A dynamic configuration switching system and control method in the form of serialization annotation

Abstract

The present invention discloses a dynamic configuration switching system and control method in the form of serialization annotations, which obtains the real-time operating state of the industrial device; inputs the real-time operating state into a preset parameter configuration model to output a dynamic configuration parameter set; and loads a number of configuration parameters according to the dynamic configuration parameter set; wherein the dynamic configuration parameter set is a number of configuration parameters output by the parameter configuration model and represented in the format of serialization annotations; the present invention automatically outputs a dynamic configuration parameter set according to the real-time operating state of the industrial device through a multi-task learning model, and represents it in the format of serialization annotations, so as to realize the real-time dynamic configuration switching of the device. In particular, the present invention simplifies the input of multiple parameters through principal component analysis while retaining the most core data structure, improving the dynamic adjustment function of the configuration parameters.

Description

Technical Field

[0001] The present invention relates to a configuration switching system, and specifically to a dynamic configuration switching system and control method in the form of serialized annotations. Background Art

[0002] During the operation of industrial equipment, the configuration parameters of the equipment (such as temperature setting, current threshold, working mode, etc.) have an important impact on the performance and efficiency of the equipment. In modern industrial equipment, there are already some methods for automatically changing configuration parameters based on the operating state. For example, machine learning adjustment based on multiple parameters, etc.; however, the multi-parameter input increases the complexity of data processing, requires processing high-dimensional data, and is easily affected by noise. In addition, the correlation between multiple parameters causes data redundancy, affecting the training efficiency and prediction accuracy of machine learning. The patent document with the patent publication number CN118467212A discloses a method and system for processing industrial equipment data, which can adjust the resource configuration of each industrial equipment.

[0003] However, it is difficult for the above-mentioned literature and the prior art to achieve the problem of retaining the core data structure when inputting multiple parameters to reduce data redundancy in the processing process. For this reason, the present invention provides a dynamic configuration switching system and control method in the form of serialized annotations. Summary of the Invention

[0004] Aiming at the deficiencies of the prior art, the present invention provides a dynamic configuration switching system and control method in the form of serialized annotations, which solves the technical problems proposed in the background art by reducing the dimensionality of multi-parameter data.

[0005] This embodiment provides a dynamic configuration switching system and control method in the form of serialized annotations, exploring how to solve the problem of retaining the core data structure when inputting multiple parameters to reduce data redundancy in the processing process.

[0006] In a first aspect, the present invention provides a dynamic configuration switching control method in the form of serialized annotations, which is used in an industrial equipment server. The method includes:

[0007] Obtain the real-time operating state of the industrial equipment;

[0008] Input the real-time operating state into a preset parameter configuration model to output a set of dynamic configuration parameters;

[0009] Load a number of configuration parameters according to the set of dynamic configuration parameters; wherein, the set of dynamic configuration parameters is output by the parameter configuration model and represents a number of configuration parameters in a serialized annotation format.

[0010] In some of these embodiments, the modeling method of the parameter configuration model includes:

[0011] S1. Obtain the multi-parameter trend features of the industrial equipment within n same time periods and their corresponding historical configuration parameter sets;

[0012] S2. Use the multi-parameter trend features as input features and the corresponding historical configuration parameter sets as target labels to construct model training samples;

[0013] S3. Aggregate n model training samples to obtain a parameter training set;

[0014] S4. Input the model training samples of the parameter training set into a multi-task learning model to train the parameter configuration model.

[0015] In some embodiments, obtaining the multi-parameter trend features of the industrial equipment within n same time periods includes:

[0016] S1-1. Collect n operating parameter variables of the industrial equipment within a reference time length, where the n operating parameter variables include: temperature change rate, current fluctuation amplitude, and other selected operating parameter variables;

[0017] S1-2. Perform principal component vector analysis on the n operating parameter variables to obtain the multi-parameter trend features.

