JP2016091312A - Fault diagnosis apparatus and fault diagnosis method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a fault diagnosis apparatus configured to accurately diagnose a fault in a power generation facility in a short learning period, and a fault diagnosis method.SOLUTION: A weighting coefficient calculation section 8 includes a prediction model generation section 10, a prediction value calculation section 12, and a fault diagnosis section 14. The weighting coefficient calculation section 8 calculates a weighting coefficient indicating a relationship between a distribution of a weather parameter in a learning period and a distribution of a weather parameter in a diagnosis period. A prediction model generation section 10 generates a prediction model of the quantity of power to be generated in the diagnosis period, on the basis of the weather parameter in the learning period, the quantity of power generated in a power generation facility in the learning period, and the weighting coefficient. A prediction value calculation section 12 calculates a prediction value of the quantity of power generated in the diagnosis period, on the basis of the prediction model and the weather parameter in the diagnosis period. The fault diagnosis section 14 compares the prediction value of the quantity of power generated in the diagnosis period with the quantity of power generated in the diagnosis period, to determine a fault of the power generation facility.SELECTED DRAWING: Figure 1

Description

  Embodiments described herein relate generally to an abnormality diagnosis apparatus and a power generation facility.

  In recent years, power generation facilities that generate power using renewable energy have become widespread. In such an abnormality diagnosis of a power generation facility, it is important to consider the weather condition during the diagnosis period. This is because when using renewable energy such as sunlight, the amount of power generated by the power generation facility changes according to the weather conditions.

  Conventionally, as a method for diagnosing abnormalities in such power generation facilities, a prediction model of power generation amount according to weather conditions is created, and the predicted value of power generation amount in the diagnosis period predicted by the prediction model and measurement of power generation amount in the diagnosis period A method for comparing values with each other has been proposed. The prediction model is created based on the meteorological state and the measurement value of the power generation amount during a predetermined learning period.

  However, the above-described abnormality diagnosis method has a problem that the prediction accuracy of the power generation amount decreases when the learning period is short. This is because the weather condition has seasonality, and therefore, if the learning period is short, the prediction model is specialized in the weather condition of the learning period. For example, a prediction model created based on summer weather conditions is specialized in predicting summer power generation, and cannot predict winter power generation accurately. For this reason, the conventional abnormality diagnosis method requires long-term meteorological conditions and measured values of the power generation amount in order to improve the accuracy of abnormality diagnosis.

JP2013-191672A JP 2011-142790 A

  An abnormality diagnosis device and an abnormality diagnosis method are provided that can accurately diagnose an abnormality in a power generation facility in a short learning period.

  An abnormality diagnosis apparatus according to an embodiment includes a weighting factor calculation unit, a prediction model creation unit, a prediction value calculation unit, and an abnormality diagnosis unit. The weighting factor calculation unit calculates a weighting factor indicating the relationship between the weather parameter distribution during the learning period and the weather parameter distribution during the diagnosis period. The prediction model creation unit creates a prediction model of the power generation amount during the diagnosis period based on the weather parameters during the learning period, the power generation amount of the power generation facility during the learning period, and the weighting factor. The predicted value calculation unit calculates a predicted value of the power generation amount during the diagnosis period based on the prediction model and the weather parameters during the diagnosis period. The abnormality diagnosis unit compares the predicted value of the power generation amount during the diagnosis period with the power generation amount during the diagnosis period to diagnose an abnormality in the power generation facility.

The block diagram which shows the function structure of an abnormality diagnosis apparatus. The figure which shows the preparation method of a prediction model. The figure which shows the hardware constitutions of an abnormality diagnosis apparatus. The flowchart which shows operation | movement of an abnormality diagnosis apparatus. The figure which shows an example of a learning period and a diagnosis period. The figure which shows an example of the weather data memorize | stored in the weather data storage part. The figure which shows an example of the electric power generation data memorize | stored in the electric power generation data storage part. The figure which shows an example of the weather data of a learning period. The figure which shows an example of the weather data of a diagnostic period. The figure which shows an example of the electric power generation data of a learning period. The figure which shows an example of the electric power generation data of a diagnostic period. The figure which shows an example of the weighting coefficient memorize | stored in the weighting coefficient memory | storage part. The figure which shows an example of the weather data of a learning period, electric power generation data, and a weighting coefficient. The figure which shows an example of the parameter of a prediction model. The figure which shows an example of the predicted value of electric power generation amount. The figure which shows an example of the abnormality diagnosis result of power generation equipment.

