Power interchange system

Abstract

To provide a power interchange system consisting of a solar battery and a storage battery capable of reducing the battery capacity as low as possible while ensuring surplus electricity that is capable of being sold during times when the selling price is favorable.SOLUTION: A disclosed power interchange system corrects the power generation forecast to increase in the first period when the difference between the predicted and actual solar radiation values is small, and in the second period where the difference is large, the power generation forecast is corrected downward. With this, the battery is charged during the first period, and during the second period, the battery is discharged.SELECTED DRAWING: Figure 10B

Description

[Technical Field]

[0001] The present invention relates to a technology for transmitting power generated by a power generation system configured with solar cells and storage batteries between base stations via a power transmission network. [Background technology]

[0002] Self-dispatch is when a business that owns a power generation facility uses the power grid owned by the power utility to transmit the electricity generated by that facility between bases. When implementing self-dispatch, it is necessary to submit a transmission plan to the operating organization (e.g., Organization for Cross-regional Coordination of Transmission Operators (OCCTO)) before the start of the transmission (e.g., the day before). Furthermore, while self-dispatch is being carried out, it is necessary to transmit the electricity as planned at specified time intervals (e.g., 30 minutes). If the actual transmitted power deviates from the planned power (imbalance), a penalty is generally imposed.

[0003] The following Patent Document 1 addresses the issue of "providing a power management system that controls the charging and discharging of storage batteries so that the storage batteries of power consumers can be used more effectively," and describes the following technology (see abstract): "The power management system is a power management system that manages the power supplied to the load equipment of the power consumer from at least a power grid and storage batteries, and includes a plan creation unit that creates a storage battery control plan in advance of a predetermined day, the plan including schedule information for the charging and discharging state of the storage battery on a predetermined day from the next day onwards, corresponding to a predicted power demand value for the predetermined day on or after the next day, which is predicted based on the power consumer's past power demand results, and a storage battery control unit that controls the charging and discharging state of the storage battery on the predetermined day based on the storage battery control plan and the remaining charge of the storage battery on the predetermined day."

[0004] Japanese Patent Application No. 2022-144965 describes a technology that assumes a power generation system that generates power to transmit by self-consignment and is composed of solar cells and a storage battery. In this document, for a power transmission period in which the range and frequency of fluctuations in solar radiation are predicted to be the largest, even if the actual amount of power generated by the solar cells is less than the predicted amount of power generation, the state of charge of the storage battery is determined so that the shortfall can be compensated for by discharging from the storage battery. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Japanese Patent Publication No. 2023-083813 Summary of the Invention [Problem to be solved by the invention]

[0006] Consider a case where a power generation system that generates power to be transmitted by self-dispatch is composed of, for example, solar cells and storage batteries. In order to transmit power in accordance with a pre-planned manner during self-dispatch, when the solar cells generate power in excess of the planned transmission power, it is necessary to suppress the amount of power generated so that it is less than the maximum power point and to use the surplus power to charge the storage battery. In other words, when the amount of power generated by the power generation system exceeds the planned transmission power, the solar cells are instructed to suppress the amount of power generated so that it is less than the maximum power point and the storage battery is instructed to charge. Conversely, when the amount of power generated by the solar cells is less than the planned transmission power, the storage battery is discharged.

[0007] In order to comply with the power transmission plan, it is desirable to ensure that the storage battery has as large a charging capacity as possible (i.e., to install a large-scale storage battery) and to use the storage battery to compensate for any excess or deficiency in the amount of power generated by the solar cells. However, since installing a large-capacity storage battery requires a considerable cost, there is a need to keep the capacity of the storage battery as low as possible.

[0008] Apart from the capacity of the storage battery, there is also room for consideration of the timing at which the storage battery is charged and discharged. If the actual transmission power is insufficient compared to the transmission plan power (i.e., a shortage imbalance occurs), a penalty fee is levied. On the other hand, if the actual transmission power is greater than the transmission plan power (i.e., a surplus imbalance occurs), it is possible to sell the surplus power to the power grid. Since it is necessary for the power grid as a whole to have a balance between power demand and power supply, during times when this balance is likely to be disrupted, the selling price tends to be favorable for the seller. Therefore, there is a need for an electricity interchange system to be operated so that surplus power can be transmitted to the power grid during times when the selling price is as favorable as possible.

[0009] The present invention has been made in consideration of the above-mentioned problems, and aims to minimize the storage battery capacity in a power interchange system composed of solar cells and storage batteries, while securing surplus electricity that can be sold during times when the selling price is favorable. [Means for solving the problem]

[0010] The power interchange system of the present invention corrects the predicted power generation amount to be higher during a first time period when the discrepancy between the predicted and actual solar radiation values ​​is small, and corrects the predicted power generation amount to be lower during a second time period when the discrepancy is large, thereby charging the storage battery during the first time period and discharging the storage battery during the second time period. [Effects of the Invention]

[0011] According to the power interchange system of the present invention, in a power interchange system composed of solar cells and storage batteries, it is possible to minimize the storage battery capacity and secure surplus power that can be sold during times when the selling price is favorable. Other issues, configurations, effects, etc. will become clear from the description of the following embodiments. [Brief explanation of the drawings]