[0018] In some embodiments, collecting n operating parameter variables of the industrial equipment within a reference time length, where the n operating parameter variables include: temperature change rate, current fluctuation amplitude, and other selected operating parameter variables, includes:

[0019] S1-1-1. Define a reference time length, which includes a start time point and an end time point;

[0020] S1-1-2. Collect the equipment temperature, current, and other selected operating parameters at the start time point and the end time point respectively to calculate the operating parameter variables within the reference time length.

[0021] In some embodiments, the principal component vector analysis of the n operating parameter variables includes:

[0022] S1-2-1. Standardize the n operating parameter variables to obtain standardized parameters and combine them into a standardized matrix;

[0023] S1-2-2. Calculate the covariance matrix according to the standardized matrix;

[0024] S1-2-3. Perform singular value decomposition on the covariance matrix to obtain eigenvalues and their corresponding eigenvectors;

[0025] S1-2-4. Arrange the eigenvalues in descending order, determine the principal component vectors according to the preset variance interpretation rate, and construct a principal component vector matrix based on the principal component vectors;

[0026] Further, the construction of the principal component matrix is as follows: Arrange each selected principal component vector as a column vector in sequence to form a principal component vector matrix.

[0027] S1-2-5. Project the standardized matrix onto the principal component vector matrix to obtain a matrix after dimensionality reduction, where the elements of the matrix after dimensionality reduction are the multi-parameter trend features.

[0028] The expression of the matrix after dimensionality reduction is: ;

[0029] where represents the matrix after dimensionality reduction, represents the principal component vector matrix, with the dimension of , and each of its elements is expressed as , which represents the projection value of the i-th parameter sample on the j-th principal component vector, and the projection value is the numerical representation of the multi-parameter trend feature.

[0030] In some of these embodiments, the standardization process for n operating parameter variables is performed to obtain standardized parameters and combine them into a standardized matrix, including:

[0031] S1-2-1-1. Collect n operating parameter variables as parameter samples and construct an independent data set for each operating parameter;

[0032] where the expression of the independent data set for each operating parameter is: , represents the k-th operating parameter variable of the i-th parameter sample;

[0033] S1-2-1-2. Calculate the mean and standard deviation of each independent data set of the operating parameter to obtain the mean and standard deviation of the operating parameter variable;

[0034] The expression of the mean of the operating parameter variable is: ;

[0035] represents the mean of the k-th operating parameter variable, represents the k-th operating parameter variable of the i-th parameter sample;

[0036] The expression of the standard deviation of the operating parameter variable is: ; represents the standard deviation of the k-th operating parameter variable;

[0037] S1-2-1-3. Standardize each operating parameter variable using the mean and standard deviation of each operating parameter variable to obtain n standardized operating parameter variables;

[0038] The expression of the standardized operating parameter variable is: ; represents the k-th standardized operating parameter variable of the i-th parameter sample;

[0039] S1-2-1-4. Combine the n standardized operating parameter variables into a standardized matrix ;

[0040] ;

[0041] represents the index of the operating parameter, represents the index of the parameter sample, m represents the total number of operating parameter variables, n represents the number of parameter samples, represents the k-th standardized operating parameter variable of the i-th parameter sample.

[0042] In some embodiments, calculating the covariance matrix according to the standardized matrix includes:

[0043] S1-2-2-1. Transpose the standardized matrix to obtain a transposed matrix ;

[0044] The expression of the transposed matrix is: ;

[0045] S1-2-2-2. Multiply the transposed matrix and the standardized matrix to obtain a product matrix ;

[0046] The expression of the product matrix is: ;

[0047] Among them, the elements of the product matrix are characterized as the unnormalized values of the elements in the covariance matrix;

[0048] S1-2-2-3. Unbiasedly estimate the unnormalized value of the product matrix to obtain the covariance matrix ;

[0049] The expression of the covariance matrix is: ;

[0050] Among them, the covariance matrix has elements defined as follows: ;

[0051] ;

[0052] represents the covariance between the k-th standardized operating parameter variable and the j-th standardized operating parameter variable.