  Embodiments of the present invention will be described below with reference to the drawings.

  The abnormality diagnosis apparatus and abnormality diagnosis method according to the present embodiment are used for abnormality diagnosis of power generation equipment that generates power using renewable energy whose power generation amount changes according to weather conditions. The power generation facilities to be diagnosed are, for example, a solar power generation facility, a wind power generation facility, and a solar thermal power generation facility, but are not limited thereto.

  First, the functional configuration of the abnormality diagnosis apparatus will be described with reference to FIG. FIG. 1 is a block diagram showing a functional configuration of the abnormality diagnosis apparatus. As shown in FIG. 1, this abnormality diagnosis apparatus includes a weather data storage unit 1, a power generation data storage unit 2, a learning period weather data extraction unit 3, a diagnosis period weather data extraction unit 4, and a learning period power generation data extraction. Unit 5, diagnosis period power generation data extraction unit 6, data density calculation unit 7, weighting factor calculation unit 8, weighting factor storage unit 9, prediction model creation unit 10, prediction model storage unit 11, prediction value A calculation unit 12, a predicted value storage unit 13, an abnormality diagnosis unit 14, and a diagnosis result storage unit 15 are provided.

  The weather data storage unit 1 stores weather data. The meteorological data is historical data of measured values of meteorological parameters indicating the weather conditions. The weather parameters are, for example, temperature, solar radiation amount, solar altitude, solar orientation, wind speed, wind force, and wind direction, but are not limited thereto. The weather data includes historical data of measured values of one or more weather parameters.

  The weather data is input to the abnormality diagnosis device from an external device such as a weather prediction device or a power generation facility, and is stored in the weather data storage unit 1. The abnormality diagnosis apparatus may be configured not to include the weather data storage unit 1.

  The power generation data storage unit 2 stores power generation data. The power generation data is historical data of measured values of the power generation amount of the power generation facility for which the abnormality diagnosis device diagnoses an abnormality. The amount of power generation is, for example, the amount of power generated by the power generation facility, DC voltage, DC current, AC voltage, AC current, DC power, and AC power, but is not limited thereto. The power generation data includes historical data of one or more power generation measurement values.

  The power generation data is automatically extracted by, for example, a data collection system, or when the user operates an operation terminal of the power generation facility, the power generation data is input from the power generation facility to the abnormality diagnosis device and stored in the power generation data storage unit 2. The abnormality diagnosis device may be configured not to include the power generation data storage unit 2.

  The learning period weather data extraction unit 3 (hereinafter referred to as “extraction unit 3”) extracts the weather data of the learning period from the weather data storage unit 1. The learning period is a period of meteorological data and power generation data used to create a prediction model described later. The learning period may be set in advance or may be input by the user. The learning period is usually selected from periods during which the power generation facility is considered to be operating normally. When the abnormality diagnosis device does not include the weather data storage unit 1, the extraction unit 3 communicates with an external device that stores the weather data and extracts the weather data during the learning period.

  The diagnosis period weather data extraction unit 4 (hereinafter referred to as “extraction unit 4”) extracts the weather data of the diagnosis period from the weather data storage unit 1. The diagnosis period is a period for which an abnormality of the power generation facility is diagnosed by the abnormality diagnosis device. The diagnosis period may be set in advance or may be input by the user. When the abnormality diagnosis device does not include the weather data storage unit 1, the extraction unit 4 communicates with an external device that stores the weather data and extracts weather data for the diagnosis period.

  The learning period power generation data extraction unit 5 (hereinafter referred to as “extraction unit 5”) extracts the power generation data of the learning period from the power generation data storage unit 2. When the abnormality diagnosis device does not include the power generation data storage unit 2, the extraction unit 5 communicates with the external device that stores the power generation data and extracts the power generation data in the learning period.

