KR20230114818A - Iot based intelligent monitoring and diagnosis method and device for machine tool - Google Patents

Abstract

In the present invention, the machine information collection unit 10 acquires machine information data (machine_data) of a machine tool from at least one sensor selected from among rotation speed, vibration, sound, temperature, current, voltage, and load, and a state determination result ( Learning data acquisition step (S10) of acquiring learning data including normal, abnormal, failure, pre-failure, score); A condition diagnosis model learning step in which the AI learning unit 20 uses the learning data to learn a condition diagnosis model that outputs condition diagnosis information (abnormal score) of the machine tool by inputting machine information data (machine_data) ( S20) and; A state diagnosis information output step (S30) of outputting state diagnosis information of the machine tool when the machine tool information data (machine_data) is input to the learned state diagnosis model 30; In the learning step (S20), machine information data (machine_data) for which the state determination result is determined to be “normal” is learned, and in the state diagnosis information output step (S30); the state diagnosis model 30 is machine information data ( It relates to an intelligent monitoring and diagnosis method of an IoT-based machine tool, characterized in that it outputs an "abnormality score" by taking machine_data) as an input.

Description

IoT based intelligent monitoring and diagnosis method and device for machine tool {Iot based intelligent monitoring and diagnosis method and device for machine tool}

The present invention relates to an apparatus and method for intelligent monitoring and diagnosis of an IoT-based machine tool.

Registration Patent No. 10-1343403 Name of Invention: Abnormality detection method during machine tool operation is a method for detecting abnormality during machine tool operation that detects poor material seating, tool wear and damage during material processing, A current sensor and a voltage sensor are respectively installed on the power line of the motor driving unit and connected to the sensor processor, the sensor processor is connected to the monitoring controller, and the monitoring controller is connected to the host computer and the machine tool controller, respectively. A preparation step of setting the type and processing method; While processing a product while the output of the machine tool is normal, the drive voltage and drive current required for product processing are measured by the current sensor and the voltage sensor, respectively, to obtain a reference waveform that serves as a standard for determining tool abnormalities. a step of obtaining a reference waveform; A monitoring interval setting step of automatically calculating a maximum load value and a minimum load value by setting a monitoring interval for the reference waveform obtained in the reference waveform acquisition step; a tolerance setting step of setting maximum and minimum allowable limits so that an area in which an alarm occurs is set based on the maximum load value and the minimum load value calculated in the monitoring section setting step; The processing load generated during actual processing of the material is continuously acquired, and it is determined whether the difference between the maximum load value and the minimum load value among the obtained processing loads within the monitoring section is outside the maximum allowable limit or the minimum allowable limit value. A monitoring step for outputting whether or not it is detected, and the monitoring step includes a monitoring data storage step for storing data according to machine tool operation.

NC and CNC machine tools perform unmanned processing and pre-determined tasks by themselves through automation programs. Even when a machine tool is in an abnormal state, such as a tool being damaged, a damaged tool may collide with a workpiece or a machine body due to an automation program, resulting in great damage.

So, as in the prior art described above, the processing load generated during actual processing of the material is continuously obtained, and the difference between the maximum load value and the minimum load value among the obtained processing loads within the monitoring interval is the maximum allowable limit or the minimum allowable limit A technique of determining whether the value is out of range and outputting whether or not it is normal is being used.

However, if the difference in load between the spindle motor during normal and abnormal machining is smaller than the change in load on the spindle feed axis during normal machining due to changes in work per hour, such as material change, work type (shape), cutting amount, and cutting speed, constant It is difficult to determine an abnormality by exceeding a threshold value.

The present invention is an IoT-based machine tool capable of determining whether or not the machine tool is abnormal in advance by monitoring not only load current and voltage but also machine state information such as rotation speed, vibration, temperature, and sound in a machine tool such as a planer miller. It relates to an intelligent monitoring diagnosis device and method.

