CN107511718A - Single product high-volume repeats the intelligent tool state monitoring method of process - Google Patents

Abstract

The present invention provides a method for monitoring the state of an intelligent cutting tool in a large-batch repeated processing process of a single product, comprising the following steps: S1, establishing a sample library, and collecting characteristic signal samples of workpiece processing when the cutting tool is in different life stages; S2, multi-sensor signal fusion ; S3. Factors affecting characteristic signals during machine tool processing; S4. Endow tool life monitoring algorithm with self-learning ability; S5. Close, start real-time monitoring when there are still H items from the last setting N. The beneficial effect of the present invention is that misjudgment can be better avoided.

Description

Intelligent Tool Condition Monitoring Method for Single Product Mass Repeated Machining Process

technical field

The invention relates to a tool state monitoring method, in particular to an intelligent tool state monitoring method for repeated processing of a single product in large batches.

Background technique

From its appearance to its development, tool condition monitoring has mainly experienced two development processes:

1) Traditional monitoring stage

In the traditional cutting process, the identification of the tool state is judged by the processing personnel to distinguish the cutting sound, chip color, cutting time, etc., or by measuring the degree of damage and wear after disassembling the tool between processing steps.

2) Intelligent monitoring stage

The so-called intelligent monitoring means that in the process of product processing, the computer predicts the wear and damage status of the tool by detecting the changes of various sensor signals, so as to determine whether the tool needs to be replaced. Commonly used tool monitoring signals include vibration signals, temperature signals, spindle Current signal, acoustic emission signal, etc., the description of tool wear state is shown in Figure 1, which is divided into: initial wear stage, normal wear stage and sharp wear stage.

Initial wear stage: The flank surface of a newly sharpened tool often has defects such as roughness, microcracks, oxidation, etc., and the cutting edge is relatively sharp. During the cutting process, the contact area between the flank surface and the machined surface is relatively small, and the contact pressure Larger, therefore, the initial wear time is shorter.

Normal wear stage: When entering normal wear, the flank of the tool has been basically ground flat, its contact area with the machined surface is larger, the contact pressure becomes smaller, and the wear is uniform and relatively slow. The wear amount of the tool in the normal wear stage is basically proportional to the processing time of the workpiece.

Acute wear stage: When the tool wears to a certain extent, the surface roughness value of the workpiece increases, the cutting force and cutting temperature increase, and the tool wears sharply. The sharp wear stage is usually accompanied by strong vibration and abnormal noise. At this stage, it is necessary to stop the machine in time to replace the tool.

The tool state monitoring system can tell which wear stage the tool is currently in. When it is in a sharp wear stage, it will prompt the CNC system to change the tool.

Traditional tool condition monitoring often relies on the long-term accumulated production experience of skilled workers, so the following problems will inevitably arise:

1) If the tool wear is lower than the blunt standard but has been removed, the actual life of the tool is not fully utilized, resulting in waste and increased processing costs;

2) If the tool wear is higher than the blunt standard, that is, the tool has been worn or damaged, it will affect the surface quality and dimensional accuracy of the workpiece, and even damage the machine tool.

3) It causes a waste of personnel. Now the trend of industrial production is unmanned production. It can no longer meet the needs of modern industrial production to find the wear and damage of the tool through personnel, and how to detect the wear of the tool by disassembling the tool will lead to The pause of processing affects the production efficiency.

Traditional tool status monitoring systems often monitor by setting thresholds. Take the spindle power signal as an example: the spindle power of the machine tool is not constant at every moment during the machining process. When it is in a heavy cutting state, its The cutting power will increase accordingly, so the power of the spindle of the machine tool is in a dynamic process during the machining process, so we can set an appropriate threshold value, and when this value is exceeded, the system will be reminded that the tool wear exceeds the normal limit, and the machine tool will complete the processing. The workpiece should be changed again; at the same time, we also need to pay attention to the possibility of tool chipping during the machining process of the machine tool, which causes more damage than tool wear. We need to set a limit threshold. When the threshold is exceeded, stop immediately to prevent tool damage. It will cause harm to the machine tool itself and the operator.

During the normal processing of the machine tool, the peak value of the machine tool spindle power is stable within a certain range, and there will not be too much jump, and the normal processing level threshold is low; when the machine tool spindle power exceeds the wear threshold, it indicates the wear value of the tool If the machining allowance is exceeded, when the machine tool finishes processing the workpiece, it will stop and change the tool; when the spindle power of the machine tool exceeds the limit threshold, it means that the tool is damaged, and tool wear is a slowly changing process. It itself and the operator cause great harm, so when the spindle power curve exceeds the limit threshold, the tool cannot be changed until the workpiece is processed, and the tool needs to be stopped immediately to avoid serious damage caused by tool damage. So this signal will appear in a kind of abrupt form, as shown in Figure 2 for the later curve of processing.

