Numerical control machine tool fault prediction and diagnosis system and method based on deep learning

Abstract

The invention relates to the technical field of numerical control machine tool fault prediction and diagnosis, in particular to a numerical control machine tool fault prediction and diagnosis system and method based on deep learning, and the method comprises the steps: obtaining first target data of a numerical control machine tool, carrying out the processing and analysis of the first target data, and constructing a standard attenuation model; a convergence analysis method is used for optimizing the standard attenuation model, and the output accuracy of the standard attenuation model is improved; and calculating the real-time standard attenuation amount of the numerical control module through the standard attenuation model, inputting the real-time standard attenuation amount into the fault prediction model, and outputting the safety state and maintenance suggestions of the numerical control module. According to the method, data monitoring and fault prediction are carried out on the numerical control machine tool in real time, potential fault hidden dangers are found in advance, maintenance suggestions are provided before faults occur, long-time shutdown caused by sudden faults is avoided, and therefore the production efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of CNC machine tool fault prediction and diagnosis, and in particular to a CNC machine tool fault prediction and diagnosis system and method based on deep learning. Background Art

[0002] A CNC machine tool is an automated device that controls the movement and processing of machine tools through digital information. It uses computer programs and preset instructions to accurately control the relative position and movement of tools and workpieces, thereby achieving high-precision and high-efficiency processing operations.

[0003] Since CNC machine tools require high precision, when some functional components in the machine tools are worn or aged, the precision of CNC machine tool processing may be greatly affected, and the CNC machine tools need to be maintained; traditional regular maintenance may not be able to detect structural component failures of CNC machine tools in time, affecting the overall quality of processing. At the same time, it may also lead to unnecessary maintenance work and increase maintenance costs; therefore, there is an urgent need for a CNC machine tool fault prediction and diagnosis system and method based on deep learning to detect potential fault hazards in advance and make maintenance suggestions before the fault occurs, so as to avoid long downtime caused by sudden failures and improve production efficiency. Summary of the invention

[0004] The purpose of the present invention is to address the problems existing in the background technology and to propose a CNC machine tool fault prediction and diagnosis system and method based on deep learning.

[0005] The technical solution of the present invention is a deep learning-based CNC machine tool fault prediction and diagnosis system and method, comprising the following steps: S1. Acquire first target data of a CNC machine tool, process and analyze the first target data, construct a standard attenuation model, and obtain the influence of wear and aging of relevant structural components of the CNC machine tool on the function of the CNC machine tool through the standard attenuation model; S2. Use the convergence analysis method to optimize the standard attenuation model to improve the accuracy of the standard attenuation model output; S3. Obtain the second target data of the CNC machine tool, process the second target data, determine the first maintenance value and the second maintenance value, calculate the real-time standard attenuation of the CNC module through the standard attenuation model, and input it into the fault prediction model, and output the safety status and maintenance suggestions of the CNC module.

[0006] Preferably, with respect to S1, the method of acquiring first target data of a numerically controlled machine tool, processing and analyzing relevant data, and constructing a standard attenuation model comprises the following steps: S11, obtaining full modular information of the CNC machine tool, determining the CNC module of the CNC machine tool, and the module function, structural component and component sub-function corresponding to the CNC module; S12, determining function attenuation information according to module functions, performing in-depth analysis based on component sub-functions, determining each possible factor that causes the function attenuation information, and combining all possible factors as an attenuation mapping set of the numerical control module; S13, acquiring first operating data of the CNC machine tool, and evaluating the CNC machine tool based on the first operating data to obtain a function degradation tolerance range; S14. Acquire the structural component data of the CNC machine tool, calculate the attenuation influence weight of each possible factor in the attenuation mapping set based on the functional attenuation tolerance range and the structural component data, obtain the attenuation weight set, and establish a standard attenuation model for the CNC machine tool.

[0007] Preferably, for S13, the method for obtaining the first operation data of the CNC machine tool is: Determine the functional dimension of the CNC module based on the functional degradation information; In a target time period, functional parameter change values ​​of the numerical control module under different functional dimensions are obtained, and first operation data of the structural components of the numerical control module are generated based on the functional parameter change values.

