Xenomai-based real-time Linux controller and method thereof

Abstract

The present invention relates to the technical field of Linux controllers, and specifically discloses a real-time Linux controller based on Xenomai and a method thereof. On the A40i-H dual-bus controller platform, the Xenomai real-time patch is first transplanted to the mainline Linux kernel and configured, compiled and tested, then the libevl library is compiled in a real-time environment, and the jitter of the evl core thread switching is tested to verify the integrity and correctness of the real-time patch transplantation, and then the real-time driver is adapted and connected to the driver, and finally an application for controlling the jitter of a processing machine tool is developed and tested to perform actual machine tool jitter control. In this way, by transplanting the Xenomai real-time patch and optimizing the configuration, the jitter of the system under no-load conditions can be reduced, thereby meeting the needs of precision machining and high-precision control applications. At the same time, it has a higher real-time response speed when processing high-frequency tasks, and improves the real-time performance of the control.

Description

Technical Field

[0001] The present invention relates to the technical field of Linux controllers, and more specifically, to a Xenomai-based real-time Linux controller and a method thereof. Background Art

[0002] The A40i-H dual-bus controller platform plays a key role in the field of industrial automation, and its performance directly affects the efficiency of the production line and the quality of the product. Although the original PREEMPT RT real-time patch solution used by the A40i-H dual-bus controller platform can provide certain real-time performance, its jitter performance under no-load state is poor and fails to meet strict industrial requirements.

[0003] Specifically, under no-load conditions, the jitter measured using the PREEMPT RT patch averaged 13 microseconds (μs), with a maximum jitter of 25 μs, which is insufficient in precision machining and high-precision control applications, as larger jitter can lead to reduced machining quality and production efficiency. In addition, the PREEMPT RT patch improves the real-time response speed of the system through a preemptive scheduling mechanism, but its response speed is still limited when processing high-frequency tasks and cannot meet the requirements of some high-precision real-time control.

[0004] Therefore, an optimized controller is desired. Summary of the invention

[0005] The present invention provides a real-time Linux controller based on Xenomai and a method thereof, which can reduce the jitter of the system under no-load conditions by transplanting the Xenomai real-time patch and optimizing the configuration, thereby meeting the requirements of precision machining and high-precision control applications. At the same time, it has a higher real-time response speed when processing high-frequency tasks, and improves the real-time performance of the control.

[0006] In a first aspect, a real-time Linux control method based on Xenomai is provided, comprising: transplanting a Xenomai real-time patch in the mainline Linux kernel on an A40i-H dual-bus controller platform, adapting relevant configuration options, and compiling, packaging and testing the transplanted mainline Linux kernel; compiling a libevl library in a Linux real-time environment, and compiling and packaging to test the jitter of evl core thread switching, so as to verify the integrity and correctness of the transplantation of the Xenomai real-time patch; adapting the real-time driver and connecting to the driver; writing a machining machine tool jitter control application, and performing functional testing; and using the machining machine tool jitter control application to perform jitter control.

[0007] In a second aspect, a Xenomai-based real-time Linux controller is also provided, wherein the Xenomai-based real-time Linux controller is used to execute the aforementioned Xenomai-based real-time Linux control method.

[0008] Compared with the prior art, the present invention has the following beneficial effects: on the A40i-H dual-bus controller platform, firstly, the Xenomai real-time patch is transplanted to the mainline Linux kernel and configured, compiled and tested, then the libevl library is compiled in a real-time environment, and the jitter of the evl core thread switching is tested to verify the integrity and correctness of the real-time patch, then the real-time driver is adapted and connected to the driver, and finally an application for controlling the jitter of the processing machine tool is developed and tested to perform actual machine tool jitter control. In this way, by transplanting the Xenomai real-time patch and optimizing the configuration, the jitter of the system under no-load conditions can be reduced, thereby meeting the needs of precision machining and high-precision control applications. At the same time, it has a higher real-time response speed when processing high-frequency tasks, and improves the real-time performance of the control. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] The above and other purposes, features and advantages of the present invention will become more apparent by describing the embodiments of the present invention in more detail in conjunction with the accompanying drawings. The accompanying drawings are used to provide a further understanding of the embodiments of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0010] Figure 1 The figure is a schematic flow chart of a real-time Linux control method based on Xenomai according to an embodiment of the present invention.

[0011] Figure 2 It is a schematic flow chart of using the machining machine tool jitter control application to perform jitter control in the Xenomai-based real-time Linux control method according to an embodiment of the present invention.

[0012] Figure 3 The figure is a data flow diagram of performing jitter control by using the machining machine jitter control application in the real-time Linux control method based on Xenomai according to an embodiment of the present invention.

[0013] Figure 4 The present invention is a schematic flow chart of a real-time Linux control method based on Xenomai in an embodiment of the present invention, in which the sequence of the motor temperature timing feature vectors is passed through a node feature integrator guided by significance attenuation to obtain a motor temperature timing node information aggregation representation vector.

[0014] Figure 5In the real-time Linux control method based on Xenomai in an embodiment of the present invention, the sequence of the motor temperature timing node information aggregation representation vector and the processing requirement word granularity semantic encoding feature vector is dynamically fused through a cross-domain feature attention gate guided by a key matrix to obtain a schematic flow chart of the motor temperature-processing requirement semantic fusion feature vector. DETAILED DESCRIPTION

[0015] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work also fall within the scope of protection of the present invention.

[0016] Based on this, the present application proposes a real-time Linux control method based on Xenomai, which, on the A40i-H dual-bus controller platform, first transplants the Xenomai real-time patch to the mainline Linux kernel and configures, compiles and tests it, then compiles the libevl library in a real-time environment, and tests the jitter of the evl core thread switching to verify the integrity and correctness of the real-time patch, then adapts the real-time driver and connects to the driver, and finally develops and tests an application for controlling the jitter of the processing machine tool to perform actual machine tool jitter control. In this way, by transplanting the Xenomai real-time patch and optimizing the configuration, the jitter of the system under no-load conditions can be reduced, thereby meeting the needs of precision machining and high-precision control applications. At the same time, it has a higher real-time response speed when processing high-frequency tasks, which improves the real-time performance of the control.

[0017] Figure 1 FIG. 1 is a schematic flow chart of a real-time Linux control method based on Xenomai according to an embodiment of the present invention. Figure 1 As shown, the real-time Linux control method based on Xenomai includes: S1, transplanting the Xenomai real-time patch in the mainline Linux kernel on the A40i-H dual-bus controller platform, and adapting the relevant configuration options, and compiling, packaging and testing the transplanted mainline Linux kernel; S2, compiling the libevl library in the Linux real-time environment, and compiling and packaging to test the jitter of the evl core thread switching, so as to verify the integrity and correctness of the Xenomai real-time patch transplantation; S3, adapting the real-time drive and connecting the drive; S4, writing a machining machine tool jitter control application, and performing a functional test; S5, using the machining machine tool jitter control application to perform jitter control.