[0053] In some embodiments, the singular value decomposition of the covariance matrix is performed to obtain eigenvalues and their corresponding eigenvectors, including:

[0054] S1-2-3-1. Perform singular value decomposition on the covariance matrix to obtain a decomposed matrix;

[0055] The expression of the decomposed matrix is: ;

[0056] Among them, represents the first orthogonal matrix, D represents a diagonal matrix, and its diagonal elements are singular values, is the second orthogonal matrix;

[0057] S1-2-3-2. Select the column vectors of the first orthogonal matrix as the eigenvectors of the covariance matrix, and the eigenvectors are used to characterize the principal component vectors;

[0058] S1-2-3-3. Select the squares of the singular values of the diagonal matrix as the eigenvalues of the covariance matrix, and the eigenvalues are used to characterize the variance magnitudes of the principal component vectors.

[0059] In some embodiments, the eigenvalues are sorted in descending order, and the principal component vectors are determined according to a preset variance interpretation rate, including:

[0060] S1-2-4-1. Sort the several eigenvalues obtained after singular value decomposition in descending order to obtain a sorted eigenvalue sequence;

[0061] S1-2-4-2. Calculate the cumulative variance interpretation rate of the eigenvalues in descending order in the eigenvalue sequence until the cumulative variance interpretation rate reaches the preset variance interpretation rate;

[0062] S1-2-4-3. Select the eigenvectors corresponding to the eigenvalues that satisfy the cumulative variance interpretation rate and use them as the principal component vectors.

[0063] Compared with the prior art, a dynamic configuration switching control method in the form of serialized annotations according to the present invention uses a multi-task learning model to automatically output a set of dynamic configuration parameters based on the real-time operating state of industrial equipment, and represents them in the format of serialized annotations to achieve real-time dynamic configuration switching of the equipment. In particular, the present invention simplifies the input of multiple parameters through principal component analysis while retaining the most core data structure, improving the training efficiency and prediction accuracy of the model. It solves the problems in the prior art such as complex multi-parameter input, insufficient generalization ability of single-task models, and limited ability of real-time monitoring systems to handle complex relationships, and improves the dynamic adjustment function of configuration parameters.

[0064] In a second aspect, the present invention provides a dynamic configuration switching system in the form of serialized annotations, including:

[0065] A real-time operating state acquisition module for acquiring the real-time operating state of the industrial equipment;

[0066] A configuration parameter set output module for inputting the real-time operating state into a preset parameter configuration model and outputting a set of dynamic configuration parameters;

[0067] A configuration parameter loading module for loading a number of configuration parameters according to the set of dynamic configuration parameters; wherein, the set of dynamic configuration parameters is a number of configuration parameters output by the parameter configuration model and represented in the format of serialized annotations.

[0068] Compared with the prior art, the beneficial effects of a dynamic configuration switching system in the form of serialized annotations according to the present invention are the same as those of the above-mentioned dynamic configuration switching control method in the form of serialized annotations, so they will not be elaborated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1 It is a flowchart of a dynamic configuration switching control method in the form of serialized annotations according to the present invention;

[0070] Figure 2 It is a modeling flowchart of the parameter configuration model according to the present invention;

[0071] Figure 3 It is a structural block diagram of a dynamic configuration switching system in the form of serialized annotations according to the present invention; DETAILED DESCRIPTION OF THE EMBODIMENTS

[0072] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0073] Example 1: Please refer to Figure 1 , the present invention provides a dynamic configuration switching control method in the form of serialization annotation, which is used in an industrial equipment server. The method includes:

[0074] Obtain the real-time operating status of the industrial equipment;

[0075] Input the real-time operating status into a preset parameter configuration model to output a set of dynamic configuration parameters;

[0076] Load a number of configuration parameters according to the set of dynamic configuration parameters; wherein, the set of dynamic configuration parameters is output by the parameter configuration model and represents a number of configuration parameters in the format of serialization annotation.

