s, but the experimental results show that the size of the resulting s is sufficient.
The simulation results obtained by the order optimization method. Figure 2 shows the bidding curve for Experiment 1 during time period 20. Table 2 lists the size of s. It can be seen that s is much smaller than N. In other words, the method of order optimization only needs to solve s = 39 scheduling problems, and the calculation time is greatly shortened. Moreover, we only use interpreted language to calculate very slowly. If we adopt C, we believe that the efficiency will be much higher.
Experimental profit/dollar In order to explain that the order optimization method has certain immunity to noise, the intersection of the sufficiently good subset G in the four experiments and the bidding strategy subset S obtained by the rough calculation is 31, respectively. 18, 26, 16, it can be seen that the intersection of G and S is much larger than 5. At the same time, in order to verify the influence of noise on λ, a uniformly distributed disturbance variable k= U~[ 1, 1]. Experiments 3 and 4 show that due to the presence of noise, G∩S is correspondingly smaller than the data in Experiments 1 and 2, that is, due to the presence of noise, the solution of the bidding strategy subset is sufficiently good, but G∩ The size of S is still much larger than 5, again indicating the effectiveness of the order optimization method.
4 Summary This paper adopts a new intelligent computing method-order optimization method to study the selection of an independent power supplier bidding strategy. Firstly, the simulation model of the electricity market is established and a small and good bidding strategy search space S is constructed. At the same time, it is guaranteed that S has a good enough solution in the sense of probability, and then through the accurate calculation of each element in S. The best bidding strategy, and use this as the final bidding strategy. Simulation results based on an independent power supplier with 10 thermal power units and the historical market price of the California power market show that sequencing optimization is an effective method.
Western Shirts,Popular Long-Sleeve Shirt,Western Men'S Shirts,White Western Shirts
SHAOXING UNIVERSE IMPORT AND EXPORT CO.,LTD , https://www.zjshirts.com