Intelligent machining system

• US 5,473,532 A
• Filed: 07/17/1991
• Issued: 12/05/1995
• Est. Priority Date: 07/17/1990
• Status: Expired due to Fees

• Current Assignee: Toyoda Koki Kabushiki Kaisha, Yukinori Kakazu
• Original Assignee: Toyoda Koki Kabushiki Kaisha, Yukinori Kakazu
• Inventors: Unno, Kunihiko, Kakazu, Yukinori, Yamanaka, Masashi, Hattori, Shiho, Sakakura, Moriaki, Yoneda, Takao
• Primary Examiner(s): Envall, Jr., Roy N.
• Assistant Examiner(s): OAKES, BRIAN
• Application Number:
  US07/731,472
• Time in Patent Office: 1,602 Days
• Field of Search: 395/23, 395/24, 395/25, 395/26, 395/27, 395/10, 395/11, 395/9, 395/89, 395/21, 395/903, 395/904, 364/152, 364/148, 364/149, 364/474.11, 364/474.17, 364/474.16
• US Class Current: 700/32
• CPC Class Codes: G05B 19/40937 concerning programming of machining or material parameters, pocket machining
• CPC Class Codes: G05B 19/4163 Adaptive control of feed or cutting velocity without NC B23Q15/12
• CPC Class Codes: G05B 2219/33027 Artificial neural network controller
• CPC Class Codes: G05B 2219/34065 Fuzzy logic, controller
• CPC Class Codes: G05B 2219/36284 Use of database for machining parameters, material, cutting method, tools
• CPC Class Codes: G05B 2219/49061 Calculate optimum operating, machining conditions and adjust, adapt them
• CPC Class Codes: G05B 2219/49088 As a function of, regulate feed as function of material, tool
• CPC Class Codes: Y02P 90/02 Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]
• CPC Class Codes: Y10S 706/904 Manufacturing or machine, e.g. agricultural machinery, machine tool

Abstract

An intelligent machining system employs a neural network for calculating machining conditions on the basis of attribute data or a workpiece and a grinding machine. The system comprises a reference machining condition calculating unit, a neural network which receives the attribute data and provides corrections, and a correcting unit for correcting the reference machining conditions by using the corrections. Corrections which cannot be determined by means of empirical expressions or theoretical expressions are determined appropriately by the neural network previously learned. The system determines corrections for the machining conditions on the basis of machining errors decided by a neural network. The system also detects time-series machining phenomena by sensors, processes the output detection signals of the sensors by a neural network to obtain machining circumstance data. The feed rate of the tool is controlled by a fuzzy inference on the basis of the machining circumstance data.

Claims

1. An intelligent machining system for automatically determining machining conditions and for controlling a numerically controlled machine tool to machine a workpiece in accordance with the machining conditions, said intelligent machining system comprising:

attribute data storage means for storing attribute data indicating characteristics of at least one of said workpiece and machining requirements; and

machining condition calculating means for calculating the machining conditions on the basis of the attribute data, said machining condition calculating means comprising;

reference value calculating means for calculating reference machining conditions on the basis of the attribute data by the use of predetermined calculating expressions; and

a first neural network responsive to the attribute data and associated with said reference value calculating means so that the reference machining conditions calculated by the reference value calculating means are corrected by said first neural network so as to output the corrected machining conditions from said machining condition calculating means to said machine tool to machine said workpiece.

2. An intelligent machining system according to claim 1, wherein said first neural network receives the attribute data for correcting the reference machining conditions and comprises at least three layers including an input layer, an intermediate layer and an output layer, and the respective outputs of the elements of the output layer are fed back to the elements of one of said input and intermediate layers, and the intelligent machining system further comprises:

first operating means for operating said first neural network in a mode where no output of the output layer is fed back to said one of said input and intermediate layers; and

second operating means for applying the attribute data to the input layer to make the output layer provide the corrections in a state where the outputs of the output layer are fed back to said one of said input and intermediate layers, after said first neural network has been operated by said first operating means.