[0018] Specifically, S1 transplants the Xenomai real-time patch in the mainline Linux kernel on the A40i-H dual-bus controller platform, adapts relevant configuration options, and compiles, packages and tests the transplanted mainline Linux kernel, including: ensuring that a cross-compilation tool chain has been installed in the development environment, and the cross-compilation tool chain is suitable for the architecture of the A40i-H dual-bus controller platform; downloading the mainline Linux kernel and the Xenomai real-time patch; determining the compatibility between the Xenomai real-time patch and the mainline Linux kernel; using the patch tool to apply the Xenomai real-time patch to the mainline Linux kernel; using make menuconfig or make xconfig configures kernel options and enables real-time options required by the Xenomai real-time patch; adjusts relevant configurations according to the hardware characteristics of the A40i-H dual-bus controller platform, and the relevant configurations include interrupt processing and scheduling strategies; executes the make command to compile the kernel and ensure that there are no compilation errors; generates a kernel image and module package suitable for the A40i-H dual-bus controller platform; deploys the kernel image and module package to the A40i-H dual-bus controller platform; starts the system to ensure that the mainline Linux kernel can run normally and load the Xenomai real-time patch; verifies the basic functions of the system, and the basic functions include device driver loading and network connection.

[0019] Specifically, S2 compiles the libevl library in the Linux real-time environment, and compiles and packages the test evl core thread switching jitter to verify the integrity and correctness of the Xenomai real-time patch transplantation, including: downloading the latest source code of libevl; setting up a cross-compilation environment to ensure that the compilation is for the A40i-H dual-bus controller platform; executing the compilation command to generate the libevl library file; using the libevl library file to write a test program to measure the switching jitter of the core thread; running the test program on the A40i-H dual-bus controller platform to record the jitter data; comparing whether the jitter data meets the real-time performance requirements.

[0020] It should be understood that the jitter control of machine tools is closely related to parameter control, and the spindle speed control is an important means to reduce jitter and improve machining accuracy. By accurately controlling the spindle speed, the machine tool jitter caused by speed fluctuations can be reduced, thereby improving the accuracy and performance of the product, reducing the scrap rate, and thus reducing production costs.

[0021] However, traditional control systems often fail to provide sufficient real-time response capabilities, especially when dealing with complex machining tasks, which may cause machining errors due to response delays. Although PLC can handle basic logic control, its computing power and real-time performance are insufficient when dealing with continuous control tasks that require high precision and fast response. This may result in the system being unable to adjust the spindle speed in a dynamically changing working environment, causing greater jitter and affecting machining accuracy and surface quality.

[0022] Accordingly, when using a processing machine tool jitter control application to perform jitter control, the technical concept of the present application is to transmit the processing requirements input by the user extracted by the human-computer interaction system and the time series of the motor temperature values ​​of the processing machine tool collected by the temperature sensor deployed on the processing machine tool to the real-time enhanced industrial computer, and use artificial intelligence-based data analysis and text processing technology in the real-time enhanced industrial computer to perform word segmentation and semantic encoding on the processing requirements, and extract and aggregate the time series features of the motor temperature values, so as to automatically recommend the real-time spindle speed of the processing machine tool according to the dynamic fusion features between the granularity semantic encoding features of each processing requirement word and the motor temperature aggregation features. In this way, by using a real-time enhanced industrial computer based on artificial intelligence technology for data processing and jitter control, it is possible to provide an instant response, and has a higher real-time control capability than the traditional PLC control system. At the same time, the spindle speed can be automatically recommended according to the dynamic fusion features of the processing requirements and the motor temperature, thereby improving the degree of automation of the processing.

[0023] Figure 2 It is a schematic flow chart of using the machining machine tool jitter control application to perform jitter control in the Xenomai-based real-time Linux control method according to an embodiment of the present invention. Figure 3 FIG. 1 is a data flow diagram of a real-time Linux control method based on Xenomai in an embodiment of the present invention, in which the jitter control application of a machining machine tool is used to perform jitter control. Figure 2 and Figure 3As shown, the jitter control application for the processing machine tool is used to perform jitter control, including: S51, extracting the processing requirements input by the user from the human-computer interaction system; S52, obtaining the time series of the motor temperature value of the processing machine tool collected by the temperature sensor deployed on the processing machine tool; S53, performing word segmentation processing on the processing requirements and then passing through a semantic encoder including a word embedding layer to obtain a sequence of semantic encoding feature vectors of the processing requirement word granularity; S54, performing data regularization on the time series of the motor temperature value of the processing machine tool according to the time dimension and then passing through a motor temperature timing encoder based on 1D-CNN to obtain the motor temperature time series. S55, passing the sequence of motor temperature timing feature vectors through a node feature integrator guided by significance attenuation to obtain a motor temperature timing node information aggregation representation vector; S56, passing the motor temperature timing node information aggregation representation vector and the sequence of processing requirement word granularity semantic encoding feature vectors through a cross-domain feature attention gate dynamic fusion network guided by a key matrix to obtain a motor temperature-processing requirement semantic fusion feature vector; S57, based on the motor temperature-processing requirement semantic fusion feature vector, obtaining a control result, and the control result is used to represent the real-time spindle speed recommendation value of the processing machine tool.

[0024] In the above-mentioned real-time Linux control method based on Xenomai, the S51 extracts the processing requirements input by the user from the human-computer interaction system. It should be understood that the processing requirements directly reflect the user's intentions and the specific work content to be completed. The processing requirements include specific process parameters, material types, processing dimensions and other information, which are important bases for determining how to control the machine tool during the processing process. By analyzing the processing requirements, the actual needs of the user can be further understood and parsed, so that the control system can respond to the user's instructions more intelligently and accurately.

[0025] In the above-mentioned real-time Linux control method based on Xenomai, the S52 obtains the time series of the motor temperature value of the processing machine tool collected by the temperature sensor deployed on the processing machine tool. It should be understood that considering that the motor temperature is an important indicator reflecting the operating state of the processing machine tool, it is related to the working efficiency, reliability and life of the motor. If the motor temperature is too high, it may cause the motor performance to decline or even be damaged. By considering the important factor of motor temperature, the spindle speed in the processing process is adjusted accordingly. Doing so can not only avoid performance degradation or failure caused by motor overheating, but also ensure that the processing task is carried out under optimal conditions, thereby improving the degree of automation of processing.

[0026] In the above-mentioned Xenomai-based real-time Linux control method, the S53, after word segmentation of the processing requirements, passes through a semantic encoder including a word embedding layer to obtain a sequence of processing requirement word granularity semantic coding feature vectors. It should be understood that, considering that the processing requirements contain the user's intentions and demand information, these information are composed of multiple words or phrases. Therefore, in order to understand the semantic information in each word more carefully and accurately, in the technical solution of the present application, the processing requirements are subject to word segmentation. Furthermore, in order to determine the specific semantic meaning and importance of each word in the processing requirements, the processing requirements after word segmentation are further passed through a semantic encoder including a word embedding layer to capture and extract the contextual semantic information and meaning of each word, and obtain a sequence of processing requirement word granularity semantic coding feature vectors.

