Method and apparatus for detecting machining abnormality of machine tool

Abstract

To realize highly accurate machining abnormality detection without human intervention by automatically selecting a machining abnormality detection algorithm / machining abnormality detection model for accurately determining a machining abnormality.SOLUTION: A machining abnormality detection method for a machine tool acquires image data of a tool and a work material in a machine tool by an imaging device provided in a machine tool (S11), estimates types of the tool and the work material from the acquired image data (S14 and S15), selects a machining abnormality model based on the estimated types of the tool and the work material (S16) and inputs machining data to the selected abnormality detection model to determine presence or absence of the machining abnormality (S17 and S18).SELECTED DRAWING: Figure 4

Description

Technical Field

[0001] The present disclosure relates to a method and an apparatus for detecting machining anomalies from machining data of a machine tool.

Background Art

[0002] There is a technique for determining machining anomalies using machining data related to machine control obtained from a numerical control device. Machining data refers to information such as the rotational command value and position command of a motor, and torque data of a motor that controls a machine, which changes as machining progresses. An algorithm obtained by analyzing machining data in advance or a machine learning, probabilistic, or statistical model that has learned normal machining data is loaded to predict the machining data, and the difference between the predicted value and the actual value is obtained as a prediction error. The prediction error may be data for the number of machining data used, or may be a single norm. The prediction error is compared with a preset threshold value. If the prediction error is greater than the threshold value, it is regarded as a machining anomaly, and machining is interrupted or the machine is stopped. If the prediction error is smaller than the threshold value, it is regarded that there is no machining anomaly, and machining is continued. When using the model, in order to improve the detection accuracy of machining anomalies, it may be necessary to prepare a plurality of models that have learned machining data under specific machining conditions and select a model. Machining conditions are determined by the type of tool, the type of workpiece material, etc. On the other hand, there is a technique for obtaining the shape of a tool or a workpiece from image data of the tool or the workpiece obtained by imaging with a camera (see, for example, Patent Document 1). An area indicating the characteristics of the tool or the workpiece is extracted from the image, its shape, etc. are obtained, the dimensions of the shape are calculated from the distance between the camera and the tool or the workpiece, and the shape of the actual size is obtained. This is realized by using an algorithm for image processing or a machine learning, probabilistic, or statistical model that has learned the image data of the tool or the workpiece. Although the model shown here is also one of the algorithms, in the present disclosure, the deductive algorithm obtained from the theoretical formula is called an algorithm, and the algorithm inductively created based on data is called a model. The algorithm for image processing is called an image processing algorithm, and the algorithms and models for detecting processing abnormalities are called a processing abnormality detection algorithm and a processing abnormality detection model, respectively.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] When using a model for detecting processing abnormalities, the detection accuracy often has a trade-off with the generalization performance of the model. When creating a model for detecting processing abnormalities under many conditions, it is necessary to pre-learn a lot of processing data. Although a certain degree of effectiveness is obtained for the detection of processing abnormalities under each condition, when high detection accuracy is required, a method of switching to a model that predicts well for specific processing conditions may be selected. In this case, since it is necessary to manually select a model or describe the model switching in the processing program, the processing process cannot be automated without labor and the possibility of mistakes. However, on the other hand, when using a model with high generalization performance but low detection accuracy, processing abnormalities may be over-detected or missed, so the machine cannot be automatically operated while maintaining the processing quality. This is the same when using a processing abnormality detection algorithm.

[0005] Therefore, an object of the present disclosure is to provide a method and apparatus for detecting processing abnormalities of a machine tool that can realize highly accurate detection of processing abnormalities without human intervention by automatically selecting a processing abnormality detection algorithm / processing abnormality detection model for accurately determining processing abnormalities.

