JPH04106145U - Machine tool failure prevention device - Google Patents

Abstract

(57) [Summary] [Purpose] To prevent machine tool failures. To accurately perform preventive diagnosis of a failure by setting a failure judgment level with relatively low accuracy. [Configuration] Vibration, sound, load, etc. during continuous operation of machine tool 1,
The cycle time and the like are statistically processed to cause the normal state learning means 10 to learn the normal state. Based on this learning result, a failure diagnosis rule is set in the rule setting section 12, and the failure prevention diagnosis means 11 predicts the occurrence of a failure. Based on the judgment results, the machine will be stopped before it breaks down and maintenance will be required. When diagnosing failure prevention, try to catch abnormalities statistically.

Description

[Detailed explanation of the idea]

0001

[Industrial application field]

This idea enables preventive diagnosis of failures in machine tools such as lathes and machining centers. This invention relates to failure prevention equipment for machine tools.

0002

[Conventional technology]

Traditionally, when trying to prevent machine tool failures, abnormal noises from the machine were used to indicate that the machine was about to break down. It can be determined based on the operator's experience, or the load of the feed shaft motor can be set to In comparison, the machine is automatically stopped in the event of an abnormal load.

0003

[Problem that the idea aims to solve]

However, when making decisions based on the worker's experience, it is uncertain and difficult to make decisions. There are some problems. In addition, if the judgment is based on an abnormal load on the motor, check the condition of each part of the machine. The level at which an abnormality should be determined changes depending on the interrelationship between There are problems in that setting is difficult and accurate diagnosis is difficult.

0004 The purpose of this invention is to set a relatively low-accuracy failure determination level and to perform preventive diagnosis of failures. We provide a failure prevention device for machine tools that allows accurate cutting and prevents failures from occurring. It is to be.

[0005]

[Means to solve the problem]

The configuration of this invention will be explained with reference to FIG. 1 which corresponds to an embodiment. Noise, vibration, load, and noise generated during continuous operation of the machine tool (1), and noise at specified time intervals. At least two or more of the detection elements are statistically processed to determine whether each detection element is in a normal state. A normal state learning means (10) for storing state quantities is provided. This normal state learning means ( 10) The fault diagnosis rules created based on each state quantity memorized in step 10) are set as rules. Set to fixed part (12). Each detection element is monitored and the rule setting unit (12) is monitored. Failure preventive diagnosis means (11) that outputs a failure prediction signal (a) according to failure diagnosis rules will be established.

[0006]

[Effect]

According to this configuration, the normal state learning means (10) allows the noise generated during continuous operation of the machine to be , vibration, load, and at least two of the detected elements at a predetermined time interval are statistically The state quantity of each detection element in a normal state is stored. I remembered this Create a failure diagnosis rule based on each state quantity and set it in the rule setting section (12). Ru. The failure diagnosis rule is based on the state quantities of the normal state and adds the operator's judgment. It may be created manually or automatically generated.

[0007] When the machine is operating, the failure prevention diagnosis means (11) checks each of the detection elements. monitor and generate failure prediction signals according to the failure diagnosis rules set in the rule setting section (12). Output number (a).

[0008]

【Example】

An embodiment of this invention will be described based on FIGS. 1 to 3. The machine tool 1 is an NC lathe that will be explained later in conjunction with FIG. A failure prevention control unit 2 is provided as a part of the functional means of the programmable controller. be. The machine tool 1 includes a sound detector 3, a vibration detector 4, a load detector 5, and a timer. Means 6 is provided, and the detection signals of the respective detectors 3 to 5 are respectively sampled by sampling means. The signals are input to the failure prevention control unit 2 via 7 to 9. The detection signal of the clock means 6 is directly is input to the failure prevention control unit 2.

[0009] The failure prevention control unit 2 mainly includes a normal state learning means 10 and a failure prevention diagnosis means 11. , and a rule setting section 12. The normal state learning means 10 includes each of the machine tools 1. The detected values of the detectors 3 to 5 and the timer 6 are statistically processed to determine each state in the normal state. It is a means of storing quantities. The rule setting unit 12 sets predetermined failure diagnosis rules. The failure diagnosis rule is based on each state stored in the normal state learning means 10. Created based on quantity. The failure preventive diagnosis means 11 includes each detector 3 of the machine tool 1. 5 and the detection values of the clock means 6, and set the failure diagnosis rules of the rule setting section 12. This is means for outputting a failure prediction signal a according to the following. The failure prediction signal a is the machine tool 1 This serves as a stop command for the system, and is also used as a notification command for failure prediction by the notification means 13. The notification means 13 is a message display means using a lamp, a buzzer, or a CRT screen. Consists of etc.