3. An intelligent machining system according to claim 1, further comprising:

teaching data storage means for storing input data accumulated during the operation of said first neural network, and teaching data corresponding to the input data; and

learning means for performing the learning of said first neural network by using the input data and the teaching data stored in the teaching data storage means.

4. An intelligent machining system according to claim 1, wherein said first neural network provides corrections based on the attribute data and wherein said machining condition calculating means includes means for correcting said reference machining conditions calculated by said reference value calculating means, by using the corrections provided by said first neural network so that the corrected machining conditions are output from said machining condition calculating means.

5. An intelligent machining system according to claim 1, further comprising:

machining condition storage means for storing the machining conditions determined by the machining condition calculating means;

display means for visually displaying the machining conditions stored in the machining condition storage means for evaluation by an operator; and

correction data input means for enabling the operator to input correction data so that the machining conditions stored in the machining condition storage means are manually corrected by the operator.

6. An intelligent machining system for automatically determining machining conditions and for controlling a numerically controlled machine tool to machine a workpiece in accordance with the machining conditions, said intelligent machining system comprising:

attribute data storage means for storing attribute data indicating characteristics of at least one of said workpiece and machining requirements;

a first neural network for determining, on the basis of the attribute data, elementary values used in determining the machining conditions; and

machining condition calculating means for calculating the machining conditions on the basis of the attribute data stored in the attribute data storage means and the elementary values determined by the first neural network and for outputting said machining conditions to said machine tool to machine said workpiece;

wherein the machining condition calculating means has predetermined calculating expressions which are used in calculating the machining conditions,

the elementary values determined by said first neural network are coefficients for the predetermined calculating expressions used in calculating the machining conditions on the basis of the attribute data, and

said machining condition calculating means calculates the machining conditions on the basis of the attribute data by using the predetermined calculating expressions and the coefficients determined by the first neural network.

7. An intelligent machining system for automatically determining machining conditions and for controlling a numerically controlled machine tool to machine a workpiece in accordance with the machining conditions, said intelligent machining system comprising:

attribute data storage means for storing attribute data indicating characteristics of at least one of said workpiece and machining requirements; and

machining condition calculating means for calculating the machine conditions on the basis of the attribute data, said machining condition calculating means comprising;

reference value calculating means for calculating reference machining conditions on the basis of the attribute data by the use of predetermined calculating expressions; and

a first neural network responsive to the attribute data and associated with said reference value calculating means so that the reference machining conditions calculated by the reference value calculating means are corrected by said first neural network so as to output the corrected machining conditions from said machining condition calculating means to said machine tool to machine said workpiece;

wherein said intelligent machining system further comprises;

machining condition storage means for storing the machining conditions determined by said machining condition calculating means;

error data storage means for storing machining errors relative to values representing desired machining results;

a second neural network which receives the machining errors stored in the error data storage means and provides corrections for the machining conditions;

control means for controlling said second neural network so as to receive the machining errors and to provide the corrections; and

correction storage means for storing the corrections provided by said second neural network, for use in subsequently correcting the machining conditions stored in said machining condition storage means.

8. An intelligent machining system according to claim 7, further comprising learning means for performing the learning of said first neural network, which receives the attribute data, on the basis of corrections for correcting the machining conditions, provided by said second neural network, which receives machining error data.

9. An intelligent machining system according to claim 7, further comprising:

correcting mean for correcting the machining conditions stored in said machining condition storage means by using the corrections provided by said second neural network.

10. An intelligent machining system according to claim 7, further comprising:

correction data input means for enabling an operator to input correction data so that the machining conditions stored in the machining condition storage means are manually corrected by the operator; and

learning means for performing the learning of the first neural network which is associated with the reference value calculating means for calculating the machining conditions, by using both of the correction data manually input by the operator and the correction data provided by said second neural network.