[0012] [Figure 1A] 1 is a configuration example of a power generation system provided by a business operator that performs self-consignment. [Figure 1B] 1 is a configuration example of a power generation system provided by a business operator that performs self-consignment. [Figure 2] FIG. 10 is a diagram illustrating a consignment plan. [Figure 3] Here are some examples of weather codes provided by weather information providers: [Figure 4A] FIG. 1 is a diagram comparing the actually measured amount of solar radiation and predicted solar radiation data for a certain day. [Figure 4B] FIG. 10 is a diagram comparing the actually measured amount of solar radiation and predicted solar radiation data for another day. [Figure 5] 4 is a schematic diagram showing a procedure in which the power control device 1 learns detailed weather classifications. FIG. [Figure 6] FIG. 6 is a diagram illustrating a procedure for estimating a detailed weather classification using the results of learning according to the procedure in FIG. 5. [Figure 7] 6 is a diagram illustrating an example of the relationship between the output value of a hidden layer neuron in the neural network of FIG. 5 and the amount of wheeling imbalance. [Figure 8A] An example of predicted solar radiation values ​​for four days is shown below. [Figure 8B] An example of estimating the range of fluctuations in solar radiation from the output of hidden layer neurons is shown below. [Figure 9A] An example of predicted and actual power generation from solar cells on DAY 3 is shown below. [Figure 9B] 9B shows the result of the power control device 1 correcting the power generation amount prediction of FIG. 9A. [Figure 10A] An example of the predicted and actual power generation amount for solar cells on DAY 4 is shown below. [Figure 10B] 10B shows the result of the power control device 1 correcting the power generation amount prediction of FIG. 10A. [Figure 11A] An example of predicted and actual power generation for solar cells on DAY 1 is shown below. [Figure 11B] 11B shows the result of the power control device 1 correcting the power generation amount prediction of FIG. 11A. [Figure 11C] Another example of correcting the power generation amount prediction on DAY1 is shown. [Figure 12A] An example of predicted and actual power generation for solar cells on DAY 2 is shown below. [Figure 12B] 12B shows the result of the power control device 1 correcting the power generation amount prediction of FIG. 12A. [Figure 12C] Another example of correcting the power generation forecast on DAY2 is shown. [Figure 13] 1 shows the relationship between solar cell power generation and imbalance charges. [Figure 14A] 10 is a flowchart illustrating a procedure in which the power control device 1 corrects the predicted power generation amount. [Figure 14B] 10 is a flowchart illustrating a procedure in which the power control device 1 corrects the predicted power generation amount. [Figure 15] 10 is an example of a user interface provided by the calculation unit 11. DETAILED DESCRIPTION OF THE INVENTION

[0013] 1A and 1B show an example of the configuration of a power generation system provided by a business operator that performs self-consignment. The power generation system is composed of solar cells (PV) and storage batteries. The solar cells are connected to the power grid via a DC-DC converter and a DC-AC converter. The same is true for the storage batteries. A power control device 1 controls the operation of the solar cells and storage batteries by sending operation commands to the solar cell drive device and the storage battery drive device (each composed of the above converter). There may be multiple solar cells and multiple storage batteries. The power control device 1 includes a calculation unit 11 that calculates operation commands for each drive device. A power interchange system can be configured using this power generation system and power control device 1.

[0014] FIG. 2 is a diagram illustrating a wheeling plan. Before carrying out self-wheeling (for example, by the day before), a utility submits a wheeling plan to the electric power management organization, describing the scheduled wheeling power in 30-minute increments, for example. The scheduled wheeling power is calculated, for example, by subtracting the demand forecast from the power generation forecast by the power generation system, and further multiplying it by a coefficient (≦1.0) as necessary. This coefficient is used to ensure that the power generation system has operating capacity.

[0015] If the actual generated power exceeds the planned power to be sent, the excess is balanced by suppressing the output power of the solar cells below the maximum power point and charging the storage battery.If the actual generated power is insufficient compared to the planned power to be sent, the shortage is made up by discharging from the storage battery.

[0016] When calculating the scheduled power to be wheeled, it is necessary to predict the power generated by solar cells and also to predict the power demand. The remainder, obtained by subtracting the demand forecast from the power generation forecast, is transmitted between base stations as the wheeled power. If the power generation forecast falls short of the demand forecast, the shortfall must be made up by discharging from the storage battery. Therefore, the storage battery must be maintained in a state of charge sufficient to ensure the scheduled power to be wheeled during the period in which wheeling is performed.

[0017] However, the amount of power generated by a solar cell varies depending on the weather, sometimes fluctuating significantly and sometimes remaining relatively stable. Therefore, in the present invention, solar radiation forecast data for the period during which the consignment is carried out is obtained from a weather information provider or the like, and this data is used to predict the weather during the period during which the consignment is carried out.

[0018] Figure 3 shows examples of weather codes provided by weather information providers. Weather codes are generally provided in broad categories such as 10 to 50 as shown in Figure 3. In the present invention, this is taken a step further and weather is classified in more detail. For example, sunny weather is classified into two detailed weather codes, 11 and 12.

[0019] Figure 4A is a graph comparing the measured solar radiation amount and predicted solar radiation data for a certain day. This graph shows an example of a day with a weather code of 10 (clear). In this example, it can be seen that the measured solar radiation amount and predicted solar radiation data match relatively well.

[0020] FIG. 4B is a graph comparing the measured solar radiation amount and predicted solar radiation data for another day. This graph shows an example of another day when the weather code is 10 (clear). In this example, the weather code is the same as in FIG. 4A, 10 (clear), but it can be seen that there is a large discrepancy between the measured solar radiation amount and the predicted solar radiation data. As such, even with the same weather code, the solar radiation amount can deviate significantly from the predicted data.