  The diagnosis period power generation data extraction unit 6 (hereinafter referred to as “extraction unit 6”) extracts the power generation data of the diagnosis period from the weather data storage unit 1. When the abnormality diagnosis device does not include the power generation data storage unit 2, the extraction unit 6 communicates with the external device that stores the power generation data and extracts the power generation data in the diagnosis period.

The data density calculation unit 7 calculates a first function P _train (x) indicating the distribution of the weather parameters in the learning period based on the weather data in the learning period extracted by the extraction unit 3. x is a weather parameter. The first function P _train (x) is, for example, a probability density function of weather parameters during the learning period, but is not limited thereto. The first function P _train (x) may be a non-normalized function such as N times the probability density function (N is an arbitrary real number).

Further, the data density calculation unit 7 calculates a second function P _test (x) indicating the distribution of the weather parameters in the diagnosis period based on the weather data in the diagnosis period extracted by the extraction unit 4. x is a weather parameter. The second function P _test (x) is, for example, a probability density function of weather parameters during the diagnosis period, but is not limited thereto. The second function P _test (x) may be a non-normalized function such as N times the probability density function (N is an arbitrary real number).

Based on the first function P _train (x) and the second function P _test (x) calculated by the data density calculation unit 7, the weighting factor calculation unit 8 calculates a weighting factor for each data in the learning period. The weighting coefficient calculation unit 8 first calculates a weighting function based on the first function P _train (x) and the second function P _test (x). The weight function is a function indicating the relationship between the distribution of the weather parameters during the learning period and the distribution of the weather parameters during the diagnosis period. The weight function is, for example, the ratio P _test (x) / P _train (x) (probability density ratio function) or the difference P between the probability density function of the weather parameter in the learning period and the probability density function of the weather parameter in the diagnosis period _test (x) -P _train (x), but not limited to this.

The weighting factor calculation unit 8 calculates a weighting factor in each data of the learning period based on the weighting function and the weather data of the learning period extracted by the extraction unit 3. For example, when the weight function is P _test (x) / P _train (x), the weight coefficient of the data of the weather parameter measurement value xi is P _test (xi) / P _train (xi).

  The weight coefficient storage unit 9 stores the weight coefficient of each data in the learning period calculated by the weight coefficient calculation unit 8.

  The prediction model creation unit 10 performs prediction based on the weather data of the learning period extracted by the extraction unit 3, the power generation data of the learning period extracted by the extraction unit 5, and the weighting factor stored in the weighting factor storage unit 9. Create a model. The prediction model is a model for predicting the power generation amount of the power generation facility based on the weather parameter, and is a regression equation with the power generation amount as an objective variable and the weather parameter as an explanatory variable. The prediction model creation unit 10 creates a prediction model by calculating the parameters of this regression equation. The regression equation parameters are calculated by, for example, a weighted least square method using weighting factors as weights, but the present invention is not limited thereto. The regression equation may be set in advance or may be input by the user of the abnormality diagnosis apparatus.

  Here, FIG. 2 is a diagram illustrating a method for creating a prediction model. In FIG. 2, the abscissa represents the weather parameter (irradiation amount or temperature), the ordinate represents the power generation amount, the value indicated by ◯ is the weather data during the learning period, and the value indicated by x is the weather data during the diagnosis period. In FIG. 2, the distribution of the meteorological data in the learning period is different from the distribution of the meteorological data in the diagnosis period.

  When a prediction model is created based on meteorological data and power generation data during a learning period as in a conventional abnormality diagnosis device, the prediction model is specialized for the distribution of meteorological data during the learning period. In such a prediction model, as shown in FIG. 2, it is impossible to accurately predict the power generation amount in the diagnosis period with different distributions of weather data.

  However, in the abnormality diagnosis apparatus according to the present embodiment, the distribution of the weather data during the learning period can be made similar to the distribution of the weather data during the diagnosis period using a weighting factor. For this reason, the prediction model according to the present embodiment corresponds to the distribution of meteorological data during the diagnosis period, and can accurately predict the power generation amount during the diagnosis period.

  The prediction model storage unit 11 stores the prediction model created by the prediction model creation unit 10. That is, the prediction model storage unit 11 stores the regression equation whose parameters are calculated by the prediction model creation unit 10.