The present invention is to solve the learning problem of an artificial intelligence model due to the lack of abnormal data and to provide an intelligent monitoring and diagnosis method and apparatus for an IoT-based machine tool that solves the problem of reducing the accuracy of determining abnormalities due to changes in working conditions. .

The present invention can manage field conditions without time and place restrictions through remote monitoring and automatic control system construction. Through this, it is possible to increase work/equipment operation efficiency at industrial sites. In addition, it is to provide an intelligent monitoring and diagnosis method and device for IoT-based machine tools that can maximize the efficiency of industrial site management system operation by grasping visualization data and equipment status and diagnosis information that managers and workers can easily use in real time. .

In the IoT-based machine tool intelligent monitoring and diagnosis method of the present invention, the machine information collection unit 10 acquires machine information of the machine tool from at least one sensor selected from rotation speed, vibration, sound, temperature, current, voltage, and load. learning data acquisition step (S10) of acquiring learning data including data (machine_data) and state judgment results (normal, abnormal, failure, pre-failure stage, score);

A condition diagnosis model learning step in which the AI learning unit 20 uses the learning data to learn a condition diagnosis model that outputs condition diagnosis information (abnormal score) of the machine tool by inputting machine information data (machine_data) ( S20) and; It is characterized in that it is configured to include a state diagnosis information output step (S30) of outputting state diagnosis information of the machine tool when the learned state diagnosis model 30 receives machine tool information data (machine_data).

In the intelligent monitoring and diagnosis method of an IoT-based machine tool of the present invention, in the condition diagnosis model learning step (S20), machine information data (machine_data) for which the condition determination result is determined to be “normal” is learned, and the condition diagnosis information is output. In step S30, the condition diagnosis model 30 preferably outputs an "abnormality score" by taking the machine information data (machine_data) as an input.

In the intelligent monitoring and diagnosis method of the IoT-based machine tool of the present invention, the state determination unit 40 determines the state of the machine tool as normal or abnormal based on the state diagnosis information output by the state diagnosis model 30, that is, the abnormal score. State determination step (S40) to determine as; A control alarm step in which the control unit 50 generates a machine tool control signal based on the status determination result (normal or abnormal) determined by the status determination unit 40 or provides a status diagnosis result (alarm) to a server or manager terminal. (S50); is preferably configured to further include.

In the method for intelligent monitoring and diagnosis of an IoT-based machine tool of the present invention, the state determining unit 40 determines a threshold value (C) based on an abnormality score of machine information data (Machine_Data) determined to be in a normal state. ) is set, and the state determining unit 40 determines the current state to be abnormal when the "Anormaly Score" exceeds the threshold value C.

The intelligent monitoring and diagnosis device for an IoT-based machine tool according to the present invention includes machine information data (machine_data) and status of a machine tool obtained from at least one sensor selected from among rotation speed, vibration, sound, temperature, current, voltage, and load. a machine information collection unit 10 that acquires learning data including judgment results (normal, abnormal, failure, pre-failure, score); Learning a condition diagnosis model that outputs condition diagnosis information (abnormal score) of a machine tool by inputting machine information data (machine_data) using learning data, and an AI learning unit 20;

a condition diagnosis model 30 that outputs condition diagnosis information of the machine tool as an "abnormality score" when machine tool information data (machine_data) is input; a state determination unit 40 that determines whether the state of the machine tool is normal or abnormal based on state diagnosis information output from the state diagnosis model 30, that is, "abnormal score"; a control unit 50 that generates a machine tool control signal based on the state determination result (normal or abnormal) determined by the state determination unit 40 or provides a state diagnosis result (alarm) to a server or manager terminal;

At least one of tool information, workpiece material information, processing shape information, cutting amount information per hour, cutting speed information per hour, spindle rotation speed information, feed shaft feed speed, and total operation time information of the work performed by the machine tool a working condition storage unit 60 that stores information about working conditions (work_data); A work condition classification unit 70 that classifies information (task_data) on a given work condition into the same work condition group (class, W_i) when it is within a certain range from one of the existing work condition classification groups (classes); It is characterized in that it is configured to include a.