Means for Solving the Problem

[0006] In order to achieve the above object, a first configuration of the present disclosure is a machining abnormality detection method for detecting a machining abnormality from machining information acquired during machining in a machine tool that performs machining based on a numerical control command, an image data acquisition step of acquiring image data by imaging a tool and / or a workpiece in the machine tool with an imaging device provided in the machine tool; a tool / workpiece estimation step of estimating the type of the tool and / or the workpiece from the acquired image data; an algorithm / model selection step of selecting a machining abnormality detection algorithm or a machining abnormality detection model based on the estimated type of the tool and / or the workpiece; an abnormality determination step of inputting the machining information into the selected machining abnormality detection algorithm or the aforesaid a machining abnormality detection model to determine the presence or absence of a machining abnormality; and executing together with In the tool / workpiece estimation step, when estimating the type of the workpiece, the workpiece is operated to estimate the weight, and the type of the workpiece is estimated from the size and shape of the workpiece obtained in the image data acquisition step and the estimated weight of the workpiece characterized in that. Another aspect of the first configuration is that, in the above configuration, in the algorithm / model selection step, based on the estimated type of the tool and / or the workpiece, machining conditions that are pre-associated are discriminated, and based on the discriminated machining conditions, the machining abnormality detection algorithm or the aforesaid a machining abnormality detection model is selected. Another aspect of the first configuration is that, in the above configuration, in the abnormality determination step, the machining information is input into the machining abnormality detection algorithm or the aforesaid a machining abnormality detection model to obtain the degree of machining abnormality and compare the degree of abnormality with a predetermined threshold value, and when the degree of abnormality is greater than the threshold value, it is determined that there is a machining abnormality. Another aspect of the first configuration is that, in the above configuration, when it is determined that there is a machining abnormality in the abnormality determination step, the machine tool is caused to replace the currently used tool with a spare tool of the same type. Another aspect of the first configuration is that, in the above configuration, when replacing the spare tool, the imaging device captures an image of the spare tool to obtain image data, and from the obtained image data, a tool discrimination step of determining whether the spare tool is of the same type as the working tool is further executed. If it is determined in the tool discrimination step that they are not of the same type, the processing is interrupted or stopped.

[0007] In order to achieve the above object, a second configuration of the present disclosure is a machining abnormality detection device that detects machining abnormalities from machining information obtained during machining in a machine tool that performs cutting machining based on a numerical control command, image data acquisition means for acquiring image data by imaging a tool and / or a workpiece within the machine using an imaging device provided in the machine tool; tool / workpiece estimation means for estimating the type of the tool and / or the workpiece from the acquired image data; algorithm / model selection means for selecting a machining abnormality detection algorithm or machining abnormality detection model based on the estimated type of the tool and / or the workpiece; abnormality determination means for inputting the machining information into the selected machining abnormality detection algorithm or the aforesaid machining abnormality detection model to determine the presence or absence of machining abnormalities; characterized by comprising together with When the tool / workpiece estimation means estimates the type of the workpiece, the workpiece is operated to estimate the weight, and the type of the workpiece is estimated from the size and shape of the workpiece obtained by the image data acquisition means and the estimated weight of the workpiece this. Another aspect of the second configuration is that, in the above configuration, the algorithm / model selection means discriminates machining conditions associated in advance based on the estimated type of the tool and / or the workpiece, and based on the discriminated machining conditions, the machining abnormality detection algorithm or the aforesaid machining abnormality detection model is selected. Another aspect of the second configuration is that, in the above configuration, the abnormality determination means inputs the machining information into the machining abnormality detection algorithm or the aforesaid machining abnormality detection model to obtain the degree of machining abnormality and compares the degree of abnormality with a predetermined threshold value, and determines that there is a machining abnormality when the degree of abnormality is greater than the threshold value. Another aspect of the second configuration is that, in the above configuration, when the abnormality determination means determines that there is a machining abnormality, the machine tool is caused to replace the currently used tool with a spare tool of the same type. Another aspect of the second configuration is that, in the above configuration, when the replacement with the spare tool is performed, the imaging device images the spare tool to acquire image data, and further includes a tool discrimination means for discriminating whether the spare tool is of the same type as the used tool from the acquired image data. When the tool discrimination means discriminates that they are not of the same type, the machining is interrupted or stopped.