  The predicted value calculation unit 12 calculates a predicted value of the power generation amount in the diagnostic period based on the weather data in the diagnostic period extracted by the extraction unit 4 and the predicted model stored in the predicted model storage unit 11. The predicted value of the power generation amount is calculated for each date and time of the diagnosis period. The predicted value of the power generation amount can be calculated by substituting the measured value of the weather parameter into the prediction model.

  The predicted value storage unit 13 stores the predicted power generation amount for the diagnosis period calculated by the predicted value calculation unit 12.

  The abnormality diagnosis unit 14 compares the power generation data of the diagnosis period extracted by the extraction unit 6 with the predicted value of the power generation amount of the diagnosis period stored in the prediction value storage unit 13, thereby determining the abnormality of the power generation facility. Diagnose. When the power generation facility is normal, the measured value and the predicted value of the power generation amount during the diagnosis period are close to each other. Therefore, the abnormality diagnosis unit 14 diagnoses the power generation facility as abnormal when the measured value and the predicted value of the power generation amount during the diagnosis period are different. The abnormality diagnosing unit 14 diagnoses an abnormality by, for example, comparing a predetermined threshold with an average ratio, an average difference, a standard deviation, and the like of the measured value of the power generation amount during the diagnosis period and the predicted value.

  The diagnosis result storage unit 15 stores the diagnosis result of the abnormality of the power generation facility by the abnormality diagnosis unit 14.

  As described above, this abnormality diagnosis apparatus and abnormality diagnosis method can create a prediction model corresponding to the distribution of weather data in the diagnosis period with different weather conditions from the weather data in the learning period by using the weight coefficient. it can. According to this prediction model, it is possible to suppress the influence due to the seasonality of the weather condition, that is, the fluctuation of the weather condition between the learning period and the diagnosis period, and accurately predict the power generation amount in the diagnosis period. Therefore, the abnormality diagnosis device can accurately diagnose abnormality of the power generation facility.

  In addition, since this abnormality diagnosis device can create a highly accurate prediction model even when the learning period is short, it can accurately diagnose abnormality of the power generation equipment. Therefore, according to this abnormality diagnosis device, abnormality diagnosis of the power generation facility can be performed even within one year from the start of operation of the power generation facility, and an initial failure of the power generation facility can be found.

  Next, the hardware configuration of the abnormality diagnosis apparatus will be described. FIG. 3 is a diagram illustrating a hardware configuration of the abnormality diagnosis apparatus. Each functional configuration of the above-described abnormality diagnosis device is realized by causing a general-purpose computer device to execute an abnormality diagnosis program. As shown in FIG. 3, the computer device includes a CPU (Central Processing Unit) 101, a RAM (Random Access Memory) 102, an HDD (Hard Disk Drive) 103, a graphic processing device 104, an input interface 105, and a communication interface 106. Prepare.

  The CPU 101 operates according to a program stored in the HDD 103 or ROM (not shown). When the CPU 101 executes the abnormality diagnosis program, each functional configuration of the abnormality diagnosis apparatus is realized.

  The RAM 102 stores data necessary for the CPU 101 to execute various processes as necessary. The weather data storage unit 1, the power generation data storage unit 2, the weight coefficient storage unit 9, the prediction model storage unit 11, the predicted value storage unit 13, and the diagnosis result storage unit 15 can be configured on the RAM 102.

  The HDD 103 stores programs executed by the CPU 101 and data necessary for the CPU 101 to execute various processes. The weather data storage unit 1, the power generation data storage unit 2, the weight coefficient storage unit 9, the prediction model storage unit 11, the predicted value storage unit 13, and the diagnosis result storage unit 15 can be configured on the HDD 103.

  The graphic processing device 104 causes the display 108 to display an image corresponding to the image data given from the CPU 101. The display 108 is, for example, a liquid crystal display, a plasma display, and a cathode ray tube display, but is not limited thereto. The graphic processing device 104 can display the diagnosis result stored in the diagnosis result storage unit 15 and various input screens for the abnormality diagnosis device on the display 108.