0010 FIG. 2 is a schematic plan view of a turret lathe, which is the machine tool 1. Supported by headstock 14 The main shaft 15 is driven by a main shaft motor 17 via a belt 16 hung around a pulley. Driven. The turret 18 is a drum-shaped one with various tools 19 attached to its circumference. The turret carriage 20 is provided with index rotation and axis via the turret shaft 21. It is mounted so that it can move in the direction (Z direction). The turret carriage 20 is a bed It is installed on the rail 23 of 22 so that it can move laterally (in the X direction), and the feed screw 24 and motor It is driven by the motor 25. The front of the headstock 14 and turret carriage 20 is , which is covered by a part of the machine cover 26 and in front of the main shaft 15, is provided to reduce the amount of wear of the tool 19. A blade edge verification device 27 for measurement is provided so as to be retractable via a rotating arm 28.

[0011] In FIG. 1, the sound detector 3 is a microphone that detects sound generated at a predetermined location of the machine tool 1. The vibration detector 4 is a means for detecting vibrations of the fuselage cover 26 etc. in FIG. Consisting of One or more of these sound detectors 3 and vibration detectors 4 are provided, If multiple detectors are installed, the detected values of each sound detector 3 and vibration detector 4 can be used to prevent failures. It is input to the control section 2. The vibration detector 4 detects a striped pattern formed on the machined surface of the workpiece W. It may also be a means for detecting vibration from.

0012 The load detector 5 detects the load current of the drive motor 25 of the turret carriage 20. The detector 5a and the load current detector of the Z-direction feed motor of the turret 18 (not shown) ), load current detector 5b of main shaft motor 17, tension detector 5c of belt 16 Consists of etc. The detected values of each of these detectors 5a to 5c correspond to the failure prevention control shown in FIG. The data is inputted to section 2. The load detector 5 in FIG. 1 represents these detectors 5a to 5c, etc. This is what was shown. If the machine tool 1 has a loader (not shown), If so, the detected values of the load current detectors for each axis of the loader are also input to the failure prevention control unit 2. I will do it. The timing means 6 measures the cycle time in which one workpiece W is processed. Judgment of tool life based on the results of verifying the amount of tool wear using the cutting edge verification device 27 Consists of means etc.

[0013] FIG. 3 shows a specific example of the failure prevention control section 2 shown in FIG. The failure prevention control unit 2 The support system 29, the failure prevention diagnosis means 11, and the It is composed of a machine section 31. The failure prevention diagnosis means 11 includes a fuzzy inference section 32 , the rule setting section 12, the input conversion section 33, the output conversion section 34, and the statistical processing 38.

[0014] The fault diagnosis rules set in the rule setting section 12 are inferred by the fuzzy inference section 32. It consists of a large number of control rules for performing IF~THEN type. The formula rule is used. Labels representing degrees in rules are membership functions Defined using The input converting unit 33 receives data from each of the detectors 3 to 5 in FIG. 1 as described above. Each input signal input to the failure prevention control unit 2 is corrected in order to perform fuzzy inference. It is a means of regulation. The output converter 34 outputs the normalized data obtained by fuzzy inference. This is a means of quantizing the controlled variable. For a predetermined detection element, the statistical processing unit 38 Computes statistical data of detected values a large number of times and outputs the computed results to the input conversion section 33. It is a means of

[0015] The support system 29 consists of a feature extraction section 35 and a simulation means 36. , the feature extraction section 35 is provided with the normal state learning means 10. The feature extraction unit 35 , the characteristics of the normal state and abnormal state during continuous machine operation are input to the failure prevention control unit 2. automatically generate fault diagnosis rules and membership functions. This is means for setting in the setting section 12. The simulation means 36 includes a rule setting section What kind of control output will occur when the input data is changed using the 12 control rules? It is a means of evaluating whether the

[0016] The man-machine section 31 has a rule editor 37, and the rule setting section 12 has a rule editor 37. The operator can create, change, add, and delete control rules and membership functions. It is possible by operation. In addition, the man-machine unit 31 also controls the statistical processing unit 38. It is possible to change the processing program and setting values.