[0008] In the present disclosure, the type of tool refers to those classified by material, shape, machining use such as drill or end mill, or the form of the cutting edge such as solid or insert. Further, in the present disclosure, it is assumed that it is known in advance what kind of machining conditions the machining abnormality detection model is effective for. Also, it is assumed that the correspondence between the combination of the type of tool and the type of work material and the machining conditions is known in advance. The type of tool is specified from the image data of the tool imaged by an imaging device such as a camera. A model is learned and created by inputting in advance the image data of various tools, detecting the image area of the tool, and outputting the estimated values of the diameter, length, and material of the tool. On the other hand, a scale for converting to the actual dimensions based on the positional relationship between the camera and the tool is obtained for the dimensions of the diameter and length of the tool on the image obtained from the detected image area of the tool. The imaged image of the tool is input to the model to obtain the estimated values of the material, diameter, length, tip angle, and twist angle of the tool. Then, the scale is multiplied by the diameter and length to convert to the actual material and dimensions of the tool. Since the material of the tool can be specified by the light reflection condition and brightness, it can be estimated from the image data.

[0009] It is difficult to identify the material of the workpiece only from its appearance. Therefore, the material is identified from the image data of the workpiece captured by an imaging device such as a camera and the weight of the workpiece. A model is learned and created by inputting in advance the image data of workpieces of various sizes and materials, detecting the image area of the workpiece, and outputting estimated values of the size and material of the workpiece. On the other hand, a scale for converting the size of the workpiece on the image obtained from the detected image area of the workpiece to the actual size is obtained based on the positional relationship between the camera and the workpiece. On the other hand, the axis of the machine on which the workpiece is mounted is operated, and the weight of the workpiece is obtained from the inertial force. The captured image of the workpiece is input into the model to obtain estimated values of the size and material of the workpiece. The estimated size is multiplied by the scale to convert it to the actual size of the workpiece. The specific gravity of the workpiece is obtained from the estimated size and weight, and the material of the workpiece is identified from the estimated value of the material and the specific gravity.

Advantages of the Invention

[0010] According to the present disclosure, the type of the tool and / or the workpiece can be estimated from the image data of the imaging device, and a machining abnormality detection algorithm that accurately determines machining abnormalities from the estimation result or The machining abnormality detection model can be automatically selected. Therefore, highly accurate machining abnormality detection without human intervention can be realized. Selected algorithm or The model determines the presence or absence of machining abnormalities based on the input of machining information, and when there are no machining abnormalities, it determines whether to continue machining, so that the machining intended by the user can be automatically continued. In addition, since it is not necessary to describe a command to replace the machining abnormality detection algorithm or in the machining program, the machining program can be reused. From the above, the machining abnormality detection algorithm or and the machining abnormality detection model can be automatically selected, and the machining according to the user's intention can be automatically continued. Therefore, the unmanned operation of the factory and the increase in the operating rate of the machine can be promoted. In addition, an algorithm used for manually detecting machining abnormalities or prevents incorrect selection of the model, and the machining abnormality detection algorithmor It is possible to prevent wear of tools and workpieces and damage to machine parts caused by misselection of the machining abnormality detection model.