  The input interface 105 transmits a signal corresponding to the operation of the connected input device to the CPU 101. The input device is, for example, the keyboard 109 and the mouse 110, but is not limited thereto. The learning period and the diagnosis period are input to the abnormality diagnosis device via the input interface 105 when the user operates the input device, for example.

  The communication interface 106 accesses an external device via the LAN 107. Communication between the abnormality diagnosis device and the power generation facility can be realized by the communication interface 106. Information obtained via the communication interface 106 is temporarily stored in the RAM 102.

  Next, the operation of the abnormality diagnosis apparatus will be described in detail using a specific example. In the following description, it is assumed that the power generation facility to be diagnosed is a solar power generation facility. Here, FIG. 4 is a flowchart showing the operation of the abnormality diagnosis apparatus.

  First, the user of the abnormality diagnosis apparatus inputs a learning period and a diagnosis period, and starts abnormality diagnosis processing. FIG. 5 is a diagram illustrating an example of a learning period and a diagnosis period. In FIG. 5, the learning period is a period from 2014/8/1 to 2014/8/31, and the diagnostic period is a period from 2014/9/1 to 2014/9/30. Here, it is assumed that the learning period and the diagnosis period in FIG. 5 are input by the user. The abnormality diagnosis apparatus starts the abnormality diagnosis process after inputting the learning period and the diagnosis period.

  In step S <b> 1, first, weather data and power generation data stored in the weather data storage unit 1 and the power generation data storage unit 2 on the HDD 103 are read into the RAM 102.

FIG. 6 is a diagram illustrating an example of weather data stored in the weather data storage unit 1. As shown in FIG. 6, this meteorological data is composed of hourly records from 9:00:00 on 2014/8/1 to 19:00 on 2014/9/30. Each record stores measurement values of date and weather parameters. There are three weather parameters: outside air temperature, solar radiation, and solar altitude. According to the meteorological data of FIG. 6, for example, the outdoor temperature at 9:00:00 on 2014/8/1 is 28.1 ° C., the amount of solar radiation is 681 W / m ² , and the solar altitude is 65 degrees. Here, it is assumed that the weather data shown in FIG. 6 has been read.

  FIG. 7 is a diagram illustrating an example of the power generation data stored in the power generation data storage unit 2. As shown in FIG. 7, this power generation data is composed of hourly records from 9:00:00 on 2014/8/1 to 19:00 on 2014/9/30. Each record stores measurement values of date and power generation amount. There are four power generation amounts: DC current (A), DC voltage (V), DC power (kW), and AC power (kW). According to the power generation data of FIG. 7, for example, the DC current at 9:00:00 of 2014/8/1 is 250 A, the DC voltage is 620 V, the DC power is 155.0 kW, and the AC power is 150.4 kW. Here, it is assumed that the power generation data shown in FIG. 7 has been read.

  In step S <b> 2, the extraction units 3 to 6 extract the learning period and diagnosis period data from the read weather data and power generation data.

  The extraction unit 3 extracts the weather data from 2014/8/1 to 2014/8/31, which is the learning period, from the weather data of FIG. Thereby, the weather data shown in FIG. 8 is extracted.

  The extraction unit 4 extracts weather data from 2014/9/1 to 2014/9/30, which is a diagnosis period, from the weather data of FIG. As a result, the weather data shown in FIG. 9 is extracted.

  The extraction unit 5 extracts the power generation data from 2014/8/1 to 2014/8/31, which is the learning period, from the power generation data of FIG. Thereby, the power generation data shown in FIG. 10 is extracted.

  The extraction unit 6 extracts power generation data from 2014/9/1 to 2014/9/30, which is a diagnosis period, from the power generation data of FIG. Thereby, the power generation data shown in FIG. 11 is extracted.

In step S3, the data density calculation unit 7 calculates the probability density function P _train (x) of the weather parameters in the learning period based on the weather data in the learning period shown in FIG. x is the outside temperature, solar radiation, and solar altitude. Therefore, P _train (x) is a three-dimensional probability density function.

Further, the data density calculation unit 7 calculates the probability density function P _test (x) of the weather parameter in the diagnosis period based on the weather data in the diagnosis period shown in FIG. x is the outside temperature, solar radiation, and solar altitude. Therefore, P _test (x) is a three-dimensional probability density function.