[0021] 5 is a schematic diagram showing the procedure by which the power control device 1 learns detailed weather classifications. The power control device 1 can learn the correlation between solar radiation forecast data and detailed weather classifications using a learning device constructed, for example, by a neural network. Here, an example is shown of a neural network in which the input layer receives solar radiation forecast values ​​for 12 hours, the intermediate layer (hidden layer) is made up of three neurons, and the output layer is made up of two neurons.

[0022] This neural network is constructed to further classify weather code 10 into two detailed weather categories 11 and 12. If the outputs from the two neurons in the output layer are "1" and "0", respectively, it means that the input solar radiation forecast value is for weather code 11. If the outputs from the two neurons in the output layer are "0" and "1", respectively, it means that the input solar radiation forecast value is for weather code 12.

[0023] The predicted solar radiation values ​​for 12 hours on the first day (DAY1) are input to the first neuron in the hidden layer. In this example, the correct detailed weather classification for DAY1 is weather code 11 (output layer "1" and "0"). The first neuron in the hidden layer has a weight matrix consisting of 12 elements that are multiplied by each of the 12 elements of the matrix describing the predicted solar radiation values. The result of multiplying both matrices is substituted into the sigmoid function represented by the first calculation formula in the bottom of Figure 5. The threshold of the first neuron in the hidden layer is specified as 15.02. Using the above procedure, the output y of the first neuron in the hidden layer is calculated.

[0024] The predicted solar radiation for 12 hours on the second day (DAY2) is input to the second neuron in the hidden layer. The same is true for the third and subsequent days, which are input to the corresponding neurons in the hidden layer. The output y of each hidden layer is calculated in the same way. For the sake of convenience, this example uses solar radiation prediction data up to the third day, but the same applies to the fourth and subsequent days.

[0025] The output y of the first neuron in the hidden layer is input to the first neuron in the output layer. The first neuron in the output layer has a weight matrix consisting of three elements that are multiplied by the outputs y of the first to third neurons in the hidden layer. The result of multiplying these two matrices is substituted into the sigmoid function represented by the second formula in the bottom of Figure 5. The threshold of the first neuron in the output layer is specified as 19.84. The output z of the first neuron in the output layer is calculated using the above procedure. The same process is carried out for the second neuron in the output layer.

[0026] From the above calculations, the output of the first neuron in the output layer is 1.00, and the output of the second neuron in the output layer is 0.33. On the other hand, since the correct weather code for DAY 1 is "1" and "0," the error Q between the output from the output layer and the correct answer can be calculated using the third formula in the bottom of Figure 5. The same calculation is performed for DAY 2 to DAY 3. Learning is completed by optimizing the matrix elements and thresholds of each layer so that the error Q is minimized. Neural network optimization methods such as backpropagation are well known, so a detailed explanation will be omitted.

[0027] FIG. 6 is a diagram explaining the procedure for estimating a detailed weather classification using the results of learning according to the procedure in FIG. 5. The power control device 1 inputs solar radiation forecast data to the neural network constructed according to FIG. 5. Using the same procedure as during learning, an output from each neuron in the output layer is obtained. For example, if the output from the first neuron in the output layer is greater than or equal to a threshold and the output from the second neuron is less than the threshold, the estimation result by the neural network is "1" or "0", i.e., weather code 11. This results in an estimation result for a detailed weather classification.

[0028] The power control device 1 learns in advance the correlation between the expected solar radiation and solar radiation forecast data for each detailed weather classification. The power control device 1 estimates the solar radiation on the solar cell using the correlation corresponding to the estimated detailed weather classification. The power control device 1 can predict the amount of power generated by the solar cell using the estimated solar radiation. The expected solar radiation can be estimated using the output current and output voltage of the solar cell. Any known technology can be used as the estimation method, but examples include those described in Patent Application Nos. 2020-127908 and 2021-128374.

[0029] FIG. 7 is a diagram illustrating the relationship between the output values ​​of hidden layer neurons in the neural network of FIG. 5 and the wheeling imbalance amount. The hidden layer neurons calculate output values ​​using a sigmoid function. The output values ​​are shown on the vertical axis of the upper graph of FIG. 7. The inventors have found that this sigmoid function value correlates with the range of fluctuation in solar radiation over a day. In the present invention, this correlation is utilized to estimate the range of fluctuation in solar radiation from the neural network of FIG. 5.

[0030] The larger the sigmoid function value of the hidden layer neuron, the larger the fluctuation range of the solar radiation on that day. The fluctuation range here is either (a) the largest absolute difference between the predicted solar radiation value and the actual solar radiation value at each time, or (b) the sum of the absolute differences between the predicted solar radiation value and the actual solar radiation value at each time for all times. However, since both have roughly similar trends and it is common for (b) to be large on days when (a) is large, either (a) or (b) can be used.

[0031] On days with large fluctuations in solar radiation, the difference between the power planned for transmission (planned transmission power) and the power actually transmitted (actual transmission power)—i.e., the transmission imbalance—is also large, and the time integral of this imbalance power is also large. Therefore, the sigmoid function value of the hidden layer neuron correlates with the fluctuation range of solar radiation on that day, and the fluctuation range of solar radiation is further correlated with the imbalance power amount. This makes it possible to estimate the imbalance power amount on that day using the sigmoid function value of the hidden layer neuron. The relationship between the sigmoid function value and the imbalance power amount can be set, for example, by previously associating the minimum value of the sigmoid function value (vertical axis 0 in the upper part of Figure 8) with the maximum value of the imbalance power amount (horizontal axis 2.0) based on actual values, etc.