Brief Description of the Drawings

[0011]

Figure 1

Figure 2

Figure 3

Figure 4

Modes for Carrying Out the Invention

[0012] Hereinafter, embodiments of the present disclosure will be described with reference to the drawings. First, a conventional machining abnormality detection method and device for a machine tool will be described with reference to FIGS. 1 and 2. FIG. 1 shows the configuration of a conventional machining abnormality detection device 10 for a machine tool. Here, it includes an NC device 11 provided in the machine tool and a machining data storage unit 12 that stores data on the mechanical operations of the machine tool during machining. Further, the machining abnormality detection device 10 includes a model selection screen 13 for selecting a machining abnormality detection model for detecting machining abnormalities, a model selection unit 14 for selecting and reading the model data of the name selected on the screen, a selected model storage unit 15 for storing the read model, and an abnormality degree calculation unit 16 for inputting the machining data stored in the machining data storage unit 12 into the model stored in the selected model storage unit 15 to obtain a predicted value of machining abnormalities. Furthermore, the machining anomaly detection device 10 includes a threshold setting unit 17 that sets a threshold for determining machining anomalies, a threshold storage unit 18 that stores the threshold set by the threshold setting unit 17, and a machining anomaly determination unit 19 that compares the anomaly degree obtained by the anomaly degree calculation unit 16 with the threshold stored in the threshold storage unit 18 and determines that a machining anomaly has occurred when the anomaly degree is greater than the threshold.

[0013] FIG. 2 shows the flow of a conventional machining anomaly detection method. A model used for detecting machining anomalies is selected from the model selection screen 13 and stored in the model selection unit 14 (S1), and the NC device 11 acquires data related to the machine operation (machining data) stored in the machining data storage unit 12 according to the progress of machining (S2). Then, the machining data is input into the model to obtain the degree of machining anomaly as the anomaly degree (S3), and the threshold stored in the threshold storage unit 18 is compared with the anomaly degree (S4). When the anomaly degree is less than or equal to the threshold, machining is continued, and when the anomaly degree is greater than the threshold, the NC device 11 is commanded to interrupt or stop machining.

[0014] Next, the machining anomaly detection method and device of the machine tool according to the present disclosure will be described with reference to FIGS. 3 and 4. FIG. 3 shows an example of a machining anomaly detection device. In this machining anomaly detection device 20, the model selection screen 13 and the machining anomaly determination unit 19 are not provided with respect to the conventional machining anomaly detection device 10 shown in FIG. 1. Here, an in-machine imaging unit 21, an imaging data storage unit 22, a tool estimation unit 23, a weight estimation unit 24, and a work material estimation unit 25 are provided. The in-machine imaging unit 21 includes an imaging device (here, a camera) and images the tool and the work material inside the machine. The imaging data storage unit 22 stores the image data of the tool and the work material imaged by the in-machine imaging unit 21. The in-machine imaging unit 21 and the imaging data storage unit 22 are an example of the image data acquisition means of the present disclosure. The tool estimation unit 23 estimates the type of tool from the image data of the tool stored in the imaging data storage unit 22. The tool estimation unit 23 is an example of the tool / work material estimation means of the present disclosure. The weight estimation unit 24 operates the work material according to the command of the NC device 11 and estimates the weight of the work material from the inertial force. The workpiece type estimating unit 25 estimates the type of the workpiece from the weight estimated by the weight estimating unit 24 and the image data of the workpiece stored in the photographed data storage unit 22. The weight estimating unit 24 and the workpiece type estimating unit 25 are examples of the tool / workpiece estimating means of the present disclosure.

[0015] Furthermore, the machining abnormality detection device 20 includes a tool estimation result storage unit 26, a workpiece type estimation result storage unit 27, a machining condition determination unit 28, a spare tool replacement determination unit 29, and a tool confirmation unit 30. The tool estimation result storage unit 26 stores information on the type of the tool estimated by the tool estimation unit 23. The workpiece type estimation result storage unit 27 stores information on the type of the workpiece estimated by the workpiece type estimating unit 25. The machining condition determination unit 28 determines the machining conditions necessary for model selection from the information on the type of the tool stored in the tool estimation result storage unit 26 and the information on the type of the workpiece stored in the workpiece type estimation result storage unit 27.