In step S4, the weight coefficient calculation unit 8 calculates a probability density ratio function P _test (x) / P _train (x) as a weight function. At this time, if the weather data in the learning period is xi, the weighted average is expressed by the following equation. Here, it can be seen that the weighted average using the probability density ratio on the left side converges to an expected value based on the probability distribution of the diagnosis period shown on the right side as the number of data n increases. The function A indicates a prediction model error or the like.

The weighting factor calculation unit 8 substitutes the weather parameter of each data into the probability density ratio function P _test (x) / P _train (x), and calculates the probability density ratio of each data. The calculated weighting coefficient of each data is stored in the weighting coefficient storage unit 9.

  FIG. 12 is a diagram illustrating weighting factors stored in the weighting factor storage unit 9. As shown in FIG. 12, in the weighting coefficient storage unit 9, the hourly records storing the date and the weighting coefficient are from 9: 00: 00: 00 on 2014/8/1 to 19:00 on 2014/9/30. : Memorized until 0:00. According to FIG. 12, for example, the weighting factor of 9: 00: 00: 00 of 2014/8/1 is 0.782.

  In step S <b> 5, the prediction model creation unit 10 calculates a regression equation parameter and creates a prediction model. FIG. 13 is a diagram illustrating the weather data, power generation data, and weighting factors in the learning period input to the prediction model creation unit 10 for parameter calculation. In FIG. 13, meteorological data, power generation data, and weighting factors for the same date and time are combined in the same record. The prediction model creation unit 10 calculates the following regression equation parameters based on the data shown in FIG. 12 by a weighted least square method using weighting factors as weights.

  In this regression equation, P is DC power, G is solar radiation, T is temperature, and A, B, C, and D are parameters. FIG. 14 is a diagram showing the parameters of the regression equation calculated by the prediction model creation unit 10. In FIG. 14, A is 313.8468, B is 0.7329, C is -37.127, and D is -13.528. Thus, the following prediction model is created by calculating the parameters of the regression equation.

  The created prediction model is stored in the prediction model storage unit 11. The regression equation is not limited to the above equation, and can be set arbitrarily.

  In step S <b> 6, the predicted value calculation unit 12 substitutes the weather parameters of the weather data of the diagnosis period shown in FIG. 9 into the prediction model stored in the prediction model storage unit 11, and calculates the power generation amount at each date and time of the diagnosis period. Calculate the predicted value. For example, the predicted value of DC power at 9:00:00 on 2014/9/1 is calculated by substituting the solar radiation amount G and temperature T at 9:00:00 on 2014/9/1 into the prediction model. The The predicted value of the power generation amount calculated by the predicted value calculation unit 12 is stored in the predicted value storage unit 13. FIG. 15 is a diagram illustrating the predicted value of DC power stored in the predicted value storage unit 13. In FIG. 15, the predicted value of DC power at 9:00:00 on 2014/9/1 is 142 kW.

  In step S7, the abnormality diagnosis unit 14 diagnoses an abnormality in the power generation facility. First, the abnormality diagnosis unit 14 compares the predicted value of the power generation amount at each date and time in the diagnosis period shown in FIG. 15 with the measured value of the power generation amount at each date and time in the diagnosis period shown in FIG. ) Is calculated. The Rating value is a measured value / predicted value of the power generation amount. According to FIG.11 and FIG.15, since the measured value of DC power of 9: 00: 00: 00 of 2014/9/1 is 137.0 kW and a predicted value is 142 kW, Rating value is 137/142 = 96.5%. It becomes.

  Next, the abnormality diagnosing unit 14 calculates an average value of Rating values at each date and time, compares the average value with a threshold value, and diagnoses an abnormality in the power generation facility. FIG. 16 is a diagram illustrating a diagnosis result by the abnormality diagnosis unit 14. In FIG. 16, the average value of Rating values is 96.6%, and the threshold value is 95%. The abnormality diagnosis unit 14 diagnoses normal if the average value> threshold, and diagnoses abnormality if the average value ≦ threshold. In FIG. 16, since 96.6%> 95%, the power generation facility is diagnosed as normal. The diagnosis result by the abnormality diagnosis unit 14 is stored in the diagnosis result storage unit 15.