[0008] Preferably, for S13, the method of evaluating the CNC machine tool based on the first operating data to obtain the function degradation tolerance range is: Perform data cleaning on the first operation data, remove abnormal data and low-frequency data, and optimize the data structure; performing normalization processing on the first operation data; The functionality of the CNC module is evaluated and the functional degradation characteristic value is obtained by the following formula: ; (Formula 1) In the formula, is the functional attenuation characteristic value of the CNC machine tool; u is the number of the CNC machine tool, and u is a positive integer; is the first function parameter change value; is the quantization coefficient based on the functional dimension; i is the functional dimension number, i is a positive integer; The parameter tolerance threshold R is set, and the function attenuation tolerance range Au is constructed based on the parameter tolerance threshold R and the function attenuation characteristic value AT, so that Au satisfies: ; (Formula 2) Where Au is the functional attenuation tolerance range of the CNC machine tool, u is the number of the CNC machine tool, and u is a positive integer.

[0009] Preferably, for S14, the method for obtaining the structural component data is: The structural components of the CNC module are numbered and marked as j; Count the number of possible factors that may cause the functional degradation of the CNC module due to the structural components and mark them as Q (j); In the target period, the parameter variation values ​​of the structural components in the numerical control module under different possibility factors are obtained and used as the first period characteristic values ​​of the possibility factors.

[0010] Preferably, for S14, the attenuation influence weight of each possibility factor in the attenuation mapping set is calculated to obtain the attenuation weight set by: The decay influence weights of different state dimensions of each structural component are solved by the following set of relations: ; (Formula 3) In the formula, is the first period characteristic value of the possibility factor; j is the structural component number in the numerical control module, j∈[1,n], n is the total number of structural components in the numerical control module; h is the state dimension of the structural component, h is a positive integer, h∈[1,Q(j)]; A1, A2, …, Functional degradation tolerance range for different CNC machine tools; is the attenuation influence weight of the possibility factor.

[0011] Preferably, for S14, the method for establishing a standard attenuation model for a CNC machine tool is: In the target period, obtain the corresponding period feature values ​​of all possible factors of the attenuation mapping set and establish the period feature sequence ; in, is the sequence element of the period characteristic sequence of the xth possibility factor; Establish a decay influence weight sequence based on the decay influence weights of all possible factors in the decay mapping set ; in, is the sequence element of the xth decay influence weight sequence; The standard attenuation model expression is as follows: ; (Formula 4) In the formula, is the standard attenuation, is the sequence element of the time period characteristic sequence, The sequence elements that decay the weight sequence.

[0012] Preferably, with respect to S3, the method for processing the second target data to determine the first maintenance value and the second maintenance value is: Mark the faulty module of the CNC machine tool that has generated the fault as the target module; Mark the previous target period before the target module fails as an abnormal period, obtain the parameter change values ​​of all possible factors in the abnormal period and use them as the characteristic values ​​of the second period, calculate the standard attenuation of the target module in the second period based on the standard attenuation model and mark it as the target attenuation; Determine whether the target attenuation is not greater than the maximum value of the functional attenuation tolerance range; if the target attenuation is not greater than the maximum value of the functional attenuation tolerance range, update the target duration, and based on the updated target duration, update the functional attenuation tolerance range and the standard attenuation model, and use the updated standard attenuation model to update and cover the target attenuation; The target attenuation amount and the functional attenuation tolerance range are traced once. If the target attenuation amount is still not greater than the maximum value of the functional attenuation tolerance range, the target duration is updated twice, and the target attenuation amount and the functional attenuation tolerance range are traced twice until the target duration satisfies the maximum value of the functional attenuation tolerance range that the target attenuation amount is greater than. The maximum value of the functional attenuation tolerance range is marked as the first maintenance value Y1, and the target attenuation amount is marked as the second maintenance value Y2; The real-time standard attenuation of the numerical control module of the numerical control machine tool is obtained and marked as SA, and the real-time standard attenuation SA is input into the fault prediction model.

[0013] Preferably, for S3, the prediction logic of the fault prediction model is as follows: If the real-time standard attenuation SA satisfies: SA≤Y1, the CNC module is judged to be in a safe state and no maintenance is required; If the real-time standard attenuation SA satisfies: Y1≤SA≤Y2, the CNC module is judged to be in a warning state and a maintenance suggestion is output; If the real-time standard attenuation SA satisfies: Y2≤SA, it is determined that the CNC module is in a dangerous state and requires timely maintenance.