[0032] Point (1) in the upper graph of Figure 7 is the sigmoid function value on a day with a moderate fluctuation in solar radiation. Therefore, the difference between the predicted solar cell power generation value (power generation plan) and the actual power generation value also fluctuates moderately from hour to hour. The middle graph of Figure 7 shows this situation. The imbalance power amount on this day is estimated to be 1.15 [kWh] from the horizontal axis of the graph. Point (2) in the upper graph of Figure 7 is the sigmoid function value on a day with a large fluctuation in solar radiation, and the bottom graph of Figure 7 shows the power generation plan and actual power generation on this day. Similarly, the imbalance power amount on this day is estimated to be 1.47 [kWh] from the horizontal axis of the graph.

[0033] We will now add a bit more about the fact that the output values ​​of hidden layer (intermediate layer) neurons are correlated with the range of fluctuations in solar radiation. For example, on clear days, the fluctuations in solar radiation are relatively small, while on sunny to cloudy days, the fluctuations in solar radiation are relatively large. When a weather code is determined based on the values ​​input from the intermediate layer to the output layer, the output from the intermediate layer is thought to reflect this fluctuation in solar radiation. In other words, in the process of training a neural network to estimate a weather code, the output values ​​of the intermediate layer are thought to be correlated with the range of fluctuations in solar radiation via the weather code. In this invention, we focused on this point and confirmed that the output of hidden layer neurons is actually configured in this way. Figure 7 is based on this idea.

[0034] The relationship between the sigmoid function, fluctuation range, and imbalance amount is thought to differ for each weather code. Therefore, a data table or the like that defines the relationship between these may be prepared in advance, and the insolation fluctuation range and imbalance amount may be estimated by referring to the data table. For example, a data table may be prepared in advance for each insolation fluctuation pattern (i.e., detailed weather code) such as DAY1 to DAY4 described below, and the insolation fluctuation range and imbalance amount may be estimated by referring to the data table corresponding to the estimated weather code. This data table may be used in combination with a neural network as an estimation model for the weather code and insolation fluctuation range.

[0035] 8A shows an example of predicted solar radiation values ​​for four days. Below, we will explain an example of predicting the range of solar radiation fluctuation for each day and ranking the power generation capacity of solar cells for each day based on the range of solar radiation fluctuation, using this predicted solar radiation value as an example.

[0036] FIG. 8B shows an example of estimating the range of fluctuations in solar radiation from the output of hidden layer neurons. The neural network described in FIGS. 5 and 6 outputs a detailed weather classification for each day (DAY 1 to DAY 4 in FIG. 8A). In addition, the hidden layer neurons of the same neural network represent the range of fluctuations in solar radiation for each day, as described in FIG. 7. The power control device 1 uses this to estimate the range of fluctuations in solar radiation for each of DAY 1 to DAY 4. Furthermore, it estimates the imbalance power amount from the range of fluctuations in solar radiation. In the example described in FIG. 7, the larger the sigmoid function value, the smaller the fluctuations in solar radiation. Therefore, even if the weather code is the same for a day, the larger the output value of the hidden layer neurons, the greater the fluctuations in solar radiation.

[0037] If the range of fluctuation in solar radiation is large, the power generation capacity of the solar cell will decrease accordingly (i.e., fluctuate greatly), and therefore the imbalance will also increase. In the present invention, the power generation capacity or imbalance of the solar cell is ranked on each day based on this. The higher the power generation capacity, the higher the rank, and the smaller the imbalance, the higher the rank. Even if the weather code is the same, if the range of fluctuation in solar radiation is different, the rank will also differ based on this. In the example of Figure 8B, the weather code for both DAY 3 and DAY 4 is clear 1, but the range of fluctuation in solar radiation is larger on DAY 4, so the rank for DAY 3 is higher (the smaller the rank number, the higher the rank). Similarly, the weather code for both DAY 1 and DAY 2 is clear 2, but the range of fluctuation in solar radiation is larger on DAY 2, so the rank for DAY 1 is higher. Hereinafter, this rank may also be referred to as the weather rank.

[0038] Figure 9A shows an example of the predicted and actual power generation amounts for a solar cell on DAY 3. DAY 3 has the highest rank and the smallest fluctuation range in solar radiation, as shown in Figure 8B. Therefore, the predicted and actual power generation amounts match well across all time periods.

[0039] 9B shows the result of the power control device 1 correcting the power generation forecast of FIG. 9A. The power control device 1 compares the power generation forecast with the actual power generation, and if the former is greater (i.e., if the actual power generation is insufficient), it compensates for the shortfall by discharging from the storage battery, and if the latter is greater (i.e., if the actual power generation is surplus), it absorbs the surplus by charging the storage battery. Therefore, the power generation forecast serves as a control parameter that indirectly controls the operation of the power interchange system. The power control device 1 can change the operation of the power interchange system by correcting the power generation forecast.

[0040] On the day with the weather code with the smallest fluctuation range in solar radiation (DAY3, among DAY1 to DAY4, is Sunny 1, with the smallest fluctuation range in solar radiation), it is desirable to charge the storage battery as much as possible in preparation for a possible power shortage the next day. Therefore, on DAY3, the power control device 1 generates control parameters so that the power interchange system operates in a direction that charges the storage battery as much as possible. Specifically, the predicted power generation amount is slightly lowered (in this example, by 3%) than the actual predicted value. As a result, when the predicted power generation amount is compared with the actual power generation amount, the latter tends to be higher, and the power interchange system will therefore have more control times to issue a charge command to the storage battery. Since it is desirable to maintain this operation throughout the entire day on DAY3, the power control device 1 lowers the predicted power generation amount by 3% than the actual predicted value throughout all time periods.