[0027] Optionally, in one embodiment of the present application, the processing requirements are segmented and then processed by a semantic encoder including a word embedding layer to obtain a sequence of processing requirements word granularity semantic encoding feature vectors, including: first, the processing requirements text input by the user is segmented, that is, the long text is divided into independent words or phrases. This process is to understand the semantic information in each word more carefully and accurately, because the processing requirements are composed of multiple words or phrases, and each word or phrase carries specific semantics. Next, the processing requirements after word segmentation are processed by a semantic encoder including a word embedding layer. The word embedding layer maps each word to a vector in a high-dimensional space, which not only represents the meaning of the word itself, but also takes into account their use in the context. After obtaining the word vectors, the semantic encoder will further process these vectors to capture the specific semantic meaning and importance of each word in the processing requirements. The semantic encoder may include models such as a recurrent neural network (RNN), a long short-term memory network (LSTM), a gated recurrent unit (GRU) or a Transformer architecture. Finally, through the above steps, the semantic encoder converts the processing requirements into a sequence of processing requirements word granularity semantic encoding feature vectors.

[0028] Optionally, in another embodiment of the present application, the processing requirements are contextually semantically encoded to obtain a sequence of processing requirement word granularity semantic encoding feature vectors, including: first, cleaning the text data, removing irrelevant characters, such as punctuation marks, special characters, etc., and segmenting the processing requirement description to decompose the sentence into separate words or tags. Then, use a pre-trained word embedding model (such as Word2Vec, GloVe, etc.), or train a custom word embedding model based on a corpus in a specific field. Thereby mapping each word to its corresponding word vector representation. These vectors capture the general meaning of words in different contexts. Apply context-sensitive embedding methods such as BERT (BidirectionalEncoder Representations from Transformers), ELMo (Embeddings from LanguageModels) or other Transformer architectures. These models can generate a context-based dynamic vector representation for each word, so that the same word has different vector representations in different sentences. Finally, for each processing requirement description, a series of word vectors are obtained through the above steps. These word vectors are organized into a sequence in the order in which the words appear, that is, a sequence of processing requirement word granularity semantic encoding feature vectors is obtained.

[0029] In the above-mentioned Xenomai-based real-time Linux control method, the S54, after the time series of the motor temperature value of the processing machine tool is data-regularized according to the time dimension, the motor temperature timing encoder based on 1D-CNN is used to obtain a sequence of motor temperature timing feature vectors. It should be understood that, considering that the time series of the motor temperature value of the processing machine tool has timing characteristics in the time dimension, and the timing characteristics exhibited in different local time segments are different, based on this, in the technical solution of the present application, the time series of the motor temperature value of the processing machine tool is data-regularized according to the time dimension, and the motor temperature timing encoder based on 1D-CNN is used to obtain a sequence of motor temperature timing feature vectors. In this way, the pattern and trend of the motor temperature changing over time can be identified, so as to better understand the operating temperature state of the machine tool.

[0030] Optionally, in one embodiment of the present application, the time series of the motor temperature value of the processing machine tool is data regularized according to the time dimension and then passed through a motor temperature timing encoder based on 1D-CNN to obtain a sequence of motor temperature timing feature vectors, including: first, the collected time series data of the motor temperature value is regularized according to the time dimension to ensure the consistency and correct format of the data. Then, the time series data of the motor temperature value is feature extracted through a 1D convolutional neural network (1D-CNN). Specifically, the convolution layer in the 1D-CNN performs a convolution operation on the input temperature time series data, and uses a set of filters (i.e., convolution kernels) to scan the entire sequence to capture features at different time points. Then, there is usually a pooling operation after the convolution, which can reduce the spatial dimension of the data while retaining the most important feature information. Through the above steps, 1D-CNN converts the original time series data into a sequence of motor temperature timing feature vectors.

[0031] In the above-mentioned real-time Linux control method based on Xenomai, the S55 is to pass the sequence of the motor temperature timing feature vector through a node feature integrator guided by significance attenuation to obtain a motor temperature timing node information aggregation representation vector. It should be understood that, considering that the sequence of the motor temperature timing feature vector is a dynamic change process that changes with time, and the features of different time points have different influences and contributions to the current time point, in order to be able to identify and strengthen those features that are most critical to the final task, while weakening the influence of those less important features, and improving the characterization capability of the motor temperature in the entire time domain, in the technical solution of the present application, the sequence of the motor temperature timing feature vector is passed through a node feature integrator guided by significance attenuation to obtain a motor temperature timing node information aggregation representation vector.

[0032] Figure 4 The present invention is a schematic flow chart of a real-time Linux control method based on Xenomai in which the sequence of the motor temperature time series feature vector is passed through a node feature integrator guided by significance attenuation to obtain a motor temperature time series node information aggregation representation vector. Figure 4As shown, optionally, in one embodiment of the present application, the sequence of the motor temperature timing feature vectors is passed through a node feature integrator guided by significance attenuation to obtain a motor temperature timing node information aggregation representation vector, including: S551, calculating the feature significance description factor of each motor temperature timing feature vector in the sequence of the motor temperature timing feature vectors, wherein the feature significance description factor is related to the mean and variance of each motor temperature timing feature vector; S552, constructing each motor temperature timing feature vector based on the distance span between each motor temperature timing feature vector in the sequence of the motor temperature timing feature vector and the current motor temperature timing feature vector. characteristic significance attenuation factor; S553, modulating the characteristic significance description factor of each motor temperature timing feature vector based on the characteristic significance attenuation factor of each motor temperature timing feature vector to obtain a sequence of motor temperature timing characteristic significance attenuation description factors; S554, inputting the sequence of motor temperature timing characteristic significance attenuation description factors into the gated mask module to obtain a sequence of motor temperature timing characteristic significance attenuation weight factors; S555, using the sequence of motor temperature timing characteristic significance attenuation weight factors as the sequence of weights, calculating the weighted sum of the sequence of motor temperature timing feature vectors to obtain the motor temperature timing node information aggregation representation vector.

[0033] Optionally, in one embodiment of the present application, a feature significance description factor of each motor temperature timing feature vector in the sequence of the motor temperature timing feature vectors is calculated, including: respectively calculating the mean and variance of the motor temperature timing feature vectors to obtain the motor temperature timing feature mean and the motor temperature timing feature variance; calculating the positional difference between the motor temperature timing feature vector and the motor temperature timing feature mean to obtain a motor temperature timing difference vector; calculating the fourth power of each eigenvalue in the motor temperature timing difference vector to obtain a motor temperature timing modulation difference vector; calculating the expected value of the motor temperature timing modulation difference vector to obtain a motor temperature timing expected value; dividing the motor temperature timing expected value by the square of the motor temperature timing feature variance to obtain a feature significance description factor corresponding to the motor temperature timing feature vector.