[0014] The present invention also discloses a CNC machine tool fault prediction and diagnosis system based on deep learning, which applies the above-mentioned CNC machine tool fault prediction and diagnosis method based on deep learning, and specifically includes: A standard attenuation model building module is used to obtain first target data of the CNC machine tool, process and analyze the first target data, build a standard attenuation model, and obtain the influence of wear and aging of relevant structural components of the CNC machine tool on the function of the CNC machine tool through the standard attenuation model; The standard attenuation model optimization module is used to optimize the standard attenuation model using the convergence analysis method to improve the accuracy of the standard attenuation model output; The fault prediction model building module is used to obtain the second target data of the CNC machine tool, process the second target data, determine the first maintenance value and the second maintenance value, calculate the real-time standard attenuation of the CNC module through the standard attenuation model, and input it into the fault prediction model, and output the safety status and maintenance suggestions of the CNC module.

[0015] Compared with the prior art, the above technical solution of the present invention has the following beneficial technical effects: (1) By acquiring the first target data of the CNC machine tool, processing and analyzing the first target data, and constructing a standard attenuation model, the standard attenuation model can effectively demonstrate the influence of wear and aging of the relevant structural components of the CNC machine tool on the function of the CNC machine tool; (2) Use the convergence analysis method to optimize the standard attenuation model, improve the accuracy of the standard attenuation model output, make the standard attenuation of the CNC machine tool output more consistent with the data, and enhance the prediction ability of the fault prediction model; (3) The real-time standard attenuation of the CNC module is calculated through the standard attenuation model and input into the fault prediction model, which outputs the safety status and maintenance suggestions of the CNC module. This facilitates real-time data monitoring and fault prediction of CNC machine tools, detects potential fault hazards in advance, and makes maintenance suggestions before the fault occurs, avoiding long downtime caused by sudden faults, thereby improving production efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A flowchart of the method of Embodiment 1 proposed by the present invention; Figure 2 This is a flow chart of the method S1 in the first embodiment of the present invention. DETAILED DESCRIPTION