[0077] Specifically, by inputting the real-time operating status of the industrial equipment into the parameter configuration model, a set of dynamic configuration parameters based on the real-time operating status is obtained, so as to realize the real-time dynamic configuration switching of the industrial equipment. Among them, the set of dynamic configuration parameters includes a number of configuration parameters that conform to the current industrial equipment.

[0078] Among them, a number of configuration parameters represented in the format of serialization annotation means that the configuration parameters and their related information are serialized in a structured manner and accompanied by annotation information, thereby forming a set of dynamic configuration parameters; in the industrial equipment server, the dynamic configuration switching control method can output a set of predicted dynamic configuration parameters according to the real-time operating status of the industrial equipment through a preset parameter configuration model. The configuration parameters therein are represented in the format of serialization annotation, which can be conveniently loaded into the device to realize real-time adjustment and optimization of the operating status of the device. For example, the configuration parameters may include temperature setting, current threshold, working mode, etc. These parameters are represented in JSON format and accompanied by type, unit, and description information.

[0079] Example 2: The technical solution of this Example 2 is different from that of Example 1 in that it discloses a modeling method of the parameter configuration model. Please refer to Figure 2 , the method includes:

[0080] S1. Obtain the multi-parameter trend characteristics of the industrial equipment and their corresponding historical configuration parameter sets within n same time periods;

[0081] S2. Take the multi-parameter trend characteristics as input features and the corresponding historical configuration parameter sets as target labels to construct model training samples;

[0082] S3. Combine n model training samples to obtain a parameter training set;

[0083] S4. Input the model training samples of the parameter training set into a multi-task learning model to train the parameter configuration model.

[0084] Specifically, by using multi-parameter trend features as input features and historical configuration parameter sets as target labels, and training with a multi-task learning model, a parameter configuration model can be accurately constructed to capture the configuration parameter requirements of industrial equipment in different operating states. Among them, the historical configuration parameter set is actually a set of optimal parameters of the industrial equipment in different operating states within a historical time period. Of course, the optimal parameters are actually determined based on the performance of the industrial equipment. After receiving the feedback of the performance, the configuration parameters corresponding to the best performance are stored as historical data, and the configuration parameters in multiple historical data can be summarized to obtain the historical configuration parameter set.

[0085] S1 specifically includes:

[0086] S1-1. Collect n operating parameter variables of the industrial equipment within a reference time length. The n operating parameter variables include: temperature change rate, current fluctuation amplitude, and other selected operating parameter variables;

[0087] S1-2. Conduct principal component vector analysis on the n operating parameter variables to obtain multi-parameter trend features.

[0088] Specifically, collecting the multi-parameter trend features of the industrial equipment within the same time period can comprehensively reflect the multi-dimensional changes of the industrial equipment during a specific operating time, providing richer input features for the parameter configuration model. For example, parameters such as temperature change rate and current fluctuation amplitude can reflect the operating state and performance changes of the industrial equipment.

[0089] S1-1 specifically includes:

[0090] S1-1-1. Define the reference time length, which includes a start time point and an end time point;

[0091] S1-1-2. Collect the equipment temperature, current, and other selected operating parameters at the start time point and the end time point respectively to calculate the operating parameter variables within the reference time length.

[0092] Specifically, by defining the reference time length and collecting multiple operating parameters within this time length, the consistency of the operating parameters is ensured.

[0093] S1-2 specifically includes:

[0094] S1-2-1. Standardize the n operating parameter variables to obtain standardized parameters and combine them into a standardized matrix;

[0095] S1-2-2. Calculate the covariance matrix based on the standardized matrix;

[0096] S1-2-3. Perform singular value decomposition on the co-defense difference matrix to obtain eigenvalues and their corresponding eigenvectors;

[0097] S1-2-4. Arrange the eigenvalues in descending order, and determine the principal component vectors according to the preset variance interpretation rate; and construct a principal component vector matrix based on the principal component vectors;

[0098] Furthermore, the principal component matrix is constructed as follows: Each selected principal component vector is arranged as a column vector in sequence to form a principal component vector matrix.