[0034] Optionally, in one embodiment of the present application, based on the distance span between each motor temperature timing feature vector in the sequence of the motor temperature timing feature vector and the current motor temperature timing feature vector, a characteristic significance attenuation factor of each motor temperature timing feature vector is constructed, including: extracting the maximum value among the each motor temperature timing feature vector to obtain a sequence of motor temperature timing feature maximum values; extracting the maximum value of the current motor temperature timing feature vector to obtain the current motor temperature timing feature maximum value; calculating the positional subtraction between the current motor temperature timing feature maximum value and the sequence of the motor temperature timing feature maximum values ​​to obtain a sequence of motor temperature timing feature offset values; counting the distance span values ​​between each motor temperature timing feature vector and the current motor temperature timing feature vector to obtain a sequence of motor temperature timing distance span values, and calculating the squares of each value in the sequence of the motor temperature timing distance span values ​​to obtain a sequence of motor temperature timing distance span modulation values; dividing the sequence of motor temperature timing feature offset values ​​by the sequence of motor temperature timing distance span modulation values ​​by position to obtain the characteristic significance attenuation factors of each motor temperature timing feature vector.

[0035] Optionally, in one embodiment of the present application, the sequence of the motor temperature timing feature significance attenuation description factors is input into a gated mask module to obtain a sequence of motor temperature timing feature significance attenuation weight factors, including: performing normalization processing based on a Sigmoid function on each motor temperature timing feature significance attenuation description factor in the sequence of the motor temperature timing feature significance attenuation description factors to obtain a set of normalized motor temperature timing feature significance attenuation description factors; and inputting each normalized motor temperature timing feature significance attenuation description factor in the set of normalized motor temperature timing feature significance attenuation description factors into a gated function for masking processing to obtain a sequence of motor temperature timing feature significance attenuation weight factors.

[0036] Specifically, in the embodiment of the present application, the sequence of the motor temperature timing feature vector is processed by a node feature integrator guided by significance attenuation using the following node feature integration formula to obtain the motor temperature timing node information aggregation representation vector; wherein the node feature integration formula is: ;in, represents the sequence of the motor temperature time series feature vectors, , and are the first , and The motor temperature time series characteristic vector, To extract the maximum value of a vector, express and The distance span between For the said The characteristic significance attenuation factor of the motor temperature time series characteristic vector, It is The first of the motor temperature time series characteristic vectors The eigenvalues ​​at the positions, and They are respectively The mean and square of the variance of the motor temperature time series feature vector, It is The motor temperature timing feature vector corresponding to the motor temperature timing feature vector is the motor temperature timing feature significance attenuation description factor, yes function, is the first in the set of descriptive factors of the significant attenuation of the normalized motor temperature timing characteristics. Normalized motor temperature timing characteristic significant attenuation description factor, For masking, is the first in the sequence of significant attenuation weight factors of the motor temperature timing characteristics. The motor temperature timing characteristic significance attenuation weight factor, is the predetermined threshold, is the aggregate representation vector of the motor temperature timing node information.

[0037] Specifically, the node feature integrator guided by the significance decay performs significance calculation analysis on each node separately, and improves the overall motor temperature aggregation feature expression based on the correlation calculation with the current time point. In detail, the feature significance description factor of each feature is first calculated based on the mean and variance of each motor temperature time series feature vector. That is, the mean and variance are two basic indicators in statistics that describe the center position and discreteness of the data set, which can quantify the stability and range of each feature in the time series, thereby evaluating its importance to the overall analysis. Then, by calculating the distance span between each motor temperature time series feature vector and the motor temperature time series feature vector at the current moment, a feature significance decay factor is constructed to reflect that the influence of historical features that differ greatly from the current state gradually weakens over time. This method emphasizes that the features of the recent period are more critical and gives them greater weight, thereby highlighting the importance of the latest data in analysis and decision-making. Subsequently, the feature significance attenuation factor of each feature vector is applied to its feature significance description factor to modulate it to enhance the weight of the feature vector of recent changes, because these changes may be more critical to the current decision or prediction, while reducing the impact of noise or irrelevant information on the current analysis to obtain a set of motor temperature timing feature significance attenuation description factors corresponding to the motor temperature timing feature vector. Finally, the obtained description factor is input into the gated mask module to convert it into a weight factor to perform weighted summation on the sequence of the motor temperature timing feature vector, so that the information of multiple timing feature vectors can be aggregated into a comprehensive representation vector, which contains the information of all motor temperature timing features, and each feature is given an appropriate weight according to its importance, thereby obtaining a motor temperature timing node information aggregation representation vector.

[0038] In the above-mentioned Xenomai-based real-time Linux control method, the S56, the motor temperature timing node information aggregation representation vector and the processing requirement word granularity semantic encoding feature vector sequence are dynamically fused through a cross-domain feature attention gate guided by a key matrix to obtain a motor temperature-processing requirement semantic fusion feature vector. It should be understood that the motor temperature timing node information aggregation representation vector is the result of aggregating and representing the temperature data of the motor at different time points, capturing the temperature behavior of the motor during operation. The sequence of the processing requirement word granularity semantic encoding feature vector is obtained by semantic encoding the processing requirement text input by the user, capturing the deep meaning and contextual relationship of the text, but it may contain demand semantic information that is not satisfied with the current device status. Therefore, in order to match based on the motor temperature timing node information, the key and significant features in the processing requirement word granularity semantic encoding features are highlighted, the influence of irrelevant parts is reduced, thereby reducing the influence of irrelevant features, and improving processing stability and product quality. In the technical solution of the present application, the sequence of the motor temperature timing node information aggregation representation vector and the processing requirement word granularity semantic encoding feature vector is dynamically fused through a cross-domain feature attention gate guided by a key matrix to obtain a motor temperature-processing requirement semantic fusion feature vector.