[0041] However, there may be a momentary shortage of power generation due to a sudden drop in solar radiation, so in such cases, it may be possible to temporarily allow the storage battery to discharge. Discharge 1 around 10:00 in Figures 9A and 9B shows this.

[0042] Even if the predicted power generation amount is lowered for all time periods, the power generation amount decreases while the power demand increases in the evening and later, so it is likely that the need to discharge from the storage battery will arise regardless of the predicted power generation amount. Therefore, in the example shown in Figures 9A and 9B, it should be noted that after 16:00, the storage battery is allowed to discharge regardless of whether the predicted power generation amount is corrected or not.

[0043] Figure 10A shows an example of the predicted and actual power generation amounts of solar cells on DAY 4. DAY 4 has the same weather code (sunny 1) as DAY 3, but the range of fluctuation in solar radiation is larger than on DAY 4, so it is ranked lower.

[0044] FIG. 10B shows the result of the power generation forecast in FIG. 10A corrected by the power control device 1. Generally, power demand tends to increase in the evening, while the amount of power generated by solar cells decreases, making it easier for shortage imbalances to occur in the evening. As a result, imbalance penalty fees tend to increase in the evening (i.e., revenue from surplus imbalances also tends to increase). Furthermore, since transmission companies generally try to minimize the discrepancy between the scheduled transmission power and the planned transmission power at the end of the day, electricity market prices tend to rise in the evening. Therefore, the power control device 1 controls the power generation system to charge the storage battery during an early time period of the day (first time period: morning in this example), and controls the power generation system to discharge the storage battery during a later time period (second time period: afternoon in this example).

[0045] Specifically, the predicted power generation amount is corrected to be 45% lower than the actual predicted value in the morning, and to be 45% higher than the actual predicted value in the afternoon. As a result, the power control device 1 controls the power interchange system so that the times at which a charge command is issued are more frequent in the morning than before the correction, and the times at which a discharge command is issued are more frequent in the afternoon than before the correction. The specific value of the correction amount (45% in this example) can be determined based on the range of fluctuation in solar radiation estimated using the output of hidden layer neurons, for example. For example, the correction amount increases as the range of fluctuation in solar radiation increases. The relationship between the range of fluctuation in solar radiation and the correction amount may be determined appropriately in advance, for example, based on actual results.

[0046] 11A shows an example of predicted and actual power generation amounts of a solar cell on DAY 1. The weather code on DAY 1 is clear 2, and the range of fluctuation in solar radiation is medium.

[0047] 11B shows the result of the power generation amount prediction of FIG. 11A being corrected by the power control device 1. As with DAY 4, the power generation amount prediction is corrected to be lower than the actual predicted value in the morning, and corrected to be higher than the actual predicted value in the afternoon. As with DAY 4, the correction amount is determined based on the estimated value of the fluctuation range of the amount of solar radiation.

[0048] FIG. 11C shows another example of correcting the power generation prediction on DAY1. Power generation is typically symmetric between morning and afternoon around noon, but depending on the weather code (and even the actual weather), this symmetry may be lost, resulting in biased power generation. In the example shown in FIG. 11C, the peak power generation occurs before 12:00. Therefore, to more reliably charge the storage battery, it is desirable to extend the charging time. Therefore, in the example shown in FIG. 11C, the power control device 1 extends the charging time compared to FIG. 11B and shortens the discharging time instead.

[0049] Whether or not to extend (or shorten) the charging time can be determined based on the predicted amount of power generation. As a general rule, charging occurs in the first half (morning) of the day, centered around 12:00, and discharging occurs in the second half (afternoon). If the predicted amount of power generation is expected to be biased, this rule can be changed and the charging time can be extended. For example, on days with a weather code where the difference between the predicted amount of power generation during the charging time period and the predicted amount of power generation during the discharging time period is expected to be equal to or greater than a threshold (days where the weather classification is asymmetric as described above), the greater the predicted amount of solar radiation (or predicted amount of power generation) is, the longer the charging time will be, ensuring that the storage battery is reliably charged.

[0050] When the charging time is extended, it is desirable not to lower the predicted power generation amount too much to avoid overcharging. Therefore, in Figure 11B, the predicted power generation amount during the charging time period is lowered by 20%, while in Figure 11C, it is lowered by 10%. Furthermore, since the discharging time period is shortened as the charging time period is extended, it is necessary to raise the predicted power generation amount during the discharging time period in order to increase the discharge amount per hour. Therefore, in Figure 11B, the predicted power generation amount during the discharging time period is raised by 20%, while in Figure 11C, it is raised by 25%.

[0051] 12A shows an example of predicted and actual power generation amounts of the solar cell on DAY 2. The weather code on DAY 2 is sunny 2, the same as on DAY 1, and the range of fluctuation in solar radiation is larger than on DAY 1.

[0052] 12B shows the result of the power generation amount prediction of FIG. 12A being corrected by the power control device 1. As with DAY 4, the power generation amount prediction is corrected to be lower than the actual predicted value in the morning, and to be higher than the actual predicted value in the afternoon. As with DAY 4, the correction amount is determined based on the estimated value of the fluctuation range of the amount of solar radiation.