[0016] The model selection unit 14 selects a model used for detecting machining abnormalities based on the machining conditions determined by the machining condition determination unit 28. The machining condition determination unit 28 and the model selection unit 14 are examples of the algorithm / model selection means of the present disclosure. The spare tool replacement determination unit 29 compares the abnormality degree obtained by the abnormality degree calculation unit 16 with the threshold value stored in the threshold value storage unit 18, determines that the machining quality cannot be maintained when the abnormality degree becomes larger than the threshold value, and determines that it is necessary to replace the current tool with the spare tool. The abnormality degree calculation unit 16 and the spare tool replacement determination unit 29 are examples of the abnormality determination means of the present disclosure. The tool confirmation unit 30 estimates the type of the spare tool replaced with the current tool and checks whether the type of the current tool is the same as that of the tool. The tool confirmation unit 30 is an example of the tool discrimination means of the present disclosure.

[0017] FIG. 4 shows the flow of the machining abnormality detection method of the present disclosure. First, the tool and the workpiece mounted in the machine are imaged by the camera of the in-machine imaging unit 21, and the image data is stored in the imaging data storage unit 22 (S11: image data acquisition step). Also, the weight estimation unit 24 operates the workpiece according to the command of the NC device 11 to estimate the weight of the workpiece from the inertial force (S12). Further, the NC device 11 acquires the machining data stored in the machining data storage unit 12 according to the progress of machining (S13). Next, the tool estimation unit 23 estimates the type of the tool from the image data of the tool stored in the imaging data storage unit 22 and stores the type of the tool in the tool estimation result storage unit 26 (S14: tool / workpiece estimation step). On the other hand, the workpiece estimation unit 25 estimates the type of the workpiece from the weight of the workpiece estimated by the weight estimation unit 24 and the image data of the workpiece stored in the imaging data storage unit 22, and stores the type of the workpiece in the workpiece estimation result storage unit 27 (S15: tool / workpiece estimation step).

[0018] Next, the machining condition determination unit 28 determines the machining conditions necessary for model selection from the estimated tool type and workpiece type, and the model selection unit 14 selects a model used for detecting machining abnormalities based on the determined machining conditions (S16: algorithm / model selection step). Next, the abnormality degree calculation unit 16 inputs the machining data stored in the machining data storage unit 12 into the selected model to obtain the abnormality degree of machining (S17: abnormality determination step), and the spare tool replacement determination unit 29 compares the threshold value stored in the threshold value storage unit 18 with the abnormality degree (S18: abnormality determination step). Here, if the abnormality degree is less than or equal to the threshold value, the process returns to S13 to continue machining, and if the abnormality degree is greater than the threshold value, the current tool and the spare tool are exchanged (S19). Next, the spare tool is imaged by the camera of the in-machine imaging unit 21, and the image data is stored in the imaging data storage unit 22 (S20), and the tool type is estimated by the tool estimation unit 23 from the image data of the spare tool (S21). Then, the tool confirmation unit 30 determines whether the types of the current tool and the spare tool are the same (S22: tool determination step). If the types of the tools are the same, the process returns to S13 to continue machining, and if the types of the tools are different, the NC device 11 is commanded to interrupt or stop the machining.

[0019] As described above, the machining anomaly detection method and the machining anomaly detection device 20 of the machine tool in the above-described embodiment image the tool and the workpiece within the machine tool by the imaging device provided in the machine tool to acquire image data, estimate the types of the tool and the workpiece from the acquired image data, select a machining anomaly detection model based on the estimated types of the tool and the workpiece, and input machining data into the selected machining anomaly detection model to determine the presence or absence of a machining anomaly.