[0039] Figure 5 The present invention is a schematic flow chart of a real-time Linux control method based on Xenomai in which the sequence of the motor temperature time series node information aggregation representation vector and the processing requirement word granularity semantic encoding feature vector are fused through a cross-domain feature attention gate dynamic fusion network guided by a key matrix to obtain a motor temperature-processing requirement semantic fusion feature vector. Figure 5As shown, optionally, in one embodiment of the present application, the motor temperature timing node information aggregation representation vector and the sequence of the processing requirement word granularity semantic encoding feature vectors are passed through a cross-domain feature attention gate dynamic fusion network guided by a key matrix to obtain a motor temperature-processing requirement semantic fusion feature vector, including: S561, linearly transforming each processing requirement word granularity semantic encoding feature vector in the sequence of the processing requirement word granularity semantic encoding feature vectors to obtain a processing requirement word granularity semantic initial key matrix, wherein each processing requirement word granularity semantic row vector in the processing requirement word granularity semantic initial key matrix is ​​each linearly transformed processing requirement word granularity semantic encoding feature vector; S562, calculating the feature distribution difference energy coefficient between the motor temperature timing node information aggregation representation vector and each processing requirement word granularity semantic row vector in the processing requirement word granularity semantic initial key matrix to obtain a set of motor temperature-processing requirement feature distribution difference energy coefficients; S563, comparing each motor temperature-processing requirement feature distribution difference energy coefficient in the set of the motor temperature-processing requirement feature distribution difference energy coefficients with a predetermined threshold A comparison is performed, and in response to the motor temperature-processing requirement feature distribution difference energy coefficient being less than a predetermined threshold, the mean vector between the processing requirement word granularity semantic row vector corresponding to the motor temperature-processing requirement feature distribution difference energy coefficient and the motor temperature timing node information aggregation representation vector is used as the updated processing requirement word granularity semantic row vector to obtain an updated processing requirement word granularity semantic key matrix; S564, the motor temperature timing node information aggregation representation vector is processed using a value embedding matrix to obtain a motor temperature timing node information aggregation value vector; S565, the motor temperature timing node information aggregation representation vector is used as a query vector, each updated processing requirement word granularity semantic row vector in the updated processing requirement word granularity semantic key matrix is ​​used as a key vector, and the motor temperature timing node information aggregation value vector is input into a cross-domain attention gate query module based on a Transformer structure to obtain a sequence of motor temperature-processing requirement cross-domain query attention vectors; S566, the positional mean vector of the sequence of the motor temperature-processing requirement cross-domain query attention vectors is calculated to obtain the motor temperature-processing requirement semantic fusion feature vector.

[0040] Optionally, in one embodiment of the present application, the feature distribution difference energy coefficient between the motor temperature timing node information aggregation representation vector and each processing requirement word granularity semantic row vector in the processing requirement word granularity semantic initial key matrix is ​​calculated to obtain a set of motor temperature-processing requirement feature distribution difference energy coefficients, including: respectively calculating the position difference, position dot multiplication and position addition between the motor temperature timing node information aggregation representation vector and the processing requirement word granularity semantic row vector to obtain the motor temperature-processing requirement differential vector, the motor temperature-processing requirement dot multiplication vector and the motor temperature-processing requirement sum vector; cascading the motor temperature-processing requirement differential vector, the motor temperature-processing requirement dot multiplication vector and the motor temperature-processing requirement sum vector and performing one-dimensional convolution encoding to obtain a motor temperature-processing requirement multi-dimensional fusion vector; performing local window-based maximum pooling processing on the motor temperature-processing requirement multi-dimensional fusion vector to obtain a motor temperature-processing requirement associated interaction vector; respectively extracting the maximum values ​​of the motor temperature-processing requirement associated interaction vectors to obtain the motor temperature-processing requirement. Calculate the maximum value of the associated interaction; calculate the mean and variance of the motor temperature-processing requirement associated interaction vector respectively to obtain the motor temperature-processing requirement associated interaction mean and the motor temperature-processing requirement associated interaction variance; add the motor temperature-processing requirement associated interaction variance and the hyperparameter to obtain the motor temperature-processing requirement associated interaction first difference energy coefficient; calculate the square of the difference between the motor temperature-processing requirement associated interaction maximum value and the motor temperature-processing requirement associated interaction mean to obtain the motor temperature-processing requirement associated interaction difference value; multiply the motor temperature-processing requirement associated interaction variance by constant two to obtain the modulated motor temperature-processing requirement associated interaction variance; add the motor temperature-processing requirement associated interaction difference value and the modulated motor temperature-processing requirement associated interaction variance to the hyperparameter to obtain the motor temperature-processing requirement associated interaction second difference energy coefficient; divide the motor temperature-processing requirement associated interaction first difference energy coefficient by the motor temperature-processing requirement associated interaction second difference energy coefficient to obtain the motor temperature-processing requirement characteristic distribution difference energy coefficient.

[0041] Optionally, in one embodiment of the present application, the motor temperature timing node information aggregation representation vector is used as a query vector, each update processing requirement word granularity semantic row vector in the update processing requirement word granularity semantic key matrix is ​​used as a key vector, and the motor temperature timing node information aggregation value vector is input into a cross-domain attention gate query module based on a Transformer structure to obtain a sequence of motor temperature-processing requirement cross-domain query attention vectors, including: calculating the product between the query vector and the transposed vector of the update processing requirement word granularity semantic row vector to obtain a query-update processing requirement word granularity association matrix, and then dividing the query-update processing requirement word granularity association matrix by the square root of the length of the update processing requirement word granularity semantic row vector by position to obtain a query-update processing requirement word granularity modulation matrix; inputting the query-update processing requirement word granularity modulation matrix into a softmax function to obtain a query-update processing requirement word granularity activation matrix; multiplying the query-update processing requirement word granularity activation matrix by the motor temperature timing node information aggregation value vector to obtain the motor temperature-processing requirement cross-domain query attention vector.

[0042] Specifically, in the embodiment of the present application, the sequence of the motor temperature timing node information aggregation representation vector and the processing requirement word granularity semantic encoding feature vector is processed through a cross-domain feature attention gate dynamic fusion network guided by a key matrix, and is processed by the following cross-domain fusion to obtain the motor temperature-processing requirement semantic fusion feature vector; wherein the cross-domain fusion formula is: ;in, represents a sequence of granular semantic encoding feature vectors of the processing requirement words, is the sequence of the granular semantic encoding feature vector of the processing requirement word The processing requirement word granularity semantic encoding feature vector, is the processing requirement word granularity semantic initial key matrix, is the first in the processing requirement word granularity semantic initial key matrix The processing requirement word granularity semantic row vector, The processing requirement word granularity semantic encoding feature vector sequence The processing requirement word granularity semantic encoding feature vector, is the weight matrix, is the bias vector, is the first in the processing requirement word granularity semantic initial key matrix The processing requirement word granularity semantic row vector, For Perform linear transformation operations, is the motor temperature timing node information aggregation representation vector, , and They are position difference, position dot multiplication and position addition respectively. For cascade processing, is a one-dimensional convolutional coding operation, is the maximum pooling operation, is the first in the sequence of motor temperature-processing demand correlation interaction vectors A motor temperature-processing demand correlation interaction vector, To extract the maximum value of a vector, and They are The mean and variance of is a hyperparameter, is the first in the set of the motor temperature-processing demand characteristic distribution difference energy coefficients The energy coefficient of the difference between the motor temperature and the processing demand characteristic distribution, is a predetermined threshold, and To update the processing requirement word granularity semantic key matrix Update the semantic row vector of the word granularity required for processing, and are the value embedding matrix and value bias vector respectively, is the motor temperature timing node information aggregation value vector, for The transposed vector of is the length of the semantic row vector of each update processing requirement word, yes function, is the first in the sequence of the motor temperature-processing requirement cross-domain query attention vector Motor temperature-processing requirement cross-domain query attention vector, is the number of vectors in the sequence of the motor temperature-processing requirement cross-domain query focus vector, It is the motor temperature-processing requirement semantic fusion feature vector.