[0017] Embodiment 1, as Figure 1 As shown, the present invention proposes a method for fault prediction and diagnosis of CNC machine tools based on deep learning, comprising the following steps: S1. Acquire first target data of a CNC machine tool, process and analyze the first target data, construct a standard attenuation model, and obtain the influence of wear and aging of relevant structural components of the CNC machine tool on the function of the CNC machine tool through the standard attenuation model; For S1, the first target data of the CNC machine tool is obtained, and the relevant data is processed and analyzed. The method for constructing a standard attenuation model includes the following steps: S11, obtaining full modular information of the CNC machine tool, determining the CNC module of the CNC machine tool, and the module function, structural component and component sub-function corresponding to the CNC module; S12, determining function attenuation information according to module functions, performing in-depth analysis based on component sub-functions, determining each possible factor that causes the function attenuation information, and combining all possible factors as an attenuation mapping set of the numerical control module; The method for determining function degradation information based on module function is: Obtain the name of the CNC module, and search on the Internet using the AI ​​big model based on the name to obtain the module function of the CNC module and the results of component sub-function attenuation, and integrate the text to obtain function attenuation information; For example, a vertical processing module of a CNC machine tool is used to produce a small number of small precision parts with a variety of products. Its function degradation information includes reduced processing accuracy, reduced processing efficiency, abnormal processing lines, abnormal processing noise, etc. The method for analyzing each possible factor that leads to the generation of functional degradation information is: Using recurrent neural network technology, natural language processing is performed on the function attenuation information and component sub-functions, and the processing results are listed to obtain each possible factor; it should be noted that recurrent neural network is an existing technology and will not be described in detail here; Exemplarily, the vertical processing module includes a spindle structure and a tool structure, and the function of the spindle structure is to ensure the coordination and connection stability with the bearing, tool and other components; The possible factors that may cause the CNC machine tool function degradation information to be generated by the spindle system include insufficient spindle accuracy, spindle bearing damage, and poor sealing; The function of the tool system is to output accurate machining trajectory based on the machining program; Then the possible factors that may cause the CNC machine tool function degradation information to be generated by the tool system include tool wear, improper tool installation, etc.; S13, acquiring first operating data of the CNC machine tool, and evaluating the CNC machine tool based on the first operating data to obtain a function degradation tolerance range; For S13, the method for obtaining the first operation data of the CNC machine tool is: Determine the functional dimension of the CNC module based on the functional degradation information; Exemplarily, for the functional attenuation information of the vertical processing module, the functional dimension of the vertical processing module is 4; In a target period, obtaining function parameter change values ​​of the numerical control module under different function dimensions, and generating first operation data of the structural components of the numerical control module based on the function parameter change values; Exemplarily, at the beginning of the target period, the first function parameters under the four function dimensions are respectively obtained, and at the end of the target period, the second function parameters under the four function dimensions are respectively obtained, and the difference between the first function parameter and the second function parameter is calculated to obtain the function parameter change value under different function dimensions; It should be noted that the target period is a period based on the running time of the CNC machine tool, that is, the data for constructing the standard attenuation model is collected based on the CNC machine tools with the same running time and running normally, that is, the influence of the running time of the CNC machine tool on the accuracy of the standard attenuation model is eliminated; For example, if the target period is 100 hours, the same batch of CNC machine tools that have been running for more than 100 hours and have been running for the same amount of time and are running normally are selected as collection objects, and the first target data is collected for this batch of CNC machine tools; For S13, the method for evaluating the CNC machine tool based on the first operating data to obtain the function degradation tolerance range is: Perform data cleaning on the first operation data, remove abnormal data and low-frequency data, and optimize the data structure; performing normalization processing on the first operation data; The functionality of the CNC module is evaluated and the functional degradation characteristic value is obtained by the following formula: ; (Formula 1) In the formula, is the functional attenuation characteristic value of the CNC machine tool; u is the number of the CNC machine tool, and u is a positive integer; is the first function parameter change value; is the quantization coefficient based on the functional dimension; i is the functional dimension number, i is a positive integer; The parameter tolerance threshold R is set, and the function attenuation tolerance range Au is constructed based on the parameter tolerance threshold R and the function attenuation characteristic value AT, so that Au satisfies: ; (Formula 2) Where Au is the functional attenuation tolerance range of the CNC machine tool, u is the number of the CNC machine tool, and u is a positive integer; S14, obtaining structural component data of the CNC machine tool, calculating the attenuation influence weight of each possible factor in the attenuation mapping set based on the functional attenuation tolerance range and the structural component data, obtaining an attenuation weight set, and establishing a standard attenuation model for the CNC machine tool; For S14, the method to obtain structural component data is: The structural components of the CNC module are numbered and marked as j; Count the number of possible factors that may cause