11. An intelligent machining system for automatically determining machining conditions and for controlling a numerically controlled machine tool to machine a workpiece in accordance with the machining conditions, said intelligent machining system comprising:

attribute data storage means for storing attribute data indicating characteristics of at least one of said workpiece and machining requirements;

reference value calculating means operable in accordance with predetermined calculating expressions for calculating reference machining conditions on the basis of the attribute data stored in said attribute data storage means;

a first neural network responsive to the attribute data for correcting the reference machining conditions calculated by said reference value calculating means so as to provide the machining conditions and for outputting said machining conditions to said machine tool to machine said workpiece;

machining condition storage means for storing the machining conditions corrected by said first neural network;

error data storage means for storing machining errors relative to desired machining results;

a second neural network which receives the machining errors stored in the error data storage means and provides corrections for further correcting the machining conditions stored in the machining condition storage means;

control means for controlling the second neural network so as to receive the machining errors and to provide the corrections; and

correction storage means for storing the corrections provided by the second neural network, for use in subsequently further correcting the machining conditions stored in said machining condition storage means.

12. An intelligent machining system according to claim 11, further comprising:

teaching data storage means for storing input data accumulated during the operation of said neural network, and teaching data corresponding to the input data; and

learning means for performing the learning of said neural network by using the input data and the teaching data stored in the teaching data storage means.

13. An intelligent machining system according to claim 11, further comprising:

correcting means for correcting the machining conditions stored in the machining condition storage means by using the corrections provided by the second neural network.

14. An intelligent machining system for automatically determining machining conditions and for controlling a numerically controlled machine tool to machine a workpiece in accordance with the machining conditions, said intelligent machining system comprising:

attribute data storage means for storing attribute data indicating characteristics of at least one of said workpiece and machining requirements; and

machining condition calculating means for calculating the machining conditions on the basis of the attribute data, said machining condition calculating means comprising;

reference value calculating means for calculating reference machining conditions on the basis of the attribute data by the use of predetermined calculating expressions; and

a first neural network responsive to the attribute data and associated with said reference value calculating means so that the reference machining conditions calculated by the reference value calculating means are corrected by said first neural network so as to output the corrected machining conditions from said machining condition calculating means to said machine tool to machine said workpiece;

wherein said intelligent machining system further comprises;

sensor means for detecting at least one of instantaneous machining phenomena such as sparks, sound and force produced in machining a workpiece by the numerically controlled machine tool;

a second neural network which receives time-series data in a predetermined time interval provided by the sensor means and provides machining circumstance data in the corresponding time interval; and

feed control means for controlling the feed of a tool of the numerically controlled machine tool according to the machining circumstance data provided by said second neural network.

15. An intelligent machining system according to claim 14, further comprising:

teaching data storage means for storing input data accumulated during the operation of said first neural network, and teaching data corresponding to the input data; and

learning means for performing the learning of said first neural network by using the input data and the teaching data stored in the teaching data storage means.

16. An intelligent machining system for automatically determining machining conditions and for controlling a numerically controlled machine tool to machine a workpiece in accordance with the machining conditions, said intelligent machining system comprising:

attribute data storage means for storing attribute data indicating characteristics of at least one of said workpiece and machining requirements; and

machining condition calculating means for calculating the machining conditions on the basis of the attribute data, said machining condition calculating means comprising;

reference value calculating means for calculating reference machining conditions on the basis of the attribute data; and

a first neural network responsive to the attribute data and associated with said reference value calculating means so that the reference machining conditions calculated by the reference value calculating means are corrected by said first neural network so as to output the corrected machining conditions from said machining condition calculating means to said machine tool to machine said workpiece;

wherein said intelligent machining system further comprises;

sensor means for detecting at least one of instantaneous machining phenomena such as sparks, sound and force produced in machining a workpiece by the numerically controlled machine tool;

a second neural network which receives time-series data in a predetermined time interval provided by the sensor means and provides machining circumstance data in the corresponding time interval; and

feed control means for controlling the feed of a tool of the numerically controlled machine tool according to the machining circumstance data provided by said second neural network;

wherein said feed control means has fuzzy inference means for calculating the respective adaptations of fuzzy rules by using values representing the grades of machining circumstances so as to infer a manipulated variable for controlling the controlled variable of the feed, whereby said feed control means controls the feed of the tool also in accordance with the manipulated variable.

