Plant operation support system

• US 5,581,459 A
• Filed: 09/26/1991
• Issued: 12/03/1996
• Est. Priority Date: 09/26/1990
• Status: Expired due to Fees

• Current Assignee: Hitachi, Ltd.
• Original Assignee: Hitachi, Ltd.
• Inventors: Baba, Kenji, Yoda, Mikio, Watanabe, Shoji, Hara, Naoki, Enbutsu, Ichiro, Yahagi, Hayao
• Primary Examiner(s): Envall, Jr., Roy N.
• Assistant Examiner(s): Garland, Steven R.
• Application Number:
  US07/765,839
• Time in Patent Office: 1,895 Days
• Field of Search: 395/22-24, 395/76, 395/903, 395/906, 395/21, 364/148, 364/152, 364/188
• US Class Current: 700/32
• CPC Class Codes: C02F 1/5209 Regulation methods for flocculation or precipitation
• CPC Class Codes: C02F 1/76 with halogens or compounds of halogens C02F1/4674 takes precedence
• CPC Class Codes: C02F 2103/007 Contaminated open waterways, rivers, lakes or ponds
• CPC Class Codes: C02F 2209/11 Turbidity
• CPC Class Codes: C02F 2209/38 Gas flow rate
• CPC Class Codes: C02F 3/006 Regulation methods for biological treatment
• CPC Class Codes: C02F 3/12 Activated sludge processes
• CPC Class Codes: G05B 13/027 using neural networks only
• CPC Class Codes: G06N 3/042 Knowledge-based neural networks; Logical representations of neural networks
• CPC Class Codes: Y02W 10/10 Biological treatment of water, waste water, or sewage
• CPC Class Codes: Y10S 706/903 Control
• CPC Class Codes: Y10S 706/906 Process plant

Abstract

When amounts of operation control variables and/or operation state evaluation variables of a plant are calculated through a multilayered neural network based on data indicating a plant state, and the calculated results are indicated to an operator, a plant operation support system has functions to analyze an internal causality between neurons in the neural network, and display quantitative guidance by association in the neural network and also the analyzed result of the internal causality between the neurons as an association reason.

Claims

1. A plant operation support system in which either of a plurality of operation control variables and operation state evaluation variables of a plant are calculated through a multilayered neural network based on data indicating a plant state, and the calculated results are communicated to an operator of said plant to support plant operation, said support system comprising:

means for analyzing an internal causality between neurons in said neural network;

means for displaying quantitative guidance associated with said neural network and also an analyzed result of the internal causality between the neurons as an association reason;

data selection means for determining the magnitude of causality between input and output of said neural network and selecting data which has a causality larger than a preset reference; and

data display means for displaying the selected data,

wherein said data selection means selects such a data item that an absolute value of said value resulting by summing a product between the connection intensity in a path reaching an output layer neuron of interest and a differential value of a conversion function used at a passing neuron for all the paths via hidden layer neurons exceeds a predetermined value, as data having the large causality.

2. A plant operation support system according to claim 1, further comprising:

data predicting means for predicting a quantity of either said plurality of operation control variables and operation state evaluation variables of said plant; and

guidance means for preparing prediction information about the operation state of said plant based on the predicted result of said data predicting means, and sending the prediction information to said data display means.

3. A plant operation support system according to claim 2, wherein said guidance means outputs at least one of an operation schedule and abnormality prediction information as said prediction information about the operation state of said plant.

4. A plant operation support system according to claim 2, wherein said data predicting means obtains said prediction information by using the same neural network as for use with said data selection means.

5. A plant operation support system in which either of a plurality of operation control variables and operation state evaluation variables of a plant are calculated through a multilayered neural network based on data indicating a plant state, and the calculated results are communicated to an operator of said plant to support plant operation, said support system comprising:

means for analyzing an internal causality between neurons in said network;

means for displaying quantitative guidance associated with said neural network and also an analyzed result of the internal causality between the neurons as an association reason;

data selection means for determining the magnitude of causality between input and output of said neural network and selecting data which has a causality larger than a preset reference; and

data display means for displaying the selected data,

wherein said data display means statistically analyzes each data item with respect to a population and qualitatively evaluates each of said data items.

6. A plant operation support system in which either of a plurality of operation control variables and operation state evaluation variables of a plant are calculated through a multilayered neural network based on data indicating a plant state, and the calculated results are communicated to an operator of said plant to support plant operation, said support system comprising:

means for analyzing an internal causality between neurons in said network;

means for displaying quantitative guidance associated with said neural network and also an analyzed result of the internal causality between the neurons as an association reason;

data selection means for determining the magnitude of causality between input and output of said neural network and selecting data which has a causality larger than a preset reference; and

data display means for displaying the selected data,

wherein said data display means determines a degree of adaptation for each data item by the use of a preset membership function and qualitatively evaluates the data item selected by said data selection means.