Information processing apparatus and monitoring apparatus

• US 5,638,492 A
• Filed: 08/17/1994
• Issued: 06/10/1997
• Est. Priority Date: 09/08/1992
• Status: Expired due to Term

• Current Assignee: Hitachi, Ltd.
• Original Assignee: Hitachi, Ltd.
• Inventors: Yanagi, Mitsuo, Fujikura, Nobuyuki, Yoda, Mikio, Maeda, Akira, Funabashi, Motohisa, Yamaoka, Hiromasa, Ichimori, Toshihide
• Primary Examiner(s): Hafiz, Tariq R.
• Application Number:
  US08/291,869
• Time in Patent Office: 1,028 Days
• Field of Search: 364/422, 395/3, 395/10-11, 395/50, 395/51, 395/60-61, 395/900
• US Class Current: 706/45
• CPC Class Codes: F02D 41/1405 Neural network control
• CPC Class Codes: G05B 13/0285 using neural networks and fuzzy logic
• CPC Class Codes: G06V 10/75 Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries

Abstract

An information processing apparatus is capable of achieving high-level functions such as high-speed processing and learning without requiring a complex algorithm based on, for example, knowledge processing. A plurality of sets of known input data and correct output data corresponding thereto are beforehand stored. The stored input data is called response pattern data. Data actually inputted in the system operation is supplied to the plural sets of data. For each set of data, the input data is compared with the response pattern data to evaluate a distance (degree of similarity) therebetween. Results of evaluations for the respective sets of data are combined with the associated output data, for example, according to weighted average values on the basis of the evaluated distance, thereby generating final outputs.

Claims

1. An information processing apparatus for obtaining a missing value portion from input data values partially containing missing values, comprising:

a plurality of storage means for storing therein sets of the input data values;

means for supplying the plural storage means with the input data values;

means for comparing, for each said storage means, current input data supplied thereto with previously input data values stored therein, and for thereby evaluating a similarity degree therebetween for each said set;

means for deriving a total result using a result of the comparison/evaluation for each said set and the stored input data;

generating means for combining the input data values with the derived total result to generate data free of missing values; and

means for outputting the generated data.

2. An information processing apparatus according to claim 1, wherein said generating means includes means for replacing the missing values of the input data with values of the total result corresponding thereto to combine the input data values with the derived total result.

4. An information processing apparatus according to claim 2, wherein said means for calculating output data includes means for calculating the output data using a neural network.

3. An information processing apparatus for calculating output data having at least one element from input data having a plurality of elements, comprising:

means for storing therein a plurality of sets of input data and output data;

means for generating input data free of missing values from a stored previously input data item containing missing values and at least a portion of stored current input data;

means for calculating output data from the generated input data; and

means for modifying parameters included in the calculating means on the basis of a discrepancy between the stored output data and the calculated output data.

5. An apparatus for monitoring a system including a control object, at least one measuring means for measuring states of said control object, means for supplying control target values for said control object, and a controller for determining a manipulation amount for said control object on the basis of values measured by said measuring means and the control target values, comprising:

a plurality of storage means for storing sets of at least a plurality of items of a portion of the measured values, control target values, and the manipulation amount;

means for comparing and evaluating for each said storage means a similarity degree between at least a portion of current measured values, control target values, and manipulation amount with stored previously input values;

means for attaining for each said set a minimum value of the similarity degrees having thus undergone the comparison/evaluation; and

means for deciding a state of said control object from the minimum value of the obtained similarity degree.

6. A monitoring apparatus according to claim 5, wherein said comparing/evaluating means includes:

means for replacing a portion of the measured values with missing values to input resultant values to said apparatus;

means for calculating measured values free of missing values from the results of the comparison/evaluation for each said set and the stored values;

means for attaining a similarity degree between the current measured values corresponding to the values replaced with the missing values and the measured values thus calculated; and

means for deciding presence or absence of abnormality in the measured values replaced with the missing values on the basis of the similarity degree.

7. A monitoring apparatus according to claim 5, wherein said comparing/evaluating means includes composing means for combining the current measured values with the measured values thus calculated and thereby obtaining corrected measured values.

8. A monitoring apparatus according to claim 6, wherein said deciding means includes means for simultaneously displaying the current measured values and the calculated measured values and thereby presenting the state of said control object.

9. A monitoring apparatus according to claim 7, wherein said composing means includes means for selecting the calculated measured values for the measured values decided as abnormal and for selecting the current measured values for measured values other than the measured values decided as abnormal, thereby combining the current measured values with the calculated measured values.