According to the present invention, in a machine tool such as a planer miller, not only load current and voltage, but also machine state information such as rotation speed, vibration, temperature, and sound are monitored to determine whether or not the machine tool is abnormal in advance Based on IoT An apparatus and method for intelligent monitoring and diagnosis of a machine tool are provided.

According to the present invention, an intelligent monitoring and diagnosis method and device for IoT-based machine tools are provided that solves the problem of learning artificial intelligence models due to abnormal data collapse and reduces the problem of reducing the accuracy of determining whether abnormality occurs due to changes in working conditions. do.

In addition, according to the present invention, it is possible to manage field conditions without time and place restrictions through remote monitoring and automatic control system construction. Through this, it is possible to increase work/equipment operation efficiency at industrial sites. In addition, an intelligent monitoring and diagnosis method and device for IoT-based machine tools that can maximize the operational efficiency of an industrial site management system by grasping visualization data and equipment status and diagnosis information that managers and workers can easily use in real time are provided.

1 is a flow chart of an intelligent monitoring and diagnosis method for an IoT-based machine tool according to the present invention.
2 is a configuration diagram of an intelligent monitoring and diagnostic device for an IoT-based machine tool according to the present invention.
3 is a conceptual explanatory diagram of an abnormality detection method used in the present invention;
4 is MNIST data used in a normal abnormality detection method (Anormaly Detection).
5 is an application example of an abnormality detection method of normal MNIST data.
6 is an explanatory diagram of a test method for an intelligent monitoring and diagnosis method for an IoT-based machine tool according to the present invention.
7 is a result of RMS values of measurement data (vibration) according to FIG. 6;
8 is an Anomaly Detection analysis result of measurement data according to FIG. 6;
9 is a comparison diagram of the RMS value of the measurement data of FIG. 7 versus the results of Anomaly Detection analysis;

Hereinafter, a method and apparatus for intelligent monitoring and diagnosis of an IoT-based machine tool according to the present invention will be described in detail with reference to the accompanying drawings. 1 is a flow chart of an intelligent monitoring and diagnosing method for an IoT-based machine tool according to the present invention, and FIG. 2 is a configuration diagram of an intelligent monitoring and diagnosing device for an IoT-based machine tool according to the present invention.

As shown in FIGS. 1 and 2, the method for intelligent monitoring and diagnosis of an IoT-based machine tool according to an embodiment of the present invention includes a learning data acquisition step (S10), a state diagnosis model learning step (S20), and state diagnosis information output. It is configured to include step S30, state determination step S40, and control alarm step S50.

First, in the learning data acquisition step (S10), the machine information collection unit 10 acquires machine information data (machine_ Data) and state judgment results (normal, abnormal, failure, pre-failure stage, score) are acquired.

In the condition diagnosis model learning step (S20), the condition diagnosis model in which the AI learning unit 20 outputs condition diagnosis information (abnormal score) of the machine tool by inputting machine information data (machine_data) using the learning data. learn In the state diagnosis model learning step (S20), machine information data (machine_data) for which the state determination result is determined to be “normal” is learned.

In the state diagnosis information output step (S30), when the machine tool information data (machine_data) is input to the state diagnosis model 30 that has been learned, state diagnosis information of the machine tool is output. In the condition diagnosis information output step (S30), the condition diagnosis model 30 outputs an "Anormaly Score" by taking machine information data (machine_data) as an input.

In the state determination step (S40), the state determination unit 40 determines the state of the machine tool as normal or abnormal based on the state diagnosis information output by the state diagnosis model 30, that is, the abnormal score. In the control alarm step (S50), the control unit 50 generates a machine tool control signal based on the state determination result (normal or abnormal) determined by the state determination unit 40 or the state diagnosis result ( alarm).