[0043] Specifically, first, a linear transformation is performed on the sequence of the processing requirement word granularity semantic encoding feature vectors to obtain a processing requirement word granularity semantic initial key matrix composed of multiple linearly transformed vectors to align the dimensions of each feature. Next, by comparing the difference between the query feature vector and each row vector in the processing requirement word granularity semantic initial key matrix, a set is obtained, which reflects the semantic difference measure between the motor temperature and the processing requirement word granularity semantic features. In other words, the query vector is regarded as a reference point to quantify the semantic distance between each processing requirement word granularity semantic row vector and the query vector. In particular, since the query feature vector serves as a common reference standard, these difference measurement coefficient sets also imply the relative semantic connection between each processing requirement word granularity semantic row vector in the key matrix. Subsequently, the calculated feature distribution difference energy coefficient is compared with the preset threshold. When the difference coefficient is less than this threshold, it means that the corresponding processing requirement word granularity semantic row vector is semantically close enough to the query vector (motor temperature timing node information aggregation representation vector). Therefore, the mean of the query feature vector and the corresponding processing requirement word granularity semantic row vector is used to update the row vector in the key matrix. This step aims to dynamically enhance the features with high semantic similarity to the query vector in the key matrix, reduce the requirements that are not satisfied with the current state of the equipment, and obtain the updated processing requirement word granularity semantic key matrix. Next, the motor temperature timing node information aggregation representation vector is linearly transformed to generate the corresponding value vector, and the row vector in the updated processing requirement word granularity semantic key matrix is ​​used as the key vector. The query vector, value vector and these key vectors are input into the cross-domain attention gate query module based on the Transformer structure to realize the local cross-domain semantic query matching and semantic association encoding between the query vector and the key vector, thereby promoting the cross-domain semantic information interaction between the two, and obtaining the sequence of motor temperature-processing requirement cross-domain query attention vectors. Finally, the sequence of query attention vectors is mean-calculated to obtain the final motor temperature-processing requirement semantic fusion feature vector. In this way, the most relevant and important processing requirement word granularity semantic information can be matched based on the motor temperature timing node information aggregation representation vector, thereby improving the stability of the processing process.

[0044] In the above-mentioned real-time Linux control method based on Xenomai, the S57, based on the motor temperature-processing requirement semantic fusion feature vector, obtains a control result, and the control result is used to represent the real-time spindle speed recommendation value of the processing machine tool. Optionally, in one embodiment of the present application, based on the motor temperature-processing requirement semantic fusion feature vector, obtains a control result, and the control result is used to represent the real-time spindle speed recommendation value of the processing machine tool, including: passing the motor temperature-processing requirement semantic fusion feature vector through a spindle speed Linux controller based on a decoder to obtain the control result, and the control result is used to represent the real-time spindle speed recommendation decoded value of the processing machine tool. That is, the motor temperature-processing requirement semantic fusion feature vector obtained by cross-domain fusion of the sequence of the motor temperature time series node information aggregation representation vector and the processing requirement word granularity semantic encoding feature vector is decoded and processed to automatically recommend the real-time spindle speed of the processing machine tool. In this way, the real-time enhanced industrial computer can quickly process data and provide instant response, which has higher real-time performance than the traditional PLC control system. At the same time, it can automatically recommend the spindle speed based on the dynamic fusion characteristics of processing requirements and motor temperature, thereby improving the degree of automation in processing.

[0045] Optionally, in one embodiment of the present application, the motor temperature-processing requirement semantic fusion feature vector is passed through a decoder-based spindle speed Linux controller to obtain the control result, including: using a decoder to decode and regress the motor temperature-processing requirement semantic fusion feature vector using the following decoding formula to obtain the control result, wherein the decoding formula is: ,in, is the motor temperature-processing requirement semantic fusion feature vector, is the control result, is the decoding weight matrix, Represents matrix multiplication.

[0046] Preferably, considering that the sequence of the motor temperature timing node information aggregation representation vector and the processing requirement word granularity semantic encoding feature vector respectively represent the timing node significance attenuation aggregation features of the one-dimensional timing local correlation features of the motor temperature value of the processing machine tool and the word embedding semantic encoding features of the processing requirements, when they are dynamically fused through the cross-domain feature attention gate guided by the key matrix, there will also be a multimodal semantic feature aggregation mapping offset caused by the inconsistent cross-domain feature distribution under different feature modes, so as to further improve the multimodal semantic feature fusion comprehensibility of the motor temperature-processing requirement semantic fusion feature vector, thereby improving the accuracy of the control result obtained by the motor temperature-processing requirement semantic fusion feature vector through the decoder-based spindle speed Linux controller.

[0047] Therefore, the present application optimizes the motor temperature-processing requirement semantic fusion feature vector when passing the motor temperature-processing requirement semantic fusion feature vector through a decoder-based spindle speed Linux controller, including the steps of: subtracting the 0 norm of the motor temperature-processing requirement semantic fusion feature vector from the length of the motor temperature-processing requirement semantic fusion feature vector to obtain the motor temperature-processing requirement semantic fusion isolated representation value.

[0048] The logarithm value with base 2 of the square of the motor temperature-processing requirement semantic fusion isolated representation value and the sum of the motor temperature-processing requirement semantic fusion isolated representation value is calculated to obtain the motor temperature-processing requirement semantic fusion information order value.

[0049] A power function is calculated with each eigenvalue of the motor temperature - processing requirement semantic fusion feature vector as the base, the difference between the motor temperature - processing requirement semantic fusion isolated representation value and one as the exponent, and point multiplication is performed with the motor temperature - processing requirement semantic fusion information order value to obtain the motor temperature - processing requirement semantic fusion leading vector.

[0050] The motor temperature-processing requirement semantic fusion feature vector is dot-multiplied with the difference of the motor temperature-processing requirement semantic fusion isolated representation value minus one, and then dot-multiplied with the inverse of the motor temperature-processing requirement semantic fusion isolated representation value to obtain the motor temperature-processing requirement semantic fusion field constraint vector.

[0051] An exponential function with a natural constant as the base and each eigenvalue of the motor temperature-processing requirement semantic fusion field constraint vector as the exponent is calculated to obtain a motor temperature-processing requirement semantic fusion field bias vector.

[0052] The motor temperature-processing requirement semantic fusion leading vector is added to the motor temperature-processing requirement semantic fusion field bias vector to obtain an optimized motor temperature-processing requirement semantic fusion feature vector. Finally, the optimized motor temperature-processing requirement semantic fusion feature vector is passed through a decoder-based spindle speed Linux controller to obtain the control result.

[0053] Among them, the motor temperature-processing requirement semantic fusion feature vector The optimization process is expressed as: ;in, represents the motor temperature-processing requirement semantic fusion feature vector, represents the length of the motor temperature-processing requirement semantic fusion feature vector, represents the zero norm of the eigenvector, Indicates the motor temperature-processing requirement semantic fusion isolated representation value, It means point multiplication by position. To calculate a power function with each eigenvalue of the motor temperature-processing requirement semantic fusion feature vector as the base and the difference between the motor temperature-processing requirement semantic fusion isolated representation value and one as the exponent, Indicates adding by position point. represents the natural exponential function, Represents the optimized motor temperature-processing requirement semantic fusion feature vector.