[0099] S1-2-5. Project the standardized matrix onto the principal component vector matrix to obtain a matrix after dimensionality reduction, where the elements of the matrix after dimensionality reduction are multi-parameter trend features.

[0100] The expression of the matrix after dimensionality reduction is: ;

[0101] where, represents the matrix after dimensionality reduction, represents the principal component vector matrix, with a dimension of , and each of its elements is expressed as , which represents the projection value of the i-th parameter sample on the j-th principal component vector, and the projection value is the numerical representation of the multi-parameter trend feature.

[0102] Specifically, principal component vector analysis is used to process multiple operating parameter variables. The multi-parameter trend features obtained through dimensionality reduction simplify the model input data structure while retaining the main change trends of the data.

[0103] S1-2-1 specifically includes:

[0104] S1-2-1-1. Collect n operating parameter variables as parameter samples and construct an independent data set for each operating parameter;

[0105] where, the expression of the independent data set for each operating parameter is: , represents the k-th operating parameter variable of the i-th parameter sample;

[0106] For example:

[0107] The expression of the first data set is: represents the i-th temperature change rate;

[0108] The expression of the second data set is: , represents the i-th current fluctuation amplitude;

[0109] S1-2-1-2. Calculate the mean and standard deviation of each independent data set of the operating parameter to obtain the mean and standard deviation of the operating parameter variable;

[0110] The mean expression of the operating parameter variable is: ;

[0111] represents the mean of the k-th operating parameter variable, represents the k-th operating parameter variable of the i-th parameter sample;

[0112] The standard deviation expression of the operating parameter variable is: ; represents the standard deviation of the k-th operating parameter variable;

[0113] S1-2-1-3. Standardize each operating parameter variable using the mean and standard deviation of each operating parameter variable to obtain n standardized operating parameter variables;

[0114] The expression of the standardized operating parameter variable is: ; represents the k-th standardized operating parameter variable of the i-th parameter sample;

[0115] S1-2-1-4. Combine the n standardized operating parameter variables into a standardized matrix ;

[0116] ;

[0117] represents the index of the operating parameter, represents the index of the parameter sample, m represents the total number of operating parameter variables, n represents the number of parameter samples, represents the k-th standardized operating parameter variable of the i-th parameter sample.

[0118] Specifically, constructing and standardizing an independent data set for each operating parameter effectively eliminates the dimensional differences between different variables and enables data analysis under the same standard.

[0119] S1-2-2 specifically includes: S1-2-2-1. Transpose the standardized matrix to obtain the transposed matrix ;

[0120] The expression of the transposed matrix is: ;

[0121] S1-2-2-2. Multiply the transposed matrix and the standardized matrix to obtain the product matrix ;

[0122] The product matrix The expression of ;

[0123] where the elements of the product matrix are characterized as the unnormalized values of the elements in the covariance matrix;

[0124] S1-2-2-3. Unbiasedly estimate the unnormalized values of the product matrix to obtain the covariance matrix ;

[0125] The covariance matrix has the following expression: ;

[0126] where the elements of the covariance matrix are defined as follows: ;

[0127] ;

[0128] represents the covariance between the k-th standardized operating parameter variable and the j-th standardized operating parameter variable.

[0129] Specifically, the calculation of the covariance matrix can reveal the correlation between parameter variables, so as to better extract features, achieve effective compression and dimensionality reduction of data, and improve the prediction accuracy of the model.