17. An intelligent machining system according to claim 16, further comprising:

correspondence relation calculating means for determining machining conditions on the basis of the outputs of said second neural network to detect the relevant controlled variable of the feed in machining a workpiece by the manual operation of the numerically controlled machine tool by a skilled operator; and

fuzzy rule producing means for producing fuzzy rules on the basis of the relation between the machining conditions determined by the correspondence relation calculating means and the controlled variable of the feed.

18. An intelligent machining system according to claim 16, further comprising learning means for determining corrections for correcting the machining conditions on the basis of the mean value of manipulated variables provided in a predetermined time interval by said fuzzy inference means, and for performing the learning of said first neural network, which receives the attribute data, on the basis of the corrections.

19. An intelligent machining system for automatically determining machining conditions and for controlling a numerically controlled machine tool to machine a workpiece in accordance with the machining conditions, said intelligent machining system comprising:

attribute data storage means for storing attribute data indicating characteristics of at least one of said workpiece and machining requirements; and

machining condition calculating means for calculating the machining conditions on the basis of the attribute data, said machining condition calculating means comprising;

reference value calculating means for calculating reference machining conditions on the basis of the attribute data; and

a first neural network responsive to the attribute data and associated with said reference value calculating means so that the reference machining conditions calculated by the reference value calculating means are corrected by said first neural network so as to output the corrected machining conditions from said machining condition calculating means to said machine tool to machine said workpiece;

wherein said intelligent machining system further comprises;

sensor means for detecting at least one of instantaneous machining phenomena such as sparks, sound and force produced in machining a workpiece by the numerically controlled machine tool;

a second neural network which receives time-series data in a predetermined time interval provided by the sensor means and provides machining circumstance data in the corresponding time interval; and

feed control means for controlling the feed of a tool of the numerically controlled machine tool according to the machining circumstance data provided by said second neural network; and

wherein said feed control means has fuzzy inference means for calculating the adaptations of fuzzy rules by using values representing the grades of machining circumstances, so as to infer a manipulated variable for controlling the controlled variable of the feed, whereby said feed control means controls the feed of the tool also in accordance with the manipulated variable.

20. An intelligent machining system for automatically determining machining conditions and for controlling a numerically controlled machine tool to machine a workpiece in accordance with the machining conditions, said intelligent machining system comprising:

attribute data storage means for storing attribute data;

automatic machining condition determining means including a first neural network and capable of automatically determining machining conditions on the basis of the attribute data stored in the attribute data storage means;

correcting means for correcting the machining conditions determined by the automatic machining condition determining means, according to instructions given thereto by an operator to provide corrected machining conditions;

machining control means for controlling the machining operation of the numerically controlled machine tool according to the machining conditions or the corrected machining conditions;

diagnostic means including a second neural network, which receives machining errors corresponding to deviations of machined results from desired machined results and provides corrections for correcting the machining conditions, and capable of correcting the machining conditions by using the corrections to determine corrected machining conditions;

learning means for performing the learning of the first neural network for determining machining conditions according to the corrected machining conditions determined by at least one of the correcting means and the diagnostic means so that the corrected machining conditions are provided for the same input attribute data as that initially applied to the automatic machining condition determining means;

sensor means for detecting at least one of instantaneous machining phenomena such as sparks, sound and force produced in machining the workpiece; and

adaptive control means for dynamically correcting the machining conditions by receiving time-series data provided by said sensor means by a third neural network, and by recognizing average machining circumstances in a time interval on the basis of the output of the third neural network.