[0054] Therefore, for the high-dimensional feature manifold of the motor temperature-processing requirement semantic fusion feature vector, the feature value of the feature set is used as the vector field representation of the aggregation dimension, and the superposition value of the vector field of the motor temperature-processing requirement semantic fusion feature vector at the isolated zero position is used as the order information to fix the local position of the feature value of its feature set, and add the bias of the reversibility of the feature regression distribution field of the motor temperature-processing requirement semantic fusion feature vector as a reward to achieve the mapping target tracking of the regression distribution of the motor temperature-processing requirement semantic fusion feature vector for the feature value position, so that the feature set of the motor temperature-processing requirement semantic fusion feature vector perceives the mapping migration to the aggregate distribution, thereby improving the accuracy of the control result obtained by the decoder-based spindle speed Linux controller by improving the comprehensibility of the multimodal feature fusion mapping of the motor temperature-processing requirement semantic fusion feature vector. In this way, the industrial computer can quickly process data and provide instant response through real-time enhancement, which has higher real-time performance than the traditional PLC control system. At the same time, the spindle speed can be automatically recommended according to the dynamic fusion characteristics of the processing demand and the motor temperature, which improves the degree of automation of the processing. Specifically, in the technical solution of the present application, first, the processing requirements input by the user are extracted from the human-computer interaction system, and the time series of the motor temperature values ​​of the processing machine tool collected by the temperature sensor deployed on the processing machine tool is obtained, and the processing requirements and the time series of the motor temperature values ​​of the processing machine tool are transmitted to the real-time enhanced industrial computer.

[0055] In summary, the real-time Linux control method based on Xenomai according to the embodiment of the present application is explained. On the A40i-H dual-bus controller platform, the Xenomai real-time patch is first transplanted to the mainline Linux kernel and configured, compiled and tested, then the libevl library is compiled in a real-time environment, and the jitter of the evl core thread switching is tested to verify the integrity and correctness of the real-time patch, and then the real-time drive is adapted and connected to the driver, and finally an application for controlling the jitter of the processing machine tool is developed and tested to perform actual machine tool jitter control. In this way, by transplanting the Xenomai real-time patch and optimizing the configuration, the jitter of the system under no-load conditions can be reduced, thereby meeting the needs of precision machining and high-precision control applications. At the same time, it has a higher real-time response speed when processing high-frequency tasks, which improves the real-time performance of the control.

[0056] Specifically, a Xenomai-based real-time Linux controller is also provided, and the Xenomai-based real-time Linux controller is used to execute the aforementioned Xenomai-based real-time Linux control method.

[0057] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference to a figure in a claim should not be considered as limiting the claim to which it relates.

[0058] In addition, it is obvious that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units stated in the device claim can also be implemented by one unit through software or hardware.

[0059] Finally, it should be noted that the above description has been given for the purpose of illustration and description. In addition, the above embodiments are only used to illustrate the technical solution of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A real-time Linux control method based on Xenomai, characterized in that: include: Port the Xenomai real-time patch to the mainline Linux kernel on the A40i-H dual-bus controller platform, adapt the relevant configuration options, and compile, package and test the ported mainline Linux kernel; Compile the libevl library in the Linux real-time environment, and compile and package the test evl core thread switching jitter to verify the integrity and correctness of the Xenomai real-time patch transplantation; Adapt to real-time drive and connect to the drive; Write machine tool jitter control applications and perform functional tests; Using the machining machine tool jitter control application to perform jitter control; Among them, porting the Xenomai real-time patch to the mainline Linux kernel on the A40i-H dual-bus controller platform, adapting the relevant configuration options, and compiling, packaging and testing the ported mainline Linux kernel, including: Ensure that a cross-compilation tool chain has been installed in the development environment, and the cross-compilation tool chain is suitable for the architecture of the A40i-H dual-bus controller platform; Download the mainline Linux kernel and the Xenomai real-time patch; Determining compatibility between the Xenomai real-time patch and the mainline Linux kernel; Apply the Xenomai real-time patch to the mainline Linux kernel using the patch tool; Use make menuconfig or make xconfig to configure kernel options and enable the real-time options required by the Xenomai real-time patch; According to the hardware characteristics of the A40i-H dual-bus controller platform, adjust relevant configurations, including interrupt processing and scheduling strategies; Execute the make command to compile the kernel and ensure that there are no compilation errors; Generate a kernel image and module package suitable for the A40i-H dual-bus controller platform; Deploy the kernel image and module package to the A40i-H dual-bus controller platform; Start the system to ensure that the mainline Linux kernel can run normally and load the Xenomai real-time patch; Verify the basic functions of the system, including device driver loading and network connection; Among them, compile the libevl library in the Linux real-time environment, and compile and package to test the jitter of the evl core thread switching to verify the integrity and correctness of the Xenomai real-time patch transplantation, including: Download the latest source code of libevl; Set up the cross-compilation environment to ensure that the compilation is targeted at the A40i-H dual-bus controller platform; Execute the compilation command to generate the libevl library file; Use the libevl library file to write a test program to measure the switching jitter of the core thread; Running the test program on the A40i-H dual-bus controller platform and recording jitter data; Compare the jitter data to see whether they meet the real-time performance requirements.

2. The Xenomai-based real-time Linux control method according to claim 1, characterized in that: The jitter control application for machining machine tools is used to perform jitter control, including: Extracting user input processing requirements from human-computer interaction systems; Acquire a time series of motor temperature values ​​of the processing machine tool collected by a temperature sensor deployed on the processing machine tool; The processing requirements are segmented and then passed through a semantic encoder including a word embedding layer to obtain a sequence of processing requirement word granularity semantic encoding feature vectors; After the time series of the motor temperature value of the machining tool is regularized according to the time dimension, the motor temperature time series encoder based on 1D-CNN is used to obtain a sequence of motor temperature time series feature vectors; The sequence of the motor temperature time series feature vectors is passed through a node feature integrator guided by significance attenuation to obtain a motor temperature time series node information aggregation representation vector; The motor temperature time series node information aggregation representation vector and the processing requirement word granularity semantic encoding feature vector are passed through a cross-domain feature attention gate dynamic fusion network guided by a key matrix to obtain a motor temperature-processing requirement semantic fusion feature vector; Based on the motor temperature-processing requirement semantic fusion feature vector, a control result is obtained, and the control result is used to represent a real-time spindle speed recommendation value of the processing machine tool.

