System identification device and system identification method

A system identification device, which identifies an unidentified system provided between a first input terminal and a second input terminal, comprising:

• a first adaptive filter in which a signal supplied from said first input terminal is inputted;
• a second adaptive filter configured to carry out system identification, in which said signal supplied from said first input terminal is inputted;
• a third adaptive filter in which said signal supplied from said first input terminal is inputted;
• a first subtracter configured to subtract an output signal of said first adaptive filter from a signal supplied from said second input terminal;
• a first output error average calculation part configured to calculate a first output error average based on an output signal of said first subtracter;
• a second subtracter configured to subtract an output signal of said second adaptive filter from said signal supplied from second input terminal;
• a second output error average calculation part configured to calculate a second output error average based on an output signal of said second subtracter;
• a third subtracter configured to subtract an output signal of said third adaptive filter from said signal supplied from second input terminal;
• a third output error average calculation part configured to calculate third output error average based on an output signal of said third subtracter;
• and
• a determination control part configured to compare said first output error average, said second output error average, and said third output error average,
• wherein said first output error average calculation part reduces at least a part of band components overlapping frequency bands of an observation noise, which is applied to said second input terminal, in frequency bands of said output signal of said first subtracter, and calculates said first output error average based on components remained after reducing,
• said second output error average calculation part reduces at least a part of band components overlapping frequency bands of an observation noise, which is applied to said second input terminal, in frequency bands of said output signal of said second subtracter and calculates said second output error average based on components remained after reducing,
• said third output error average calculation part reduces at least a part of band components overlapping frequency bands of an observation noise, which is applied to said second input terminal, in frequency bands of said output signal of said third subtracter and calculates said third output error average based on components remained after reducing, and
• at least one of updating said filter factor of said adaptive filter by said output signal of said first subtracter, updating said filter factor of said adaptive filter by said output signal of said second subtracter, and updating said filter factor of said adaptive filter by said output signal of said third subtracter is controlled based on a comparison result of said determination control part.

A system identification method for identifying an unidentified system provided between a first input terminal and a second input terminal, comprising:

• inputting a signal supplied from said first input terminal into an adaptive filter and a dependent filter which has a system identified filter factor;
• updating a filter factor of said dependent filter based on a filter factor of said adaptive filter;
• generating first output signal by subtracting an output signal of said adaptive filter from a signal supplied from said second input terminal;
• calculating a first output error average based on said first output signal;
• generating a second output signal, by subtracting an output signal of said dependent filter from said signal supplied from said second input terminal;
• calculating a second output error average based on said second output signal;
• and
• comparing said first output error average and said second output error average,
• wherein said calculating said first output error average includes, reducing at least a part of band components overlapping frequency bands of an observation noise, which is applied to said second input terminal, in frequency bands of said first output signal, and calculating said first output error average based on components remained after said reducing,
• said calculating said second output error average includes, reducing at least a part of band components overlapping frequency bands of an observation noise, which is applied to said second input terminal, in frequency bands of said second output signal, and calculating said second output error average based on components remained after said reducing, and
• further comprising, controlling at least either updating said filter factor of the adaptive filter by said output signal of said first subtracter or updating said filter factor of said dependent filter by said dependent filter factor update part, based on a comparison result of a comparator.

A system identification method for identifying an unidentified system provided between a first input terminal and a second input terminal, comprising:

• inputting a signal supplied from said first input terminal into a first adaptive filter;
• inputting said signal supplied from said first input terminal into a second adaptive filter which carries out system identification;
• inputting said signal supplied from said first input terminal into a third adaptive filter;
• subtracting an output signal of said first adaptive filter from a signal supplied from said second input terminal, by using a first subtracter;
• calculating a first output error average based on an output signal of said first subtracter, by using a first output error average calculation part;
• subtracting an output signal of said second adaptive filter from said signal supplied from said second input terminal, by using a second subtracter;
• calculating a second output error average based on an output signal of said second subtracter, by using a second output error average calculation part;
• subtracting an output signal of said third adaptive filter from said signal supplied from said second input terminal, by using a third subtracter;
• calculating a third output error average based on an output signal of said third subtracter, by using third output error average calculation part;
• and
• comparing said first output error average and said second output error average, by using a determination control part,
• wherein said calculating said first output error average includes, reducing at least a part of band components overlapping frequency bands of an observation noise, which is applied to said second input terminal, in frequency bands of said output signal of said first subtracter, and calculating said first output error average based on components which are remained after reducing,
• said calculating said second output error average includes, reducing at least a part of band components overlapping frequency bands of an observation noise, which is applied to said second input terminal, in frequency bands of said output signal of said second subtracter, and calculating said second output error average based on components which are remained after reducing,
• said calculating said third output error average includes, reducing at least a part of band components overlapping frequency bands of an observation noise, which is applied to said second input terminal, in frequency bands of said output signal of said third subtracter, and calculating said third output error average based on components which are remained after reducing, and
• further comprising, controlling at least one of updating said filter factor of said adaptive filter by said first subtracter, updating said filter factor of said adaptive filter by said second subtracter, and updating said filter factor of said adaptive filter by said third subtracter, based on a comparison result of said determination control part.