  The abnormality diagnosing unit 14 may diagnose an abnormality using a median value instead of an average value of the rating values, or may diagnose an abnormality using a rating value in a part of a time zone. Moreover, you may diagnose abnormality using the standard deviation of the Rating value shown in FIG.

  In step S8, the diagnosis result stored in the diagnosis result storage unit 15 is output. The diagnosis result is output, for example, by causing the graphic processing device 104 to display the diagnosis result shown in FIG.

  With the above operation, the abnormality diagnosis device creates a prediction model for the power generation amount in September (diagnosis period) based on the weather parameters and power generation amount in August (learning period), and diagnoses the power generation facility or more. Can do.

  Note that the present invention is not limited to the above-described embodiments as they are, and can be embodied by modifying the components without departing from the scope of the invention in the implementation stage. Moreover, various inventions can be formed by appropriately combining a plurality of constituent elements disclosed in the above embodiments. Further, for example, a configuration in which some components are deleted from all the components shown in each embodiment is also conceivable. Furthermore, you may combine suitably the component described in different embodiment.

1: weather data storage unit, 2: power generation data storage unit, 3: learning period weather data extraction unit, 4: diagnosis period weather data extraction unit, 5: learning period power generation data extraction unit, 6: diagnosis period power generation data extraction unit, 7: data density calculation unit, 8: weighting factor calculation unit, 9: weighting factor storage unit, 10: prediction model creation unit, 11: prediction model storage unit, 12: prediction value calculation unit, 13: prediction value storage unit, 14 : Abnormality diagnosis unit, 15: Diagnosis result storage unit, 101: CPU, 102: RAM, 103: HDD, 104: Graphic processing device, 105: Input interface, 106: Communication interface, 107: LAN, 108: Display, 109: Keyboard, 110: Mouse

Claims (7)

A coefficient calculation unit for calculating a weighting coefficient indicating a relationship between the distribution of the weather parameter in the learning period and the distribution of the weather parameter in the diagnosis period;
A prediction model creation unit that creates a prediction model of the power generation amount of the diagnosis period based on the weather parameter of the learning period, the power generation amount of the power generation facility of the learning period, and the weighting factor;
A predicted value calculation unit that calculates a predicted value of the power generation amount in the diagnosis period based on the prediction model and the weather parameter in the diagnosis period;
A diagnostic unit for diagnosing an abnormality in the power generation facility by comparing the predicted value of the power generation amount during the diagnosis period and the power generation amount during the diagnosis period;
An abnormality diagnosis device comprising: The abnormality diagnosis device according to claim 1, wherein the weighting coefficient is a probability density ratio between the weather parameter in the learning period and the weather parameter in the diagnosis period. The abnormality diagnosis apparatus according to claim 1, wherein the prediction model creation unit calculates a regression equation parameter using the power generation amount as an objective variable and the weather parameter as an explanatory variable. The abnormality diagnosis device according to any one of claims 1 to 3, wherein the weather parameter includes temperature, solar radiation amount, solar altitude, solar orientation, wind speed, wind force, and wind direction. The abnormality diagnosis device according to any one of claims 1 to 4, wherein the power generation amount includes power amount, DC voltage, DC current, AC voltage, AC current, DC power, and AC power. The abnormality diagnosis apparatus according to any one of claims 1 to 5, wherein the power generation facility includes a solar power generation facility, a wind power generation facility, and a solar thermal power generation facility. Calculating a weighting factor indicating the relationship between the distribution of the weather parameters during the learning period and the distribution of the weather parameters during the diagnosis period;
Based on the weather parameter of the learning period, the amount of power generation of the power generation facility in the learning period, and the weighting factor, create a prediction model of the power generation amount of the diagnostic period,
Based on the prediction model and the meteorological parameter of the diagnosis period, to calculate a predicted value of the power generation amount of the diagnosis period,
An abnormality diagnosis device comprising diagnosing an abnormality of the power generation facility by comparing the predicted value of the power generation amount during the diagnosis period and the power generation amount during the diagnosis period.

Images