[0017] The operation of the above configuration will be explained. During continuous operation of machine tool 1 under normal conditions, each Sound, vibration, load, and predetermined time detected by detectors 3 to 5 and timing means 6 The interval detection signal is learned by the normal state learning means 10 of the failure prevention control section 2. . In this case, for example, continuous processing from the start of machining to the completion of machining of one workpiece W. Detection signals b to d consisting of analog values of each detector 3 to 5 outputted from the sample The digital data is sampled by ring means 7 to 9 at an appropriate sampling period. They are input to the failure prevention control unit 22 as rule values e to g.

[0018] The normal state learning means 10 statistically processes detection signals from multiple workpiece machining operations. The state quantities and characteristics of each detection element in the normal state and the state between each detection element are Extract and memorize relationships between physical quantities. Based on this memorized state quantity and characteristics, Fault diagnosis rules and membership functions are automatically generated and set in the rule setting section 12. Ru. When automatically generating these fault diagnosis rules and membership functions, the necessary data is required. The data is input by the rule editor 37, and the automatically generated failure diagnosis rule Modifications and the like are also performed using the rule editor 37.

[0019] Diagnosis for failure prevention is performed as follows. Failure prediction when operating machine tool 1 The diagnostic prevention means 11 monitors the detection values detected by each of the detectors 3 to 5 and the timer means 6. the failure prediction signal a according to the failure diagnosis rule set in the rule setting unit 12. Output. This failure prediction signal a causes the machine tool 1 to stop, and the notification means 1 3 to notify and prompt maintenance.

[0020] In the failure diagnosis, statistical elements calculated by the statistical processing unit 38, e.g. For example, fluctuations in the load value each time the operation is repeated many times are also considered factors for judgment. . In addition, the types of detection elements such as sound, vibration, load, etc., and the degree of detection elements, etc. Therefore, even in the case of one abnormality, the failure prediction signal a is output.

[0021] This failure prevention device for machine tools controls vibrations, noise, etc. during continuous machine operation. A failure prediction signal a can be generated before a failure by calculating and comparing it with the learned normal state. By outputting the information, it is possible to prevent failures from occurring. in this case The system comprehensively evaluates multiple detection elements such as sound, vibration, load, and a predetermined time interval. This allows accurate failure prediction and judgment for each detection element. The failure determination level can be set with relatively low accuracy.

[0022] In the above embodiment, the failure prevention control unit 2 was assumed to perform fuzzy control. It is also possible to perform failure prediction diagnosis using only processing using normal crisp set calculations. stomach.

[0023] Furthermore, although the above embodiment has been explained with reference to a case where it is applied to a lathe, this invention is applicable to a machine tool. It can be applied to machines in general.

[0024]

[Effect of the idea]

This machine tool failure prevention device is designed to prevent noise and vibration generated during continuous machine tool operation. The normal state is learned by statistically processing detection elements such as , load, and predetermined time intervals. A fault diagnosis rule is created based on the learning results. According to the rules, each detection element during operation is monitored and a failure prediction signal is output. This allows you to prevent failures from occurring. In this case, sound, vibration, load, and can comprehensively judge multiple detection elements within a predetermined time interval. Therefore, accurate failure prediction judgment can be made, and the failure judgment level can be set with relatively low precision. It has the effect of

[Brief explanation of drawings]

FIG. 1 is a conceptual diagram showing the configuration of a failure prevention device according to an embodiment of the invention.

FIG. 2 is a plan view of an example of a machine tool using the failure prevention device.

FIG. 3 is a block diagram showing a specific example of a part of the failure prevention device.

[Explanation of symbols]

1...Machine tool, 2...Failure prevention control unit, 3...Sound detector, 4
... Vibration detector, 5... Load detector, 6... Timing means, 10...
Normal state learning means, 11... Failure prevention diagnosis means, 12... Rule setting section, 13... Notification means, a... Failure prediction signal

Claims (1)

[Scope of utility model registration request] [Claim 1] Sound and vibration generated during continuous operation of a machine tool,
a normal state learning means for statistically processing at least two or more detection elements of the load and a predetermined time interval and storing state quantities of each of the detection elements in a normal state; and each state stored by the normal state learning means. A machine tool comprising: a rule setting section that sets a failure diagnosis rule created based on quantity; and a failure preventive diagnosis means that monitors each of the detection elements and outputs a failure prediction signal according to the failure diagnosis rule of the rule setting section. failure prevention device.