[0053] Comparing DAY1 (Figure 11B) with DAY2 (Figure 12B), the correction amount is larger for DAY2. This is because the estimated range of insolation fluctuation is larger for DAY2 than for DAY1. In other words, even for the same weather code, the greater the range of insolation fluctuation, the greater the correction amount. However, because the patterns of insolation fluctuation are completely different between different weather codes, it may not be appropriate to simply increase the correction amount the greater the range of insolation fluctuation. Therefore, only when comparing between the same weather codes, the correction amount is increased the greater the range of insolation fluctuation.

[0054] 12C shows another example of correcting the predicted power generation amount on DAY 2. As with DAY 1, the amount of power generation is not symmetrical between the morning and afternoon, but is biased. Therefore, as with DAY 1, the power control device 1 extends the charging time length, does not reduce the predicted power generation amount during the charging time period too much, and increases the predicted power generation amount during the discharging time period.

[0055] FIG. 13 shows the relationship between the power generated by a solar cell and the imbalance charge. Generally, in the morning and evening hours when the amount of power generated is small, the shortage imbalance charge is high, and the selling price for the surplus imbalance also becomes high accordingly. However, if the consumer is unable to consume the surplus power transmitted to the power grid, a penalty charge may be temporarily imposed for the surplus imbalance. Since it is usually highly likely that surplus power can be sold in the evening hours, in the present invention, the power generation system is controlled so that the power can be sold by discharging it from the storage battery as early as possible in the evening. If there is a low possibility that a penalty will be imposed for the surplus power, the power may be sold in the early morning.

[0056] In addition to the above, the imbalance fee tends to increase as the difference between the predicted power generation amount on the day before the power transmission and the predicted power generation amount on that day (or the difference between the predicted power generation amount on the day before and the actual power generation amount on that day) increases. This tendency also appears in the ranking estimated using a neural network. In other words, if the rank on that day is lower than the rank on the previous day, the imbalance fee tends to increase. Therefore, in steps S1404 to S1405 (described later), it is determined whether the weather rank on the day is sufficiently lower than the weather rank on the previous day, and if it is sufficiently low, the power interchange system is controlled so as to sell as much electricity as possible.

[0057] 14A and 14B are flowcharts illustrating the procedure for correcting the power generation amount prediction by the power control device 1. This flowchart is executed by the calculation unit 11 on the day when the power interchange system is scheduled to transmit power to the demand base. Each step of this flowchart will be explained below.

[0058] (FIG. 14A: Step S1401) The calculation unit 11 acquires weather forecast data (including solar radiation forecast data) for the previous day for multiple power generation bases (preferably all power generation bases under its jurisdiction) under the power interchange system. The calculation unit 11 uses the acquired weather forecast data to determine the weather rank of each power generation base. The ranking criteria are as explained in FIG. 8. The calculation unit 11 calculates the weather rank for the previous day for the entire power interchange system by adding up (or averaging) the weather ranks of each power generation base.

[0059] (FIG. 14A: Step S1402) The calculation unit 11 acquires weather forecast data for the day at the same power generation base as in S1401. The calculation unit 11 obtains the weather rank for each power generation base using the same procedure as in S1401, and further obtains the weather rank for the day for the entire power interchange system.

[0060] (FIG. 14A: Step S1403) The calculation unit 11 determines whether the weather rank of the current day for the entire power interchange system, calculated in S1401-S1402, is lower than the weather rank of the previous day by a predetermined value or more. In the example of Fig. 8, the higher the rank number, the lower the rank, so if the weather rank number of the current day is higher than the weather rank number of the previous day by a predetermined value or more, the condition of this step is met (i.e., the weather rank of the current day is lower). If the condition is met, proceed to S1404, otherwise proceed to S1405.

[0061] (FIG. 14A: Steps S1404 to S1405) If the weather rank of the day is sufficiently lower than the weather rank of the previous day (i.e., S1403: Yes), the calculation unit 11 sets the forced power sale flag. Otherwise (S1403: No), the calculation unit 11 sets the forced power sale flag. The forced power sale flag is a control parameter that instructs the power interchange system to control so that power is sold unconditionally from a certain time period onwards (third time period: in this example, from 17:00 onwards). If the rank of the day is sufficiently lower than the rank of the previous day, the imbalance fee will be high (i.e., the power sale price will be high), as explained in FIG. 13, so on such days it is desirable to sell power in the evening or later if possible. The forced power sale flag can be used for this purpose.

[0062] (FIG. 14B: Step S1406) When the weather rank is 1 (DAY 3 described in FIGS. 9A and 9B), the calculation unit 11 reduces the power generation forecast by 3% from the actual forecast value over all time periods, as described in FIG. 9B. When the forced power sale flag is set in S1404, the power generation forecast is set to 0 so that power is sold unconditionally after 17:00. This makes it possible to control the power generation system in the direction of transmitting power to the demand base without relying on the correction amount of the power generation forecast. The same applies to S1407 to S1409.

[0063] (FIG. 14B: Step S1407) When the weather rank is 2 (DAY 4 described in Figures 10A and 10B), the calculation unit 11 reduces the morning power generation forecast by 45% from the actual predicted value, and increases the afternoon power generation forecast by 45% from the actual predicted value, as described in Figure 10B.

[0064] (FIG. 14B: Step S1408) When the weather rank is 3 (DAY 1 described in FIGS. 11A to 11C), the calculation unit 11 lowers the predicted amount of power generation in the morning from the actual predicted value and raises the predicted amount of power generation in the afternoon from the actual predicted value, as described in FIGS. 11B to 11C. Whether to use the correction amount shown in FIG. 11B or 11C can be determined based on whether the difference between the predicted amount of power generation in the charging period and the predicted amount of power generation in the discharging period is equal to or greater than a threshold. The same applies to S1409.