Although the intelligent tool condition monitoring developed in recent years has got rid of the limitation of personnel and realized intelligent online monitoring, there are still problems of inaccurate monitoring and difficult determination of tool change time, because the existing intelligent monitoring can only distinguish Finding out whether the current tool is in the initial wear stage, normal wear stage or rapid wear stage does not give the specific life of the tool, so it often causes the two problems of insufficient tool use and excessive tool use.

Moreover, the intelligent monitoring of modern tools often leads to misjudgment, because the monitoring system does not know what kind of processing state the tool is currently in, whether it is in high-speed processing or low-speed processing, whether it is heavy cutting or light cutting, the monitoring system has no idea at all. When the tool performs high-speed, large-cut, heavy-volume cutting, its characteristic signal often changes more drastically than when the tool performs low-speed, small-cut, and light-weight cutting. At this time, the tool monitoring system may think that the tool is damaged or has exceeded The standard of tool bluntness is exceeded, and the tool needs to be changed.

Contents of the invention

In order to solve the problems in the prior art, the present invention provides an intelligent tool state monitoring method for a large batch of single product repeated processing.

The invention provides a method for monitoring the state of an intelligent tool in a large-batch repeated processing process of a single product, comprising the following steps:

S1. Establish a sample library to collect characteristic signal samples of workpiece processing when the tool is in different life stages;

S2, multi-sensor signal fusion;

S3. Factors affecting characteristic signals during machine tool processing;

S4. Endow the tool life monitoring algorithm with the ability of self-learning;

S5. In the life monitoring of the tool in the early stage, the monitoring is carried out every m pieces of processing, and the monitoring is turned off during other processing periods. When the monitoring reaches the final setting of N and there are H pieces, the real-time monitoring starts.

As a further improvement of the present invention, in step S1, the characteristic signal includes a spindle current signal, a vibration signal, a sound signal, and a temperature signal.

As a further improvement of the present invention, in step S3, factors affecting the characteristic signal include cutting width W, cutting depth D, and feed speed V.

The beneficial effect of the present invention is: through the above solution, the occurrence of misjudgment can be better avoided.

Description of drawings

Fig. 1 is a schematic diagram of the state of tool wear in the prior art.

Fig. 2 is a graph of a later stage of processing in the prior art.

Fig. 3 is a schematic diagram of multi-sensor signal fusion of an intelligent tool state monitoring method for a large batch of repeated processing of a single product according to the present invention.

FIG. 4 is a schematic diagram of factors affecting characteristic signals during machine tool processing of a method for intelligent tool state monitoring of a single product in large batches of repeated processing in the present invention.

Fig. 5 is a flow chart of establishing a sample library of an intelligent tool state monitoring method for a large-scale repeated processing process of a single product.

Fig. 6 is a training flow chart for inventing a method for monitoring the state of an intelligent tool in a large-batch repetitive machining process for a single product.

detailed description

The present invention will be further described below in conjunction with the description of the drawings and specific embodiments.

As shown in Figure 3 to Figure 6, in view of the phenomenon that more and more modern machine tools are engaged in the repeated processing of products of a single specification in large quantities, in order to improve the accuracy of tool status monitoring and reduce the rate of misjudgment, the self-learning ability is also added , to truly realize the on-line monitoring of the tool state. The monitoring result is not limited to what kind of wear state the tool is in, but can get the remaining number of workpieces that can be processed by the current tool. The invention provides a large-scale repetitive processing of a single product A process intelligent tool condition monitoring method includes the following steps:

1) Establish a sample library to collect characteristic signal samples of workpiece processing when the tool is in different life stages. Such characteristic signals can be spindle current signals, vibration signals, sound signals, temperature signals, etc. It is necessary to find the characteristic signals closely related to the tool wear state from these characteristic signals, and exclude the insensitive signals. Among them, the sample selection must be comprehensive, from the beginning of processing the first workpiece with a new tool to the tool wear to make the processed workpiece unqualified, that is, the sample data of the last workpiece are all needed to ensure the integrity of the sample library. At the same time, the number of samples should be large and can be Avoid the influence of accidental factors on decision-making.