[0130] S1-2-3 specifically includes:

[0131] S1-2-3-1. Perform singular value decomposition on the covariance matrix to obtain the decomposed matrix;

[0132] The expression of the decomposed matrix is: ;

[0133] where represents the first orthogonal matrix, D represents the diagonal matrix, whose diagonal elements are singular values, is the second orthogonal matrix;

[0134] S1-2-3-2. Select the column vectors of the first orthogonal matrix as the eigenvectors of the covariance matrix, and the eigenvectors are used to represent the principal component vectors;

[0135] S1-2-3-3. Select the squares of the singular values of the diagonal matrix as the eigenvalues of the covariance matrix, and the eigenvalues are used to represent the variance magnitudes of the principal component vectors.

[0136] Specifically, perform singular value decomposition on the covariance matrix to obtain eigenvalues and corresponding eigenvectors, which provides a basis for subsequent principal component analysis. Singular value decomposition can extract the main features from high-dimensional data, simplify the data structure, and retain the core information of the data.

[0137] S1-2-4 specifically includes:

[0138] S1-2-4-1. Arrange the several eigenvalues obtained after singular value decomposition in descending order to obtain an ordered eigenvalue sequence;

[0139] S1-2-4-2. Calculate the cumulative variance explanation rate of the eigenvalues in descending order in the eigenvalue sequence until the cumulative variance explanation rate reaches a preset variance explanation rate;

[0140] S1-2-4-3. Select the eigenvectors corresponding to the eigenvalues that satisfy the cumulative variance explanation rate and use them as the principal component vectors.

[0141] Specifically, select the principal component vectors according to the variance explanation rate of the eigenvalues, so that the features after dimensionality reduction can fully express the main information of the original data.

[0142] S4 specifically includes:

[0143] S4-1. Receive the multi-parameter trend features in the model training samples as the input of the multi-task learning model, and output a configuration parameter prediction set through forward propagation;

[0144] S4-2. Use the multi-task loss function to calculate the loss between the configuration parameter prediction set and the historical configuration parameter set;

[0145] S4-3. Perform backpropagation in the multi-task learning model to iteratively obtain the model update parameters;

[0146] S4-4. Use the model update parameters for the next forward propagation, output the next configuration parameter prediction set, and repeat the forward propagation and backpropagation until the convergence condition is met.

[0147] S4-5. Export the multi-task learning model that meets the convergence condition as a parameter configuration model.

[0148] Specifically, through the forward propagation and backpropagation of the multi-task learning model, efficient training of the model and accurate prediction of dynamic configuration parameters are achieved. Among them, the convergence condition can be reaching a preset number of iterations, or the multi-task loss reaching a preset loss threshold, etc.

[0149] In this embodiment, the multi-task loss function is: ;

[0150] Among them, represents the multi-task loss, where n is the number of model training samples, is the set of historical configuration parameters in the i-th model training sample, the predicted set of configuration parameters for the i-th model training sample; C represents the total number of categories of the set of historical configuration parameters, represents the one-hot encoding of the c-th category in the i-th model training sample, represents the predicted probability of the c-th category in the i-th model training sample; represents the weight of the mean squared error loss, is the weight of the cross-entropy loss.

[0151] Specifically, by adopting a multi-task loss function and integrating different types of losses, the prediction accuracy of the model on the mixed configuration parameters of continuous values and discrete categories is improved. This solution further enhances the generalization ability of the model and its performance in different tasks through the weight adjustment of different loss functions, making the dynamic configuration more accurate and reliable.

[0152] Example 3: Refer to Figure 3 , the technical solution of this Example 3 is different from that of Example 1 and Example 2 in that it discloses a dynamic configuration switching system in the form of serialization annotations for implementing a dynamic configuration switching control method in the form of serialization annotations; the system includes:

[0153] A real-time operating state acquisition module for acquiring the real-time operating state of industrial equipment;

[0154] A configuration parameter set output module for inputting the real-time operating state into a preset parameter configuration model and outputting a dynamic configuration parameter set;

[0155] A configuration parameter loading module for loading a number of configuration parameters according to the dynamic configuration parameter set; wherein, the dynamic configuration parameter set is a number of configuration parameters output by the parameter configuration model and represented in the format of serialization annotations.