[0065] (FIG. 14B: Step S1409) When the weather rank is 4 (DAY 2 described in Figures 12A to 12C), the calculation unit 11 lowers the predicted power generation amount in the morning from the actual predicted value and raises the predicted power generation amount in the afternoon from the actual predicted value, as described in Figures 12B to 12C.

[0066] (FIG. 14B: Step S1410) If the weather rank is not one of 1 to 4, the amount of solar radiation is considered to be insufficient for wheeling, so wheeling is not carried out.

[0067] 15 is an example of a user interface provided by the calculation unit 11. The user interface can be presented, for example, on a display provided in the power control device 1. The user interface can present, for example, a power generation amount forecast and a corrected power generation amount forecast for each time period. Other appropriate parameters (for example, an estimated weather rank) may also be presented.

[0068] <Summary of the present invention> The power interchange system according to the present invention increases the power generation forecast higher than the actual forecast value during time periods when it is expected that there will be a large discrepancy between the forecasted and actual solar radiation values ​​(for example, the afternoon in FIG. 11B, after 14:00 in FIG. 11C, and after 15:00 in FIG. 12C), and decreases the power generation forecast lower than the actual forecast value during time periods before those. By controlling the power generation system based on this corrected power generation forecast, as explained in FIG. 9B, it is possible to increase the timing at which commands to transmit power to demand bases (i.e., to sell power) compared to when no correction is made. In other words, the system can be controlled to sell power as much as possible during time periods when it is advantageous to sell power.

[0069] The power interchange system according to the present invention fixes the charging and discharging operations of the storage battery as much as possible for each time period by determining in advance the time period for charging (first time period: for example, morning) and the time period for discharging (second time period: for example, afternoon). This makes it possible to assume that the capacity required for the storage battery will be sufficient within the range of the fixed operation. In other words, since the capacity of the storage battery can be kept within the assumed range, the installation cost of the storage battery can be reduced.

[0070] <Modifications of the present invention> The present invention is not limited to the above-described embodiments and includes various modifications. For example, the above-described embodiments have been described in detail to clearly explain the present invention, and are not necessarily limited to those including all of the described configurations. Furthermore, it is possible to replace part of the configuration of one embodiment with the configuration of another embodiment, or to add the configuration of another embodiment to the configuration of one embodiment. Furthermore, it is possible to add, delete, or replace part of the configuration of each embodiment with other configurations.

[0071] In the above embodiment, the correction amount for the predicted power generation amount (for example, 3% in FIG. 9B) is just an example, and it should be noted that the calculation unit 11 can determine these correction amounts based on the range of fluctuation in the amount of solar radiation.

[0072] In the above embodiment, even if the forced power selling flag is set, if the state of charge of the storage battery is not sufficient (that is, if it is below the threshold), discharging may be prohibited.

[0073] In the above embodiment, it has been described that the timing for issuing a command to charge or discharge a storage battery is adjusted by correcting the power generation forecast. Increasing or decreasing the power generation forecast not only affects the operation of the storage battery, but also adjusts the output power of the solar cell. The solar cell output is determined based on the sum of the scheduled transmission power and the power demand. Since the scheduled transmission power is determined based on the power generation forecast, if the power generation forecast is increased and the solar cell is controlled based on this, a control command to increase the solar cell output will be executed. Therefore, in the above embodiment, when the power generation forecast is corrected, as long as the power that the power interchange system can transmit to the demand base can be secured according to the amount of correction, the power may be secured by either the storage battery or the solar cell (or a combination of both).

[0074] In the above embodiments, extending the charging time length in FIG. 11C or FIG. 12C has the following significance. The power transmission network operator may request that the power interchange operator cover the power transmitted to the power transmission network with power generated by the power interchange operator itself. This is because, depending on the electricity market price, retransmitting power received from the power transmission network may be disadvantageous to the power transmission network operator. If this condition cannot be met, there is a possibility that use of the power transmission network will not be permitted. The significance of extending the charging time length is to ensure that this condition is met by reliably charging the storage battery.

[0075] In the above embodiment, an example has been described in which the output value of the second neuron among the hidden layer neurons 1 to 3 shown in Fig. 8B is used to estimate the fluctuation range of solar radiation, but which hidden layer neuron to use can be designed as appropriate. Furthermore, the output values ​​of the output layer neurons may be appropriately combined as necessary to estimate the fluctuation range of solar radiation.

[0076] In the above embodiments, the calculation unit 11 can be configured by hardware such as a circuit device that implements its functions, or can be configured by a calculation device such as a CPU (Central Processing Unit) executing software that implements its functions.

[0077] In the above embodiment, a sigmoid function is used as the output function of the hidden layer neurons (internal variables), but other similar functions may be used. For example, it is possible to use a function other than the sigmoid function, which has a nonlinear relationship between input and output and in which the output increases monotonically as the input increases.