As shown in FIGS. 1 and 2 , in the method for intelligent monitoring and diagnosis of an IoT-based machine tool according to an embodiment of the present invention, the state determination unit 40 determines machine information data (machine_Data) determined to be in a normal state. A threshold value (C) is set based on the abnormality score of . The state determination unit 40 determines the current state to be abnormal when the "Anormaly Score" exceeds the threshold value C.

As shown in FIGS. 1 and 2, in the method for intelligent monitoring and diagnosis of an IoT-based machine tool according to the first embodiment of the present invention, machine information data (machine_data) is machine operation information data (machine_driving data). ) and machine status information data (machine_status data).

The machine operation information data (machine_operation data) may include at least one selected from load information obtained from a current sensor, a voltage sensor, or a load sensor that monitors the load of the main shaft motor or the transfer shaft motor, and the spindle rotation speed information. can

Machine state information data (machine_status data) includes vibration information, temperature information, It may include at least one selected from among sound information. The state determination unit 40 sets a threshold value related to machine operation data based on an abnormality score for each of the load information and the spindle rotation speed information among the machine operation information data (machine_driving data) determined to be in a normal state. C_j, C_1, C_2) are created. In addition, the threshold value (C_j , C_3, C_4, C_5).

Here, the condition diagnosis model 30 outputs each of the abnormal scores (AS_j, AS_1, AS_2, AS_3, AS_4, AS_5) by taking load information, spindle speed information, vibration information, temperature information, and sound information as inputs. The state determination unit 40 receives the load information, the spindle speed information, the vibration information, the temperature information, and the sound information as inputs, and obtains the respective abnormal scores (AS_j, AS_1, AS_2, AS_3, AS_4, AS_5), load information, The respective threshold values (C_i, C_i, C_1, C_2, C_3, C_4) for the spindle speed information, vibration information, temperature information, and sound information are matched and compared. The state determination unit 40 may evaluate that the reliability of the determination increases as the number of types of information in which the abnormality score AS_j exceeds the threshold value C_j increases.

For example, the state determination unit 40 determines that the abnormal score AS_1 output as load information exceeds the threshold value C_1 for matched load information, and the abnormal score AS_3 output as vibration information matches. When the load information critical value (C_3) is exceeded at the same time, it can be estimated with higher reliability that it is an abnormal state.

<Second Embodiment>

As shown in FIG. 2 , the method and apparatus for intelligent monitoring and diagnosis of an IoT-based machine tool according to the second embodiment of the present invention further include a working condition storage unit 60 and a working condition classifying unit 70.

As shown in FIG. 2, the working condition storage unit 60 includes tool information, workpiece material information, processing shape information, cutting amount information per hour, cutting speed information per hour, spindle rotation speed information, Stores information about work conditions including at least one of feed axis feed speed and total operation time information (work_data).

The working condition classification unit 70 classifies information (task_data) on a given working condition into the same working condition group (class, W_i) if it is within a certain range from one of the existing working condition classification groups (classes). .

At this time, as shown in FIG. 2, the AI learning unit 20 learns the machine information data (machine_data) in a state in which it is classified for each working condition group (class, W_i). The state diagnosis model 30 outputs "Anormaly Score (AS_i)" by inputting machine information data (machine_data) for each working condition group (class, W_i).

As shown in FIG. 2, machine information data (machine_data) is at least one selected from load information obtained from a current sensor, a voltage sensor, or a load sensor, spindle rotation speed information, vibration information, temperature information, and sound information. It is machine information data containing one. The AI learning unit 20 learns the machine information data (machine_data) in a state in which it is classified for each working condition group (class, W_i). The state determination unit 40 determines the machine data-related threshold value for each working condition group (class, W_i) based on the load information, spindle speed information, vibration information, temperature information, and sound information determined to be in a normal state ( C_ij, C_i1, C_i2, C_i3, C_i4, C_i5).

