Intelligent cloud platform tool wear determination method based on fuzzy neural network

Abstract

The present application relates to a kind of intelligent cloud platform tool wear determination methods based on fuzzy neural network, S1: numerical control system gathers tool and the relevant monitoring data of the tool processing, fixed interval time is stored in numerical control system, when processing is completed or processing alarm, send data system monitoring data to cloud data storage platform;S2: planning cloud storage space, for receiving the data of numerical control system monitoring, according to rule constructs cloud database;S3: with cloud database establishes fuzzy neural network model, the fuzzy neural network model is automatically updated according to fixed interval time;S4: the monitoring data is input into the neural network model that has completed self-learning and self-adaptation, corresponding tool wear result is output after hidden layer and defuzzification module, display on numerical control system.The present application provides accurate and reliable tool wear determination for numerical control operator without high-precision sensor, so as to maintain efficient operation, improve efficiency.

Description

Technical Field

[0001] This invention relates to the field of CNC tool inspection technology, and in particular to a tool wear determination method based on a fuzzy neural network in an intelligent cloud platform. Background Technology

[0002] With the continuous development of automobiles, aerospace, energy, medicine, new materials, and new technologies, cutting tools are trending towards composite, specialized, high-speed, high-precision, and diversified products. In addition to improvements in cutting tool manufacturing technology, regular inspection of tool wear is indispensable.

[0003] In general, tool wear is difficult to measure directly and requires sophisticated instruments and complex measurement methods. Therefore, indirect prediction methods have become a common approach.

[0004] To prevent excessive tool wear and subsequent workpiece damage, in addition to visual inspection, more precise instrument-based detection is also necessary. However, traditional sensors are no longer sufficient for high-precision tool condition monitoring systems. While high-precision sensing instruments can achieve excellent detection results, they are expensive and cumbersome to install, use, and maintain, making them unsuitable for individual businesses and small manufacturers.

[0005] Currently, most methods for predicting tool wear conditions are data-driven. Data-driven methods primarily involve building predictive models, mining operational data during the machining process to uncover the implicit relationship between operational data and tool wear, and thus achieving prediction.

[0006] In 2018, patent application number 201810081967.5, entitled "A tool wear monitoring method based on convolutional neural network", disclosed a tool wear monitoring method based on convolutional neural network, which monitors the tool wear state based on vibration signal; in 2019, patent application number 201910277738.5, entitled "A CNC machine tool wear state prediction method based on parallel deep neural network", disclosed a CNC machine tool wear state prediction method based on parallel deep neural network, which uses multiple sensors to acquire parallel data to establish a deep neural network model.

[0007] Existing tool wear monitoring models are all offline models, designed for a specific tool and not applicable to all tool models. The degree of tool wear is difficult to determine, there are no standardized calculation rules, and the amount of preparation work required to obtain enough data to build a model is too large. The tolerance for tool wear varies for different accuracy requirements, which the above models do not take into account. The above models cannot be updated automatically and are difficult to adapt to new working conditions. Summary of the Invention

[0008] The purpose of this invention is to provide a tool wear determination method for intelligent cloud platforms based on fuzzy neural networks, so as to solve the problems encountered in the above-mentioned background technology.

[0009] To achieve the above objectives, the technical solution of the present invention is as follows:

[0010] A method for determining tool wear in an intelligent cloud platform based on a fuzzy neural network includes the following steps:

[0011] S1: The CNC system collects the tool and related monitoring data of the tool processing, stores it in the CNC system at fixed intervals, and sends the data monitored by the data system to the cloud data storage platform when processing is completed or when a processing alarm is triggered.

[0012] The tool and related monitoring data for its machining include: workpiece material, tool model, machining accuracy, cutting quantity, tool speed, feed rate, spindle, vibration frequency, vibration amplitude, and completion level; wherein, workpiece material, tool model, and machining accuracy serve as data identifiers; cutting quantity, tool speed, feed rate, spindle, vibration frequency, and vibration amplitude serve as input vectors for the model; and completion level serves as the output vector.

[0013] When machining is successfully completed, the completion level is set to 1; when a tool-related alarm occurs, the completion level is set to -1; if machining is interrupted due to other circumstances, no data is sent to the cloud platform. Using the completion level as the output of the prediction model instead of the wear amount makes the data more practical. Relevant data can be collected during daily machining to expand the database and achieve automatic database updates.

[0014] S2: Plan cloud storage space to receive data monitored by the CNC system and build a cloud database according to rules.

[0015] Each workpiece material, tool type, and machining accuracy is used as an identifier to create a separate storage space. All other collected data are stored within the same storage space. If the input variables are the same, the data is merged, and the completion level is accumulated using the formula: sum = sum + k, where k is the accumulation coefficient, and the value of k varies depending on the completion level.

[0016] S3: A fuzzy neural network model is built using a cloud database. This model is automatically updated at fixed intervals. The input vector is first normalized, and then the output vector is fuzzified. The input and output vectors are then fed into the neural network model for training and learning.