the functional degradation of the CNC module due to the structural components and mark them as Q (j); In the target period, the parameter change values ​​of the structural components in the numerical control module under different possibility factors are obtained and used as the first period characteristic values ​​of the possibility factors; For S14, the attenuation influence weight of each possibility factor in the attenuation mapping set is calculated, and the method for obtaining the attenuation weight set is as follows: The decay influence weights of different state dimensions of each structural component are solved by the following set of relations: ; (Formula 3) In the formula, is the first period characteristic value of the possibility factor; j is the structural component number in the numerical control module, j∈[1,n], n is the total number of structural components in the numerical control module; h is the state dimension of the structural component, h is a positive integer, h∈[1,Q(j)]; A1, A2, …, Functional degradation tolerance range for different CNC machine tools; is the attenuation impact weight of the possibility factor; For S14, the method to establish the standard attenuation model of CNC machine tools is: In the target period, obtain the corresponding period feature values ​​of all possible factors of the attenuation mapping set and establish the period feature sequence ; in, is the sequence element of the period characteristic sequence of the xth possibility factor; Establish a decay influence weight sequence based on the decay influence weights of all possible factors in the decay mapping set ; in, is the sequence element of the xth decay influence weight sequence; The standard attenuation model expression is as follows: ; (Formula 4) In the formula, is the standard attenuation, is the sequence element of the time period characteristic sequence, is the sequence element that attenuates the weight sequence; S2. Use the convergence analysis method to optimize the standard attenuation model to improve the accuracy of the standard attenuation model output; For S2, the method of optimizing the standard attenuation model using the convergence analysis method is: Set the reduction radius r of the parameter tolerance threshold R, based on the formula The parameter tolerance threshold R is reset, and the function attenuation tolerance range Au is updated; wherein m is the reduction ratio, and m is a positive integer; Determine the attenuation influence weight of the possibility factor Satisfactory relationship with the updated functional degradation tolerance range Au; If the attenuation of the probability factor affects the weight If the updated function attenuation tolerance range Au is not met, the satisfaction relationship is yes, and the attenuation influence weight is Adjust the weights; if the attenuation of the possibility factor affects the weights If the updated functional degradation tolerance range Au is met, the satisfiability relationship is no; According to the satisfaction relationship being yes, the reduction ratio m is adjusted step by step until the preset parameter tolerance threshold R is equal to the standard parameter tolerance threshold R0, and the convergent solution set of the attenuation influence weights of different possibility factors is obtained; it should be noted that the accuracy of the standard parameter tolerance threshold R0 should be the same as the minimum accuracy of the parameters required by the CNC machine tool; S3, obtaining the second target data of the CNC machine tool, processing the second target data, determining the first maintenance value and the second maintenance value, calculating the real-time standard attenuation of the CNC module through the standard attenuation model, and inputting the calculated value into the fault prediction model, and outputting the safety status and maintenance suggestions of the CNC module; With respect to S3, the second target data is processed to determine the first maintenance value and the second maintenance value by: Mark the faulty module of the CNC machine tool that has generated the fault as the target module; Mark the previous target period before the target module fails as an abnormal period, obtain the parameter change values ​​of all possible factors in the abnormal period and use them as the characteristic values ​​of the second period, calculate the standard attenuation of the target module in the second period based on the standard attenuation model and mark it as the target attenuation; As explained above for the exemplary target duration, similarly, a CNC machine tool whose running time is the same as that of the CNC machine tool as the collection object of the first running data and has a fault is selected as the collection object of the second target data, so as to eliminate the influence of the running time of the CNC machine tool on the fault prediction model; Determine whether the target attenuation is not greater than the maximum value of the functional attenuation tolerance range; if the target attenuation is not greater than the maximum value of the functional attenuation tolerance range, update the target duration, and based on the updated target duration, update the functional attenuation tolerance range and the standard attenuation model, and use the updated standard attenuation model to update and cover the target attenuation; The target attenuation amount and the functional attenuation tolerance range are traced once. If the target attenuation amount is still not greater than the maximum value of the functional attenuation tolerance range, the target duration is updated twice, and the target attenuation amount and the functional attenuation tolerance range are traced twice until the target duration satisfies the maximum value of the functional attenuation tolerance range that the target attenuation amount is greater than. The maximum value of the functional attenuation tolerance range is marked as the first maintenance value Y1, and the target attenuation amount is marked as the second maintenance value Y2; The real-time standard attenuation of the numerical control module of the numerical control machine tool is obtained and marked as SA, and the real-time standard attenuation SA is input into the fault prediction model; For S3, the prediction logic of the fault prediction model is as follows: If the real-time standard attenuation SA satisfies: SA≤Y1, the CNC module is judged to be in a safe state and no maintenance is required; If the real-time standard attenuation SA satisfies: Y1≤SA≤Y2, the CNC module is judged to be in a warning state and a maintenance suggestion is output; If the real-time standard attenuation SA satisfies: Y2≤SA, it is determined that the CNC module is in a dangerous state and requires timely maintenance.