The state diagnosis model 30 takes load information, spindle rotation speed information, vibration information, temperature information, and sound information for each work condition group (class, W_i) as input, and each abnormal score (AS_ij, AS_i1, AS_i2, AS_i3, AS_i4 , AS_i5). The state determining unit 40 takes load information, spindle rotation speed information, vibration information, temperature information, and sound information for each working condition group (class, W_i) as inputs, and generates respective abnormal scores (AS_ij, AS_i1, AS_i2, AS_i3, AS_i4 , AS_i5) and the respective threshold values (C_ij, C_i1, C_i2, C_i3, C_i4, C_i5) for load information, spindle rotation speed information, vibration information, temperature information, and sound information are matched and compared. The state determination unit 40 evaluates that the reliability of the determination increases as the number of types of information in which the abnormality score AS_j exceeds the threshold value C_j increases.

3 is a conceptual diagram of an abnormality detection method used in the present invention, FIG. 4 is MNIST data used in a normal abnormality detection method, and FIG. 5 is an abnormality detection method of normal MNIST data (Anormaly Detection). Detection) application example. Figure 6 shows the results and distribution when the artificial intelligence model is trained with only "0" and randomly calculated abnormal scores such as 0, 9, 6, 4, 8, and 3. The contents related to the artificial intelligence abnormality detection method (Anormaly Detection) described in the papers and patent prior documents known before the filing date of the present invention are considered to be described in the specification of the present invention, and the specific description of the artificial intelligence abnormality detection method (Anormaly Detection) As for the detailed description in the specification of the present invention, it cannot be seen that the present invention is obscured for this reason.

6 is a diagram illustrating a test method for an intelligent monitoring and diagnosis method for an IoT-based machine tool according to the present invention, FIG. 7 is an RMS value result of measurement data (vibration) according to FIG. 6, and FIG. 8 is an Anomaly Detection of measurement data according to FIG. As a result of the analysis, FIG. 9 is a comparison diagram of the RMS value of the measurement data of FIG. 7 vs. the Anomaly Detection analysis result.

As shown in FIG. 6, the specifications of the bearing deterioration test apparatus for obtaining data used in the present invention are as follows.

-. Load: radial/axial (50kN), maximum rotational speed: 3000 rpm, acceleration, temperature, torque measurement, load profile application

-. Test content (bearing deterioration test): Measurement time: March 18th to June 17th, total number of data: 20,516, data information, 4-channel vibration signal, storage time: 60 sec, number of test bearings: 18, damage Bearing: 18

FIG. 7 shows the RMS value results of the measured data (vibration) according to FIG. 6 . The vibration level gradually increases and then spikes just before bearing failure. FIG. 8 shows the Anomaly Detection analysis result using the state diagnosis model 30 for the vibration value of FIG. 7 . FIG. 9 is a comparison diagram of RMS values of the measured data of FIG. 7 versus Anomaly Detection analysis results. 18 bearing damage occurred, and the condition diagnosis model 30 and the condition determination unit 40 detected (predicted) bearing damage with an accuracy of about 72%. However, the above-mentioned test and analysis results are only one embodiment (applying the acquired data to a model) for explaining the present invention, and the damage prediction and diagnosis performance using the present invention is based on the above results. It is not limited.

Although the present invention has been described in relation to the preferred embodiments mentioned above, the scope of the present invention is not limited to these embodiments, and the scope of the present invention is defined by the following claims, and the scope equivalent to the present invention It will contain various modifications and variations pertaining to.

The reference numerals described in the claims below are merely to aid understanding of the invention and do not affect the interpretation of the scope of rights, and the scope of rights should not be interpreted narrowly by the reference numerals described.