[0017] S4: Input the monitoring data into the neural network model that has completed self-learning and self-adaptation, and output the corresponding tool wear results through the hidden layer and defuzzification module, which are then displayed on the CNC system.

[0018] The output tool wear results specifically include the following steps:

[0019] S401. Assign different neural network models to the data collected by the CNC system using identifiers;

[0020] S402. Input the input vector into the neural network model after normalization.

[0021] S403. After normalization, the input vector is processed by the algorithm through the hidden layer in the neural network model.

[0022] S404, the neural network model outputs any real number between (-1, 1), then performs defuzzification processing, and outputs the tool wear determination result.

[0023] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention provides a tool wear determination method based on a fuzzy neural network for an intelligent cloud platform. It provides CNC operators with accurate and reliable tool wear determination without requiring high-precision sensors, thereby maintaining efficient operation and improving efficiency. Furthermore, this invention uses the completion of workpiece machining as the standard for evaluating wear, and combines cloud storage and cloud computing technologies to achieve online data storage and model updates, adapting to new working conditions. Attached Figure Description

[0024] The disclosure of this invention is illustrated with reference to the accompanying drawings. It should be understood that the drawings are for illustrative purposes only and are not intended to limit the scope of protection of this invention. In the drawings, the same reference numerals are used to refer to the same parts. Wherein:

[0025] Figure 1 This is a schematic diagram of the method steps of the present invention;

[0026] Figure 2 This is a schematic diagram of the system architecture of the present invention;

[0027] Figure 3 This is a membership function diagram illustrating the degree of completion of the present invention;

[0028] Figure 4 The cloud computing platform neural network algorithm model learning flowchart provided by this invention;

[0029] Figure 5 A flowchart for situation prediction of a cloud computing platform based on a neural network algorithm provided for this invention. Detailed Implementation

[0030] To make the technical means, creative features, objectives, and effects of this invention readily understandable, the invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the relevant components of the invention.

[0031] According to the technical solution of the present invention, without changing the essential spirit of the present invention, those skilled in the art can propose various interchangeable structural methods and implementations. Therefore, the following detailed embodiments and accompanying drawings are merely exemplary descriptions of the technical solution of the present invention, and should not be regarded as the entirety of the present invention or as a limitation or restriction of the technical solution of the present invention.

[0032] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments.

[0033] like Figure 1 and Figure 2 As shown, a method for determining tool wear in an intelligent cloud platform based on a fuzzy neural network includes the following steps:

[0034] S1: The CNC system collects relevant monitoring data on the cutting tool and its machining process, including: workpiece material, tool type, machining accuracy, cutting quantity, tool speed, feed rate, spindle speed, vibration frequency, vibration amplitude, and completion level. Among these, workpiece material, tool type, and machining accuracy serve as data identifiers; cutting quantity, tool speed, feed rate, spindle speed, vibration frequency, and vibration amplitude serve as the model's input vectors; and completion level serves as the output vector.

[0035] When machining is successfully completed, the completion level is set to 1; when a tool-related alarm occurs, the completion level is set to -1; if machining is interrupted due to other circumstances, no data is sent to the cloud platform. Fixed intervals are stored in the CNC system, and relevant monitoring data is sent to the cloud data storage platform upon completion of machining or when a machining alarm is triggered.

[0036] This invention uses the degree of completion as the output of the prediction model instead of the amount of wear, making the data more practical. Relevant data can be collected during daily processing to expand the database and complete automatic database updates.

[0037] S2: Plan cloud storage space to receive data monitored by the CNC system and build a cloud database according to rules. Establish separate storage spaces based on workpiece material, tool type, and machining accuracy. Within the same storage space, if the input variables are the same (i.e., cutting amount, tool speed, feed rate, vibration frequency, vibration amplitude, and tool axis torque), the data is merged, the input variables remain unchanged, and the completion level is accumulated. The accumulation formula is: sum = sum + k, where k is the accumulation coefficient. The value of k varies depending on the completion level; typically, k is positive when the completion level is 1 and negative when the completion level is -1, with a larger absolute value.

[0038] S3: A fuzzy neural network model is built using a cloud database. This model is automatically updated at fixed intervals. First, the input vector is normalized, and then the completion level of the output vector is fuzzified. The input and output vectors are then fed into the neural network model for training and learning, resulting in a fully trained neural network model.

[0039] S4: Input the monitoring data into the neural network model that has completed self-learning and self-adaptation, and output the corresponding tool wear results through the hidden layer and defuzzification module, which are then displayed on the CNC system.

[0040] Figure 3 This is a membership function graph used in this invention to fuzzify the degree of completion.

[0041] The completion level is normalized to (-1, 1) using a membership function, where b is a set threshold. That is, when the completion level exceeds this value, the output vector will remain at 1 and will not increase further; the same applies to negative values. Figure 3 The membership function graph can be used to fuzzify the degree of completion to avoid the impact of extreme cases on the overall data, such as mass production of specific workpieces or batch damage of tools due to special circumstances.