[0018] Embodiment 2, a CNC machine tool fault prediction and diagnosis system based on deep learning proposed by the present invention, which is applied to a CNC machine tool fault prediction and diagnosis method based on deep learning proposed in Embodiment 1, specifically includes: A standard attenuation model building module is used to obtain first target data of the CNC machine tool, process and analyze the first target data, build a standard attenuation model, and obtain the influence of wear and aging of relevant structural components of the CNC machine tool on the function of the CNC machine tool through the standard attenuation model; The standard attenuation model optimization module is used to optimize the standard attenuation model using the convergence analysis method to improve the accuracy of the standard attenuation model output; The fault prediction model building module is used to obtain the second target data of the CNC machine tool, process the second target data, determine the first maintenance value and the second maintenance value, calculate the real-time standard attenuation of the CNC module through the standard attenuation model, and input it into the fault prediction model, and output the safety status and maintenance suggestions of the CNC module.

[0019] The embodiments of the present invention are described in detail above with reference to the accompanying drawings, but the present invention is not limited thereto, and various changes can be made within the knowledge scope of technicians in the relevant technical field without departing from the purpose of the present invention.

Claims

1. A method for fault prediction and diagnosis of CNC machine tools based on deep learning, characterized in that: The following steps are involved: S1. Acquire first target data of a CNC machine tool, process and analyze the first target data, construct a standard attenuation model, and obtain the influence of wear and aging of relevant structural components of the CNC machine tool on the function of the CNC machine tool through the standard attenuation model; S2. Use the convergence analysis method to optimize the standard attenuation model to improve the accuracy of the standard attenuation model output; S3. Obtain the second target data of the CNC machine tool, process the second target data, determine the first maintenance value and the second maintenance value, calculate the real-time standard attenuation of the CNC module through the standard attenuation model, and input it into the fault prediction model, and output the safety status and maintenance suggestions of the CNC module.

2. The method for fault prediction and diagnosis of CNC machine tools based on deep learning according to claim 1, characterized in that: For S1, the first target data of the CNC machine tool is obtained, and the relevant data is processed and analyzed. The method for constructing a standard attenuation model includes the following steps: S11, obtaining full modular information of the CNC machine tool, determining the CNC module of the CNC machine tool, and the module function, structural component and component sub-function corresponding to the CNC module; S12, determining function attenuation information according to module functions, performing in-depth analysis based on component sub-functions, determining each possible factor that causes the function attenuation information, and combining all possible factors as an attenuation mapping set of the numerical control module; S13, acquiring first operating data of the CNC machine tool, and evaluating the CNC machine tool based on the first operating data to obtain a function degradation tolerance range; S14. Acquire the structural component data of the CNC machine tool, calculate the attenuation influence weight of each possible factor in the attenuation mapping set based on the functional attenuation tolerance range and the structural component data, obtain the attenuation weight set, and establish a standard attenuation model for the CNC machine tool.

3. A method for fault prediction and diagnosis of CNC machine tools based on deep learning according to claim 2, characterized in that: For S13, the method for obtaining the first operation data of the CNC machine tool is: Determine the functional dimension of the CNC module based on the functional degradation information; In a target time period, functional parameter change values ​​of the numerical control module under different functional dimensions are obtained, and first operation data of the structural components of the numerical control module are generated based on the functional parameter change values.

4. The method for fault prediction and diagnosis of CNC machine tools based on deep learning according to claim 3, characterized in that: For S13, the method for evaluating the CNC machine tool based on the first operating data to obtain the function degradation tolerance range is: Perform data cleaning on the first operation data, remove abnormal data and low-frequency data, and optimize the data structure; performing normalization processing on the first operation data; The functionality of the CNC module is evaluated and the functional degradation characteristic value is obtained by the following formula: ； (Formula 1) In the formula, is the functional attenuation characteristic value of the CNC machine tool; u is the number of the CNC machine tool, and u is a positive integer; is the first function parameter change value; is the quantitative coefficient based on the functional dimension; i is the function dimension number, i is a positive integer; Set the parameter tolerance threshold R, and based on the parameter tolerance threshold R and the function attenuation characteristic value Construct the functional attenuation tolerance range Au so that Au satisfies: ; (Formula 2) Where Au is the functional attenuation tolerance range of the CNC machine tool, u is the number of the CNC machine tool, and u is a positive integer.

5. The method for fault prediction and diagnosis of CNC machine tools based on deep learning according to claim 2, characterized in that: For S14, the method to obtain structural component data is: The structural components of the CNC module are numbered and marked as j; Count the number of possible factors that may cause the functional degradation of the CNC module due to the structural components and mark them as Q (j); In the target period, the parameter variation values ​​of the structural components in the numerical control module under different possibility factors are obtained and used as the first period characteristic values ​​of the possibility factors.