[0078] In the above embodiment, a neural network is used as an example of an estimation model for the weather code and the fluctuation range of solar radiation, but other estimation models may also be used. However, the estimation model must be configured to output the weather code and have an internal variable that correlates with the fluctuation range of solar radiation (and the imbalance amount). As long as the estimation model is configured in this way, estimation models other than a neural network can be used. [Explanation of symbols]

[0079] 1: Power control device 11: Arithmetic section

Claims

1. A power interchange system that transmits power generated by a power generation system composed of solar cells and storage batteries between bases via a power transmission network, a calculation unit that specifies a first time period during which the storage battery is charged using power generated by the solar cell and a second time period during which the storage battery is discharged; The calculation unit identifies time periods in which a deviation between a predicted value and an actual value of the amount of solar radiation is larger in the second time period than in the first time period, as the first time period and the second time period, respectively; the calculation unit determines a first correction value that corrects the predicted amount of power generation by the solar cell in the first time zone in a decreasing direction and a second correction value that corrects the predicted amount of power generation by the solar cell in the second time zone in an increasing direction, The calculation unit corrects a predicted amount of power generation by the solar cell using the first correction value and the second correction value, thereby controlling the power generation system to execute a command to charge the storage battery with power generated by the solar cell during the first time period, and controls the power generation system to execute a command to transmit power to the power transmission network by discharging the storage battery during the second time period. The power interchange system is characterized by:

2. the calculation unit estimates a weather classification when the power generation system generates power by using an estimation model that estimates a weather classification by receiving a predicted solar radiation value as an input; The calculation unit ranks the weather classifications estimated using the estimation model in order of power generation capacity of the solar cell; the calculation unit, on the day with the highest rank, regards all time periods as the first time periods and regards the second time periods as not existing, and corrects the power generation amount prediction by the solar cell; The calculation unit adopts the first correction value and the second correction value, which are different for each rank, on a day on which the rank is second or higher.

2. The power interchange system according to claim 1 .

3. The estimation model is configured to have, as an internal variable, a variable correlated with a fluctuation range of the amount of solar radiation in one day, The calculation unit estimates the fluctuation range using the estimation model, The calculation unit determines a magnitude of the first correction value and a magnitude of the second correction value in response to a magnitude of the estimated fluctuation range.

3. The power interchange system according to claim 2.

4. the calculation unit identifies a first estimated day on which the fluctuation range is a first fluctuation range and a second estimated day on which the fluctuation range is larger than the first fluctuation range, among a plurality of days having the same weather classification estimated using the estimation model; The calculation unit makes the first correction value on the second estimated date larger than the first correction value on the first estimated date, The calculation unit sets the second correction value on the second estimated date to be larger than the second correction value on the first estimated date.

4. The power interchange system according to claim 3.

5. The calculation unit identifies an asymmetric day on which a difference between the power generation amount prediction in the first time slot and the power generation amount prediction in the second time slot is expected to be equal to or greater than a threshold; The calculation unit adjusts the length of the first time period on the asymmetric day such that the first time period is extended longer as the predicted solar radiation value in the first time period is larger.

4. The power interchange system according to claim 3.

6. the calculation unit estimates a weather classification when the power generation system generates power by using an estimation model that estimates a weather classification by receiving a predicted solar radiation value as an input; The calculation unit ranks the weather classifications estimated using the estimation model in order of power generation capacity of the solar cell; When an absolute difference between the rank on a scheduled power transmission date when power is scheduled to be transmitted to the power transmission network and the rank on a day before the scheduled power transmission date exceeds a threshold value and the rank on the scheduled power transmission date is lower than the rank on the day before, the calculation unit controls the power generation system to transmit power to the power transmission network from a third time slot on the scheduled power transmission date onward.

2. The power interchange system according to claim 1 .

7. The calculation unit corrects a predicted amount of power generation by the solar cell after the third time slot to zero, thereby controlling the power generation system so as to transmit power to the power transmission network after the third time slot without relying on the first correction value and the second correction value.

7. The power interchange system according to claim 6.

8. The calculation unit allows the storage battery to discharge when the state of charge of the storage battery is equal to or higher than a threshold value after the third time period, thereby enabling the storage battery to cover the power to be transmitted to the power transmission network; The calculation unit prohibits discharging from the storage battery regardless of the rank on the scheduled power transmission date and the rank on the previous day when the state of charge of the storage battery is less than the threshold value after the third time period.

7. The power interchange system according to claim 6.

9. The estimation model is configured to have, as an internal variable, a variable correlated with a fluctuation range of the amount of solar radiation in one day, The calculation unit determines the rank based on the fluctuation range represented by the internal variable.

3. The power interchange system according to claim 2.

10. The estimation model is an input layer for receiving the solar radiation forecast value; an output layer that outputs the weather classification; an intermediate layer disposed between the output layers; Equipped with The internal variables are configured as variables that the intermediate layer has, The calculation unit determines the rank by using the internal variables of the intermediate layer.

10. The power interchange system according to claim 9.

11. The first time period is a time period that occurs in a day before the second time period.

2. The power interchange system according to claim 1 .

12. The third time period is a time period in a day that is earlier than the first time period and the second time period, or a time period in a day that is later than the first time period and the second time period.

7. The power interchange system according to claim 6.

13. The calculation unit outputs a user interface; The user interface includes: A prediction of the amount of power generated by the solar cell; a corrected power generation prediction obtained by correcting the power generation prediction by the solar cell using the first correction value and the second correction value; is configured to present at least one of 2. The power interchange system according to claim 1 .

14. The power generation system includes: A first solar cell provided at the first power generation site; A second solar cell provided at the second power generation site; It is composed of The calculation unit ranks the power generation capabilities of the solar cells in the entire power generation system based on a sum of the rank at the first power generation base and the rank at the second power generation base.

3. The power interchange system according to claim 2.

Images