[0042] Figure 4 A flowchart for establishing a neural network algorithm model learning process is provided for the implementation of this invention.

[0043] Input and output data are fed into the BP neural network model; through the iteration of the hidden layers of the neural network algorithm model, the state value is output as a vector; by comparing the output vector with the desired output vector, the backpropagation process of error is started, error correction is performed, the connection weights and thresholds of the hidden layers of the neural network are adjusted, and the "sequential-reverse" iteration is repeated repeatedly. When the error tends to a minimum value, the learning of the neural network model is completed.

[0044] Figure 5 A flowchart illustrating cloud-based tool wear prediction using a neural network algorithm, provided for embodiments of the present invention. Figure 5 The steps for outputting tool wear results include:

[0045] S401. Assign different neural network models to the data collected by the CNC system using identifiers;

[0046] S402. Normalize the input vector. The input vector includes: cutting amount, tool speed, feed rate, spindle, vibration frequency, and vibration amplitude as a model. After normalization, the input vector is input into the neural network model.

[0047] S403. The normalized input vector is processed through hidden layers (one or more) in the neural network model, i.e.

[0048] y = f(x1, x2, ... x i ,) to obtain the prediction results;

[0049] S404. After processing by the hidden layer algorithm, the output of the neural network model is a real number R between (-1, 1). After defuzzification, the outputs correspond to different states of the input prediction function points, and the tool wear determination results are output.

[0050] The solution to the fuzzy equation is:

[0051]

[0052] The output of this invention is a database established based on the degree of completion, with accuracy as one of the indicators. If the machining accuracy does not meet the requirements, that is, fewer workpieces are processed, and the degree of completion is low. If the accuracy requirements are met, batch processing will be carried out, and the degree of completion will be high. Therefore, the tool wear determination method of the intelligent cloud platform of this invention can meet the requirements of tool wear determination for different accuracy requirements.

[0053] To address the limitations of existing data-driven algorithms in tool wear prediction, this invention primarily improves upon them through data selection and management. Currently, data-driven models are mostly offline, and the wear parameters required for modeling are difficult to measure. This invention provides a tool wear determination method based on a fuzzy neural network in an intelligent cloud platform. It uses the completion of workpiece machining as the standard for evaluating wear, and combines cloud storage and cloud computing technologies to achieve online data storage and model updates, adapting to new working conditions.

[0054] In addition, this invention provides a tool wear determination method for intelligent cloud platforms based on fuzzy neural networks, which provides CNC operators with accurate and reliable tool wear determination without the need for high-precision sensors, thereby maintaining efficient operation and improving efficiency.

[0055] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for determining tool wear in an intelligent cloud platform based on a fuzzy neural network, characterized in that, Includes the following steps: S1: The CNC system collects the tool and related monitoring data of the tool processing, stores it in the CNC system at fixed intervals, and sends the data monitored by the data system to the cloud data storage platform when processing is completed or when a processing alarm is triggered. The tool and related monitoring data for machining with the tool include: workpiece material, tool model, machining accuracy, cutting amount, tool speed, feed rate, spindle, vibration frequency, vibration amplitude, and completion level; wherein, workpiece material, tool model, and machining accuracy serve as data identifiers; cutting amount, tool speed, feed rate, spindle, vibration frequency, and vibration amplitude serve as input vectors for the model; and completion level serves as the output vector. S2: Plan cloud storage space to receive data monitored by the CNC system and build a cloud database according to rules; Each workpiece material, tool type, and machining accuracy is used as an identifier to establish a separate storage space. Other collected data are stored in the same storage space. If the input variables are the same, the data is merged and the completion level is accumulated. The accumulation formula is: sum = sum + k, where k is the accumulation coefficient. The value of k is different when the completion level is different. S3: A fuzzy neural network model is built using a cloud database, and this fuzzy neural network model is automatically updated at fixed intervals. S4: Input the monitoring data into the neural network model that has completed self-learning and self-adaptation, and output the corresponding tool wear results through the hidden layer and defuzzification module, which are then displayed on the CNC system; The output of tool wear results specifically includes the following steps: S401. Assign different neural network models to the data collected by the CNC system using identifiers; S402. Input the input vector into the neural network model after normalization. S403. After normalization, the input vector is processed by the algorithm through the hidden layer in the neural network model. S404. The neural network model outputs any real number between (-1, 1), then performs defuzzification processing, and outputs the tool wear determination result.

2. The method for determining tool wear in an intelligent cloud platform based on a fuzzy neural network according to claim 1, characterized in that: In step S1, when the machining is completed successfully, the completion level is set to 1; when an alarm related to the tool occurs, the completion level is set to -1; if other situations cause the machining to be interrupted, no data is sent to the cloud platform.

3. The method for determining tool wear in an intelligent cloud platform based on a fuzzy neural network according to claim 1, characterized in that: In step S3, the input vector is first normalized, and then the output vector is fuzzified. The input and output vectors are then input into the neural network model for training and learning.

Images