10: machine information collection unit
20: AI learning unit
30: condition diagnosis model
40: state determination unit
50: control unit
60: working condition storage unit
70: working condition classification unit

Claims (7)

The machine information collection unit 10 acquires machine information data (machine_data) of the machine tool from at least one sensor selected from among rotation speed, vibration, sound, temperature, current, voltage, and load, and a state determination result (normal or abnormal). Learning data acquisition step (S10) of acquiring learning data including failure, pre-failure, score);

A condition diagnosis model learning step in which the AI learning unit 20 uses the learning data to learn a condition diagnosis model that outputs condition diagnosis information (abnormal score) of the machine tool by inputting machine information data (machine_data) ( S20) and;

A state diagnosis information output step (S30) of outputting state diagnosis information of the machine tool when the learned state diagnosis model 30 receives machine tool information data (machine_data);

In the state diagnosis model learning step (S20), machine information data (machine_data) for which the state determination result is determined to be “normal” is learned;

In the state diagnosis information output step (S30), the state diagnosis model 30 takes machine information data (machine_data) as an input and outputs an "abnormality score" of the IoT-based machine tool, characterized in that Intelligent monitoring diagnostic method.
According to claim 1,
a state determination step (S40) in which the state determination unit 40 determines the state of the machine tool as normal or abnormal based on the state diagnosis information output from the state diagnosis model 30, that is, the abnormal score;

A control alarm step in which the control unit 50 generates a machine tool control signal based on the status determination result (normal or abnormal) determined by the status determination unit 40 or provides a status diagnosis result (alarm) to a server or manager terminal. (S50);
Intelligent monitoring and diagnosis method for IoT-based machine tools, characterized in that configured to further include.
According to claim 2,
The state determining unit 40 sets a threshold value C based on an abnormality score of the machine information data (Machine_Data) determined to be in a normal state;

The state determination unit 40 determines the current state as abnormal when the "Anormaly Score" exceeds the threshold value C, characterized in that the IoT-based machine tool intelligent monitoring diagnosis device diagnostic method.
According to claim 3,
The machine information data (machine_data) is
Machine operation information data (machine_operation data) including at least one selected from among load information acquired from a current sensor, a voltage sensor, or a load sensor that monitors the load of the main shaft motor or transfer shaft motor, respectively, and the spindle speed information ,

A machine state including at least one selected from vibration information, temperature information, and sound information obtained from a vibration sensor, a temperature sensor, and an acoustic sensor installed adjacent to a spindle head, a machine tool structure (body), and an acoustic sensor installed adjacent to a machine tool structure (body) It is composed of information data (machine_state data),

The state determination unit 40 sets a threshold value related to the machine operation data based on an abnormality score for each of the load information and the spindle speed information among the machine operation information data (machine_driving data) determined to be in a normal state. (C_j, C_1, C_2) is generated, and an abnormality score for each of vibration information, temperature information, and sound information among machine state information data (machine_state data) determined to be in a normal state is obtained. Based on the machine state data related thresholds (C_j, C_3, C_4, C_5) are generated,

The condition diagnosis model 30 outputs each abnormal score (AS_j, AS_1, AS_2, AS_3, AS_4, AS_5) by taking load information, spindle speed information, vibration information, temperature information, and sound information as inputs,

The state determination unit 40 receives load information, spindle speed information, vibration information, temperature information, and sound information as inputs, and each of the abnormal scores (AS_j, AS_1, AS_2, AS_3, AS_4, AS_5) and load information , Each threshold value (C_i, C_i, C_1, C_2, C_3, C_4) for spindle rotation speed information, vibration information, temperature information, and sound information is matched and compared, respectively,

The state determination unit 40 evaluates that the reliability of determination increases as the number of types of information in which the abnormal score AS_j exceeds the threshold value C_j increases, characterized in that for intelligent monitoring diagnosis of IoT-based machine tools. method.
According to claim 3,
At least one of tool information, workpiece material information, processing shape information, cutting amount information per hour, cutting speed information per hour, spindle rotation speed information, feed shaft feed speed, and total operation time information of the work performed by the machine tool a working condition storage unit 60 that stores information about working conditions (work_data);