[0156] In summary, the present invention, through a multi-task learning model, automatically outputs a dynamic configuration parameter set according to the real-time operating state of industrial equipment and represents it in the format of serialization annotations, realizing the real-time dynamic configuration switching of the equipment. In particular, the present invention simplifies the input of multi-parameters through principal component analysis while retaining the most core data structure, improving the training efficiency and prediction accuracy of the model. It solves the problems in the prior art such as complex multi-parameter input, insufficient generalization ability of single-task models, and limited ability of real-time monitoring systems to handle complex relationships, and improves the dynamic adjustment function of configuration parameters.

[0157] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present application are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by wire (such as infrared, wireless, microwave, etc.).

[0158] The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or a data center that contains a collection of one or more available media. The available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVD ), or semiconductor media. The semiconductor media can be a solid-state drive.

[0159] In several embodiments provided in the present application, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division of an underwater terrain change analysis system and method for waterways. In actual implementation, there can be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces, and the indirect couplings or communication connections of devices or units can be in electrical, mechanical, or other forms.

[0160] The above is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed in the present application can easily think of changes or substitutions, which should all be covered within the protection scope of the present application.

Claims

1. A dynamic configuration switching control method in the form of serialized annotations, used in an industrial equipment server, characterized in that: The method comprises: Obtaining the real-time operating status of the industrial equipment; Inputting the real-time operating status into a preset parameter configuration model and outputting a dynamic configuration parameter set; According to the dynamic configuration parameter set, a plurality of configuration parameters are loaded; wherein the dynamic configuration parameter set is output by a parameter configuration model and is a plurality of configuration parameters represented in a serialized annotation format; The modeling method of the parameter configuration model includes: Inputting the model training samples of the parameter training set into the multi-task learning model to train the parameter configuration model; The steps of training the parameter configuration model include: Receive multi-parameter trend features in model training samples as input to the multi-task learning model, and output a configuration parameter prediction set through forward propagation; Use a multi-task loss function to calculate the loss between the configuration parameter prediction set and the historical configuration parameter set; In the multi-task learning model, back propagation is performed and the model update parameters are iteratively obtained; Use the model to update the parameters for the next forward propagation, output the next configuration parameter prediction set, and repeat the forward propagation and backward propagation until the convergence conditions are met; Export the multi-task learning model that meets the convergence conditions as a parameter configuration model; Accurate prediction of dynamic configuration parameters is achieved through forward propagation and back propagation of multi-task learning models; The multi-task loss function is: Among them, MTL represents multi-task loss, n is the number of model training samples, is the historical configuration parameter set in the i-th model training sample, The configuration parameter prediction set of the i-th model training sample; C represents the total number of categories of the historical configuration parameter set, represents the one-hot encoding of the cth category in the i-th model training sample, represents the predicted probability of the cth category in the i-th model training sample; represents the weight of the mean squared error loss, is the weight of the cross entropy loss.

2. According to a method for dynamic configuration switching control in the form of serialized annotations as described in claim 1, it is characterized in that: Before inputting the model training samples of the parameter training set into the multi-task learning model to train the parameter configuration model, the method further includes: S1. Obtaining multi-parameter trend characteristics of n industrial equipment in the same time period and their corresponding historical configuration parameter sets; S2, using the multi-parameter trend feature as an input feature and the corresponding historical configuration parameter set as a target label to construct a model training sample; S3. Gather n model training samples to obtain a parameter training set.

3. According to a method for dynamic configuration switching control in the form of serialized annotations as described in claim 1, it is characterized in that: Obtain multi-parameter trend characteristics of n industrial equipment in the same time period, including: S1-1, collecting n operating parameter variables of the industrial equipment within a reference time length, the n operating parameter variables including: temperature change rate, current fluctuation amplitude and other selected operating parameter variables; S1-2. Perform principal component vector analysis on the n operating parameter variables to obtain the multi-parameter trend characteristics.