3. The real-time Linux control method based on Xenomai according to claim 2, characterized in that: The sequence of the motor temperature time series feature vectors is passed through a node feature integrator guided by significance attenuation to obtain a motor temperature time series node information aggregation representation vector, including: Calculating a characteristic significance description factor of each motor temperature timing feature vector in the sequence of the motor temperature timing feature vectors, wherein the characteristic significance description factor is related to the mean and variance of each motor temperature timing feature vector; Constructing a characteristic significance attenuation factor of each motor temperature time series feature vector based on the distance span between each motor temperature time series feature vector in the sequence of the motor temperature time series feature vector and the current motor temperature time series feature vector; Modulating the characteristic significance description factors of each motor temperature timing feature vector based on the characteristic significance attenuation factors of each motor temperature timing feature vector to obtain a sequence of motor temperature timing feature significance attenuation description factors; Inputting the sequence of motor temperature timing feature significance attenuation description factors into a gated mask module to obtain a sequence of motor temperature timing feature significance attenuation weight factors; The sequence of the motor temperature timing feature significance attenuation weight factors is used as a sequence of weights, and the weighted sum of the sequence of the motor temperature timing feature vectors is calculated to obtain the motor temperature timing node information aggregation representation vector.

4. The Xenomai-based real-time Linux control method according to claim 3, characterized in that: Calculating the characteristic significance description factor of each motor temperature time series feature vector in the sequence of the motor temperature time series feature vectors includes: Calculating the mean and variance of the motor temperature timing characteristic vector respectively to obtain the motor temperature timing characteristic mean and the motor temperature timing characteristic variance; Calculating the position difference between the motor temperature timing characteristic vector and the motor temperature timing characteristic mean to obtain the motor temperature timing difference vector; Calculating the fourth power of each eigenvalue in the motor temperature timing difference vector to obtain the motor temperature timing modulation difference vector; Calculating the expected value of the motor temperature timing modulation difference vector to obtain the motor temperature timing expected value; The motor temperature timing expected value is divided by the square of the motor temperature timing characteristic variance to obtain a characteristic significance description factor corresponding to the motor temperature timing characteristic vector.

5. The Xenomai-based real-time Linux control method according to claim 4, characterized in that: The motor temperature time series node information aggregation representation vector and the processing requirement word granularity semantic encoding feature vector are subjected to a cross-domain feature attention gate dynamic fusion network guided by a key matrix to obtain a motor temperature-processing requirement semantic fusion feature vector, including: Performing a linear transformation on each processing requirement word granularity semantic encoding feature vector in the sequence of processing requirement word granularity semantic encoding feature vectors to obtain a processing requirement word granularity semantic initial key matrix, wherein each processing requirement word granularity semantic row vector in the processing requirement word granularity semantic initial key matrix is ​​a processing requirement word granularity semantic encoding feature vector after each linear transformation; Calculating the characteristic distribution difference energy coefficient between the motor temperature timing node information aggregation representation vector and each processing requirement word granularity semantic row vector in the processing requirement word granularity semantic initial key matrix to obtain a set of motor temperature-processing requirement characteristic distribution difference energy coefficients; Compare each motor temperature-processing requirement feature distribution difference energy coefficient in the set of the motor temperature-processing requirement feature distribution difference energy coefficient with a predetermined threshold value, and in response to the motor temperature-processing requirement feature distribution difference energy coefficient being less than the predetermined threshold value, use the mean vector between the processing requirement word granularity semantic row vector corresponding to the motor temperature-processing requirement feature distribution difference energy coefficient and the motor temperature timing node information aggregation representation vector as the updated processing requirement word granularity semantic row vector to obtain an updated processing requirement word granularity semantic key matrix; Using a value embedding matrix to process the motor temperature timing node information aggregation representation vector to obtain a motor temperature timing node information aggregation value vector; The motor temperature timing node information aggregation representation vector is used as a query vector, each updated processing requirement word granularity semantic row vector in the updated processing requirement word granularity semantic key matrix is ​​used as a key vector, and the motor temperature timing node information aggregation value vector is input into a cross-domain attention gate query module based on a Transformer structure to obtain a sequence of motor temperature-processing requirement cross-domain query attention vectors; The position-wise mean vector of the sequence of the motor temperature-processing requirement cross-domain query focus vector is calculated to obtain the motor temperature-processing requirement semantic fusion feature vector.

6. The Xenomai-based real-time Linux control method according to claim 5, characterized in that: Calculating the characteristic distribution difference energy coefficient between the motor temperature timing node information aggregation representation vector and each processing requirement word granularity semantic row vector in the processing requirement word granularity semantic initial key matrix to obtain a set of motor temperature-processing requirement characteristic distribution difference energy coefficients, including: Respectively calculating the position difference, position dot product and position addition between the motor temperature timing node information aggregation representation vector and the processing requirement word granularity semantic row vector to obtain the motor temperature-processing requirement difference vector, the motor temperature-processing requirement dot product vector and the motor temperature-processing requirement sum vector; The motor temperature-processing requirement differential vector, the motor temperature-processing requirement point product vector and the motor temperature-processing requirement sum vector are cascaded and then one-dimensional convolution encoding is performed to obtain a motor temperature-processing requirement multi-dimensional fusion vector; Performing a maximum pooling process based on a local window on the multi-dimensional fusion vector of motor temperature and processing requirements to obtain a motor temperature and processing requirement correlation interaction vector; Extracting the maximum values ​​of the motor temperature-processing requirement correlation interaction vectors respectively to obtain the motor temperature-processing requirement correlation interaction maximum value; Respectively calculating the mean and variance of the motor temperature-processing requirement associated interaction vector to obtain the motor temperature-processing requirement associated interaction mean and the motor temperature-processing requirement associated interaction variance; Adding the motor temperature-processing requirement correlation interaction variance and the hyperparameter to obtain the motor temperature-processing requirement correlation interaction first difference energy coefficient; Calculating the square of the difference between the maximum value of the motor temperature-processing requirement association interaction and the mean value of the motor temperature-processing requirement association interaction to obtain a motor temperature-processing requirement association interaction difference value; Multiplying the motor temperature-processing requirement correlation interaction variance by a constant of two to obtain a modulated motor temperature-processing requirement correlation interaction variance; Adding the motor temperature-processing requirement association interaction difference value and the modulated motor temperature-processing requirement association interaction variance to the hyper parameter to obtain a motor temperature-processing requirement association interaction second difference energy coefficient; The first difference energy coefficient of the motor temperature-processing requirement association interaction is divided by the second difference energy coefficient of the motor temperature-processing requirement association interaction to obtain the motor temperature-processing requirement characteristic distribution difference energy coefficient.

7. The Xenomai-based real-time Linux control method according to claim 6, characterized in that: Based on the motor temperature-processing requirement semantic fusion feature vector, a control result is obtained, and the control result is used to represent the real-time spindle speed recommended value of the processing machine tool, including: passing the motor temperature-processing requirement semantic fusion feature vector through a decoder-based spindle speed Linux controller to obtain the control result, and the control result is used to represent the real-time spindle speed recommended decoded value of the processing machine tool.

8. A real-time Linux controller based on Xenomai, characterized in that: The Xenomai-based real-time Linux controller is used to execute the Xenomai-based real-time Linux control method described in any one of claims 1-7.

Images