A work condition classification unit 70 that classifies information (task_data) on a given work condition into the same work condition group (class, W_i) when it is within a certain range from one of the existing work condition classification groups (classes); It is composed of further including,

The AI learning unit 20 learns the machine information data (machine_data) in a state in which it is classified for each working condition group (class, W_i),

The state diagnosis model 30 is an IoT-based machine tool, characterized in that it outputs an "Anormaly Score (AS_i)" by inputting machine information data (machine_data) for each working condition group (class, W_i). Intelligent monitoring diagnostic method.
According to claim 4,
At least one of tool information, workpiece material information, processing shape information, cutting amount information per hour, cutting speed information per hour, spindle rotation speed information, feed shaft feed speed, and total operation time information of the work performed by the machine tool a working condition storage unit 60 that stores information about working conditions (work_data);

A work condition classification unit 70 that classifies information (task_data) on a given work condition into the same work condition group (class, W_i) when it is within a certain range from one of the existing work condition classification groups (classes); It is composed of further including,

The machine information data (machine_data) includes load information obtained from a current sensor, a voltage sensor, or a load sensor, and at least one selected from spindle speed information, vibration information, temperature information, and sound information. ego,

The AI learning unit 20 learns the machine information data (machine_data) in a state in which it is classified for each working condition group (class, W_i),

The state determination unit 40 sets a threshold value related to machine data for each working condition group (class, W_i) based on the load information, spindle speed information, vibration information, temperature information, and sound information determined to be in a normal state. create (C_ij, C_i1, C_i2, C_i3, C_i4, C_i5),

The state diagnosis model 30 takes load information, spindle speed information, vibration information, temperature information, and sound information for each working condition group (class, W_i) as inputs, and each abnormal score (AS_ij, AS_i1, AS_i2, AS_i3, AS_i4, AS_i5) output,

The state determination unit 40 receives load information, spindle speed information, vibration information, temperature information, and sound information for each working condition group (class, W_i) as inputs, and sets each abnormality score (AS_ij, AS_i1, AS_i2, AS_i3, AS_i4, AS_i5) and each threshold value (C_ij, C_i1, C_i2, C_i3, C_i4, C_i5) for load information, spindle speed information, vibration information, temperature information, and sound information are matched and compared, respectively,

The state determination unit 40 evaluates that the reliability of determination increases as the number of types of information in which the abnormal score AS_j exceeds the threshold value C_j increases, characterized in that for intelligent monitoring diagnosis of IoT-based machine tools. method.
Machine information data (machine_data) and state judgment result (normal, abnormal, failure, pre-failure stage, score) of the machine tool obtained from at least one sensor selected from among rotation speed, vibration, sound, temperature, current, voltage, and load A machine information collection unit 10 that acquires learning data including;

Learning a condition diagnosis model that outputs condition diagnosis information (abnormal score) of a machine tool by inputting machine information data (machine_data) using learning data, and an AI learning unit 20;

a condition diagnosis model 30 that outputs condition diagnosis information of the machine tool as an "abnormality score" when machine tool information data (machine_data) is input;

a state determination unit 40 that determines whether the state of the machine tool is normal or abnormal based on state diagnosis information output from the state diagnosis model 30, that is, "abnormal score";

a control unit 50 that generates a machine tool control signal based on the state determination result (normal or abnormal) determined by the state determination unit 40 or provides a state diagnosis result (alarm) to a server or manager terminal;

At least one of tool information, workpiece material information, processing shape information, cutting amount information per hour, cutting speed information per hour, spindle rotation speed information, feed shaft feed speed, and total operation time information of the work performed by the machine tool a working condition storage unit 60 that stores information about working conditions (work_data);

A work condition classification unit 70 that classifies information (task_data) on a given work condition into the same work condition group (class, W_i) when it is within a certain range from one of the existing work condition classification groups (classes); Intelligent monitoring and diagnostic device for IoT-based machine tools, characterized in that configured to include.