6. A method for fault prediction and diagnosis of CNC machine tools based on deep learning according to claim 5, characterized in that: For S14, the attenuation influence weight of each possibility factor in the attenuation mapping set is calculated, and the method for obtaining the attenuation weight set is as follows: The decay influence weights of different state dimensions of each structural component are solved by the following set of relations: ； (Formula 3) In the formula, is the first period characteristic value of the possibility factor; j is the structural component number in the numerical control module, j∈[1,n], n is the total number of structural components in the numerical control module; h is the state dimension of the structural component, h is a positive integer, h∈[1,Q(j)]; A1, A2, ..., Functional degradation tolerance range for different CNC machine tools; is the attenuation influence weight of the possibility factor.

7. The method for fault prediction and diagnosis of CNC machine tools based on deep learning according to claim 6, characterized in that: For S14, the method to establish the standard attenuation model of CNC machine tools is: In the target period, obtain the corresponding period feature values ​​of all possible factors of the attenuation mapping set and establish the period feature sequence ; in, is the sequence element of the period characteristic sequence of the xth possibility factor; Establish a decay influence weight sequence based on the decay influence weights of all possible factors in the decay mapping set ; in, is the sequence element of the xth decay influence weight sequence; The standard attenuation model expression is as follows: ; (Formula 4) In the formula, is the standard attenuation, is the sequence element of the time period characteristic sequence, The sequence elements that decay the weight sequence.

8. The method for fault prediction and diagnosis of CNC machine tools based on deep learning according to claim 7, characterized in that: With respect to S3, the second target data is processed to determine the first maintenance value and the second maintenance value by: Mark the faulty module of the CNC machine tool that has generated the fault as the target module; Mark the previous target period before the target module fails as an abnormal period, obtain the parameter change values ​​of all possible factors in the abnormal period and use them as the characteristic values ​​of the second period, calculate the standard attenuation of the target module in the second period based on the standard attenuation model and mark it as the target attenuation; Determine whether the target attenuation is not greater than the maximum value of the functional attenuation tolerance range; if the target attenuation is not greater than the maximum value of the functional attenuation tolerance range, update the target duration, and based on the updated target duration, update the functional attenuation tolerance range and the standard attenuation model, and use the updated standard attenuation model to update and cover the target attenuation; The target attenuation amount and the functional attenuation tolerance range are traced once. If the target attenuation amount is still not greater than the maximum value of the functional attenuation tolerance range, the target duration is updated twice, and the target attenuation amount and the functional attenuation tolerance range are traced twice until the target duration satisfies the maximum value of the functional attenuation tolerance range that the target attenuation amount is greater than. The maximum value of the functional attenuation tolerance range is marked as the first maintenance value Y1, and the target attenuation amount is marked as the second maintenance value Y2; The real-time standard attenuation of the numerical control module of the numerical control machine tool is obtained and marked as SA, and the real-time standard attenuation SA is input into the fault prediction model.

9. The method for fault prediction and diagnosis of CNC machine tools based on deep learning according to claim 8, characterized in that: For S3, the prediction logic of the fault prediction model is as follows: If the real-time standard attenuation SA satisfies: SA≤Y1, the CNC module is judged to be in a safe state and no maintenance is required; If the real-time standard attenuation SA satisfies: Y1≤SA≤Y2, the CNC module is judged to be in a warning state and a maintenance suggestion is output; If the real-time standard attenuation SA satisfies: Y2≤SA, it is determined that the CNC module is in a dangerous state and requires timely maintenance.

10. A CNC machine tool fault prediction and diagnosis system based on deep learning, applying a CNC machine tool fault prediction and diagnosis method based on deep learning as described in any one of claims 1 to 9 above, characterized in that: Specifically include: A standard attenuation model building module is used to obtain first target data of the CNC machine tool, process and analyze the first target data, build a standard attenuation model, and obtain the influence of wear and aging of relevant structural components of the CNC machine tool on the function of the CNC machine tool through the standard attenuation model; The standard attenuation model optimization module is used to optimize the standard attenuation model using the convergence analysis method to improve the accuracy of the standard attenuation model output; The fault prediction model building module is used to obtain the second target data of the CNC machine tool, process the second target data, determine the first maintenance value and the second maintenance value, calculate the real-time standard attenuation of the CNC module through the standard attenuation model, and input it into the fault prediction model, and output the safety status and maintenance suggestions of the CNC module.

Images