[0019] According to this configuration, the types of the tool and the workpiece can be estimated from the image data of the camera and the operation of the machine, and a machining anomaly detection model for accurately determining machining anomalies can be automatically selected from the estimation results. Therefore, highly accurate machining anomaly detection without human intervention can be realized. The selected model can automatically continue the machining intended by the user according to the threshold value by determining the continuation of machining by comparison with the threshold value. Further, since it is not necessary to describe a command for exchanging the machining anomaly detection model in the machining program, the machining program can be reused. When the degree of anomaly exceeds the threshold value, replacement with a spare tool is performed. As described above, it is possible to automatically estimate the machining conditions and automatically continue the machining according to the user's intention. Therefore, it is possible to promote the unmanned operation of the factory and increase the operating rate of the machine. Also, it is possible to prevent an incorrect selection of the model used for detecting machining anomalies by hand, and to prevent wear of tools and workpieces and damage to machine parts caused by an incorrect selection of the machining anomaly detection model.

[0020] Incidentally, as an example of an image processing algorithm for obtaining the dimensions, size, and shape of a tool or a workpiece from a captured image in the present disclosure, there is a deep learning method, which is an inductive algorithm. As a method for specifying the imaging region of a tool or a workpiece from the captured image inside the machine and detecting it as a tool or a workpiece, there is Region CNN, and there is a generation query network for generating a three-dimensional shape from the captured image. Also, when selecting processing conditions for selecting a machining anomaly detection model, if there is a model with generalization performance for any workpiece material type for a certain tool type, or if there is a model with generalization performance for any tool type for a certain workpiece material type, it may be determined which processing conditions to select the model from at least one of specifying the tool type or specifying the workpiece material type. For the tool and workpiece material types and the corresponding processing conditions, the recommended processing conditions of the tool manufacturer or the processing conditions set by the user may be used. Examples of a machining anomaly detection algorithm and a machining anomaly detection model used for detecting machining anomalies are as follows. Regarding the algorithm, the harmonic component of the nth rotation is subtracted from the motor torque of the tool spindle and used as the machining component torque for anomaly detection. The machining component torque during abnormal machining has a different tendency and magnitude from that during normal machining, so detection using a threshold value is possible. In the above embodiment, the presence or absence of machining anomalies is determined using a machining anomaly detection model, but instead of the machining anomaly detection model, the presence or absence of machining anomalies may be determined using such a machining anomaly detection algorithm. Regarding the model, there is a neural network or the like that infers and outputs future input values for the input used for anomaly detection.

[0021] In the above form, when the anomaly degree becomes larger than the threshold value in the anomaly degree calculation unit, the current tool and the spare tool are exchanged, but instead of exchanging the tool, it may be determined as an anomaly as it is and the machining may be interrupted or stopped.

Explanation of Signs

[0022] 10, 20 ··· Processing abnormality detection device, 11 ··· NC device, 12 ··· Processing data storage unit, 13 ··· Model selection screen, 14 ··· Model selection unit, 15 ··· Selected model storage unit, 16 ··· Abnormality degree calculation unit, 17 ··· Threshold setting unit, 18 ··· Threshold storage unit, 19 ··· Processing abnormality discrimination unit, 21 ··· In-machine imaging unit, 22 ··· Imaging data storage unit, 23 ··· Tool estimation unit, 24 ··· Weight estimation unit, 25 ··· Workpiece material estimation unit, 26 ··· Tool estimation result storage unit, 27 ··· Workpiece material estimation result storage unit, 28 ··· Processing condition discrimination unit, 29 ··· Spare tool change determination unit, 30 ··· Tool confirmation unit.

Claims

1. In a machine tool that performs machining based on numerical control commands, a machining abnormality detection method for detecting machining abnormalities from machining information acquired during machining, comprising: an image data acquisition step of acquiring image data by imaging a tool and / or a workpiece in the machine tool with an imaging device provided in the machine tool; a tool / workpiece estimation step of estimating the type of the tool and / or the workpiece from the acquired image data; an algorithm / model selection step of selecting a machining abnormality detection algorithm or a machining abnormality detection model based on the estimated type of the tool and / or the workpiece; an abnormality determination step of inputting the machining information into the selected machining abnormality detection algorithm or the machining abnormality detection model to determine the presence or absence of machining abnormalities; while executing, in the tool / workpiece estimation step, when estimating the type of the workpiece, the workpiece is operated to estimate its weight, and the type of the workpiece is estimated from the size and shape of the workpiece acquired in the image data acquisition step and the estimated weight of the workpiece. A machining abnormality detection method for a machine tool, characterized by this.

