| CPC H02S 50/00 (2013.01) | 13 Claims |
|
1. An anomaly factor diagnosis apparatus, comprising: a database including
measurement data items for a plurality of days, the measurement data items each including generated power and solar irradiances as measured values related to a power generation apparatus wherein the day includes a plurality of unit time slots, and
anomaly factor occurrence data including whether or not an anomaly factor occurs for the power generation apparatus for each of the days;
an identifier configured to identify first measurement data items for one or more days on which the anomaly factor occurred, and second measurement data items for one or more days on which the anomaly factor does not occur;
a distribution generator configured to
specify a maximum value of values of the generated power divided by the solar irradiance for a pair of the first and second measurement data belonging to the unit time slot for each unit time slot, and
calculate a first distribution function for each unit time slot representing a distribution of feature values related to generated power, the feature values being ratios between the maximum value divided by the solar irradiance and the generated power divided by the solar irradiance based on the first measurement data items for each unit time slot, an input variable of the first distribution function being the feature value and an output variable of the first distribution being a probability, and
calculate a second distribution function for each unit time slot representing a distribution of feature values related to generated power, the feature values being ratios between the maximum value divided by the solar irradiance and the generated power divided by the solar irradiance based on the second measurement data items for each unit time slot;
an evaluator configured to
determine whether a pair of the first distribution function and the second distribution function satisfies a separation condition that the first distribution function and the second distribution function are separate from each other for each unit time slot to identify one or more unit time slots that satisfies the separation condition,
identify a first percent point in an upper side of the first distribution function and a second percent point in a lower side of the second distribution function, detects whether an interval between the first percent point and the second percent point contains at least one of respective peaks of the first distribution function and the second distribution function, and determines that the separation condition is satisfied only when none of the peaks belong to the interval,
acquire a target measurement data item to be tested on a day which includes measured values related to the power generation apparatus, the target measurement data item including the generated power and the solar irradiances as the measured values and the day including the plurality of unit time slots,
obtain a feature value related to generated power based on the target measurement data for each of the identified unit time slots, the feature value related to generated power based on the target measurement data being a ratio between the maximum value divided by the solar irradiance and the generated power divided by the solar irradiance based on the target measurement data for each of the identified unit time slots,
calculate a score for each of the identified unit time slots, which is a ratio between the feature value calculated for each of the identified unit time slots and a maximum output value of the first distribution function corresponding to each of the identified unit time slots, and
average the score across the identified unit time periods to obtain an evaluation value indicating a possibility of occurrence of the anomaly factor in the power generation apparatus; and
an output device configured to output information indicative of whether the anomaly factor occurs for the power generation apparatus based on the evaluation value.
|