2) For multi-sensor signal fusion, one problem that must be faced squarely is that it is impossible to achieve high accuracy through the monitoring of only one signal, because each signal has its own limitations. Multi-sensor fusion means that the judgment of the tool state does not only depend on one signal, but collects and processes the data of multiple signals. Using each signal has its own advantages, and complementary advantages can get more accurate results. For example, the spindle current signal, vibration signal and processing signal in the process of processing are collected at the same time, and different eigenvalues in the same signal can also reflect the wear state of the tool to varying degrees, so different signal eigenvalues can be analyzed according to them. Different weights are given to them according to the sensitivity of tool wear, and then reasonable monitoring results are obtained. Finally, the state of tool wear is obtained through comprehensive judgment of various eigenvalues.

3) Factors affecting the characteristic signal during machine tool processing: cutting width W, cutting depth D, feed speed V and various external interference factors. What kind of cutting depth, width and feed speed are known, and the normal processing characteristic signal can be seen from the standard sample, which is convenient for the monitoring of tool life, which is also the importance of starting sample collection , the success of sample selection directly determines the accuracy of tool life monitoring.

4) Endow the tool life monitoring algorithm with the ability of self-learning. When a misjudgment occurs, for example, the algorithm judges that the tool cannot continue to process the workpiece, but after careful inspection of the replaced tool, it is found that the tool can continue Processing, which shows that the algorithm is not perfect and needs to be improved. At this time, a penalty mechanism is added to the algorithm itself. When a misjudgment occurs, the algorithm itself makes some adjustments to endow the algorithm with a self-learning ability. The monitoring of the tool life by the algorithm is realized through the training of the algorithm on the samples.

5) If the above-mentioned monitoring method is used to monitor from the first workpiece processed by the tool to the last workpiece processed by the tool, the amount of data processing will be too large, the load on the monitoring system will be too high, and it is unnecessary in actual production. Assuming that a new tool has a given life of N pieces when it comes out of the market, it is meaningless to monitor the remaining number of pieces to be processed before it processes N/2, or even 2N/3, because if the tool does not break at this time, The amount of wear must be within the allowable range, so when the number of workpieces processed by the tool is within 2N/3, the monitoring method can be simplified, and the focus of monitoring should be on whether the tool is damaged. The simplified solution is as follows:

In the life monitoring of the tool in the early stage, monitoring can be carried out for every m pieces processed, and the monitoring is turned off during other processing periods. When the monitoring reaches 10 or 20 pieces from the final setting N, real-time monitoring starts, because at this time the tool Wear and tear begins to occur sharply, and the characteristic signal will also change relatively obviously during this period. At this time, monitoring becomes relatively easy and necessary.

Because the monitoring method of multi-sensor information fusion is adopted, in order to reduce the data processing pressure and save data processing time in the early and middle stages of tool wear, only one sensor parameter can be used for monitoring. When N has 10 or 20 pieces, it starts to use multi-sensor information fusion for monitoring.

The intelligent tool state monitoring method for a large batch of single product repeated processing provided by the present invention has the following advantages:

1) Reduce the misjudgment rate

For large-scale repetitive processing of a single specification product, it is possible to know what kind of processing state the tool is in at all times, whether it is high-speed processing or low-speed processing, whether it is heavy cutting or light cutting, etc., avoiding the severe processing stage of the previous monitoring system. possible misjudgments.

2) Monitoring is more accurate

Compared with the previous tool status monitoring, the remaining number of workpieces processed by the tool can be obtained, which can avoid the failure of the processing of the work cost due to the scrapping of half of the tool. The cost will be greatly increased, and the application of new tool monitoring can reduce the unqualified rate of workpieces, thereby saving costs.

The above content is a further detailed description of the present invention in conjunction with specific preferred embodiments, and it cannot be assumed that the specific implementation of the present invention is limited to these descriptions. For those of ordinary skill in the technical field of the present invention, without departing from the concept of the present invention, some simple deduction or replacement can be made, which should be regarded as belonging to the protection scope of the present invention.

Claims (3)

1. a kind of single product high-volume repeats the intelligent tool state monitoring method of process, it is characterised in that including following step Suddenly：

S1, Sample Storehouse is established, the characteristic signal sample of work pieces process when collection cutter is in different lifetime stages；

S2, multiple sensor signals fusion；

The factor of the excessively middle effect characteristicses signal of S3, machine tooling；

S4, the ability for assigning cutter life monitoring algorithm self study；

S5, in the life-span monitoring of cutter early stage, take every processing m parts once to be monitored, other process during then monitoring Close, start to monitor in real time when monitoring and also having H parts away from last set N.

2. single product high-volume according to claim 1 repeats the intelligent tool state monitoring method of process, its feature It is：In step sl, characteristic signal includes spindle motor current signal, vibration signal, voice signal, temperature signal.

3. single product high-volume according to claim 1 repeats the intelligent tool state monitoring method of process, its feature It is：In step s3, the factor of effect characteristicses signal includes cutting width W, cutting depth D, feed speed V.