2. In the algorithm / model selection step, based on the estimated type of the tool and / or the workpiece, the machining conditions associated in advance are discriminated, and the machining abnormality detection algorithm or the machining abnormality detection model is selected based on the discriminated machining conditions. The machining abnormality detection method for a machine tool according to Claim 1, characterized by this.

3. In the abnormality determination step, the machining information is input into the machining abnormality detection algorithm or the machining abnormality detection model to obtain the degree of machining abnormality, and the degree of abnormality is compared with a predetermined threshold value. When the degree of abnormality is greater than the threshold value, it is determined that there is a machining abnormality. The machining abnormality detection method for a machine tool according to Claim 1 or 2, characterized by this.

4. When it is determined that there is a machining abnormality in the abnormality determination step, the machine tool is characterized in that the currently used tool is replaced with a spare tool of the same type. The machining abnormality detection method for a machine tool according to any one of Claims 1 to 3.

5. When replacing with the spare tool, the imaging device images the spare tool to obtain image data, and further executes a tool determination step of determining whether the spare tool is of the same type as the working tool from the obtained image data. If it is determined in the tool determination step that they are not of the same type, the machining is interrupted or stopped. The machining abnormality detection method for a machine tool according to claim 4, characterized in that.

6. In a machine tool that performs cutting machining based on a numerical control command, a machining abnormality detection device that detects machining abnormalities from machining information obtained during machining, Image data acquisition means for imaging a tool and / or a workpiece within the machine by an imaging device provided in the machine tool to obtain image data; Tool / workpiece estimation means for estimating the type of the tool and / or the workpiece from the obtained image data; Algorithm / model selection means for selecting a machining abnormality detection algorithm or a machining abnormality detection model based on the estimated type of the tool and / or the workpiece; Abnormality determination means for inputting the machining information into the selected machining abnormality detection algorithm or the machining abnormality detection model to determine the presence or absence of machining abnormalities; and is provided with When the tool / workpiece estimation means estimates the type of the workpiece, the workpiece is operated to estimate the weight, and the type of the workpiece is estimated from the size and shape of the workpiece obtained by the image data acquisition means and the estimated weight of the workpiece. The machining abnormality detection device for a machine tool according to claim 6, characterized in that.

7. The algorithm / model selection means determines machining conditions associated in advance based on the estimated type of the tool and / or the workpiece, and selects the machining abnormality detection algorithm or the machining abnormality detection model based on the determined machining conditions. The machining abnormality detection device for a machine tool according to claim 6, characterized in that.

8. The abnormality determination means inputs the machining information into the machining abnormality detection algorithm or the machining abnormality detection model to obtain the degree of machining abnormality, compares the degree of abnormality with a predetermined threshold value, and determines that there is a machining abnormality when the degree of abnormality is greater than the threshold value. The machining abnormality detection device for a machine tool according to claim 6 or 7, characterized in that.

9. When the abnormality determination means determines that there is a machining abnormality, the machine tool is caused to replace the currently used tool with a spare tool of the same type. The machining abnormality detection device for a machine tool according to any one of claims 6 to 8, characterized in that.

10. When the replacement with the spare tool is performed, the imaging device images the spare tool to acquire image data, and further includes a tool discrimination means for discriminating whether the spare tool is of the same type as the used tool from the acquired image data. When the tool discrimination means discriminates that they are not of the same type, the machining is interrupted or stopped. The machining abnormality detection device for a machine tool according to claim 9, characterized in that.

Images