4. According to a method for dynamic configuration switching control in the form of serialized annotations as described in claim 3, it is characterized in that: The collecting of n operating parameter variables of the industrial equipment within a reference time length includes: S1-1-1. Define a reference time length, where the reference time length includes a start time point and an end time point; S1-1-2. Collect the device temperature, current and other selected operating parameters at the start time point and the end time point respectively to calculate the operating parameter variables within the reference time length.

5. According to a method for dynamic configuration switching control in the form of serialized annotations as described in claim 3, it is characterized in that: The performing principal component vector analysis on the n operating parameter variables comprises: S1-2-1, standardize n operating parameter variables to obtain standardized parameters, and combine them into a standardized matrix; S1-2-2. Calculate the covariance matrix based on the standardized matrix; S1-2-3, perform singular value decomposition on the auxiliary defense difference moment to obtain the eigenvalue and its corresponding eigenvector; S1-2-4, arranging the eigenvalues ​​in descending order, determining the principal component vector according to a preset variance explanation rate; and constructing a principal component vector matrix based on the principal component vector; S1-2-5. Project the standardized matrix onto the principal component vector matrix to obtain a reduced-dimensional matrix, wherein the elements of the reduced-dimensional matrix are the multi-parameter trend features.

6. A method for dynamic configuration switching control in the form of serialized annotations according to claim 5, characterized in that: The step of standardizing the n operating parameter variables includes: S1-2-1-1, collect n operating parameter variables as parameter samples, and construct an independent data set for each operating parameter; Wherein, the expression of the independent data set of each operating parameter is: , represents the kth running parameter variable of the i-th parameter sample; S1-2-1-2. Calculate the mean and standard deviation of each independent data set of operating parameters to obtain the mean and standard deviation of the operating parameter variables; The mean expression of the operating parameter variable is: represents the mean value of the kth running parameter variable, represents the kth running parameter variable of the i-th parameter sample; The standard deviation expression of the operating parameter variable is: represents the standard deviation of the kth running parameter variable; S1-2-1-3. Standardize each operating parameter variable using its mean and standard deviation to obtain n standardized operating parameter variables; The expression of the standardized operating parameter variable is: represents the kth standardized operating parameter variable of the i-th parameter sample; S1-2-1-4, combining n standardized operating parameter variables into a standardized matrix Indicates the index of the running parameter. represents the index of the parameter sample, m represents the total number of running parameter variables, and n represents the number of parameter samples. represents the kth standardized running parameter variable of the ith parameter sample.

7. A method for dynamic configuration switching control in the form of serialized annotations according to claim 5, characterized in that: The step of calculating the covariance matrix according to the standardized matrix includes: S1-2-2-1, the standardized matrix Transpose to get the transposed matrix ; The transposed matrix The expression is: S1-2-2-2, transpose the matrix and the standardized matrix Multiply them to get the product matrix ; The product matrix The expression is: Wherein, the elements of the product matrix are represented as unnormalized values ​​of the elements in the covariance matrix; S1-2-2-3. Product Matrix The unnormalized value of is unbiasedly estimated to obtain the covariance matrix ; The covariance matrix The expression is: Among them, the covariance matrix The elements of are defined as follows: represents the covariance between the kth standardized operating parameter variable and the jth standardized operating parameter variable.

8. A method for dynamic configuration switching control in the form of serialized annotations according to claim 5, characterized in that: The performing singular value decomposition on the auxiliary defense difference moment includes: S1-2-3-1. Perform singular value decomposition on the covariance matrix to obtain a decomposed matrix; The expression of the decomposed matrix is: Where U represents the first orthogonal matrix, D represents a diagonal matrix whose diagonal elements are singular values, is the second orthogonal matrix; S1-2-3-2, selecting a column vector of the first orthogonal matrix as an eigenvector of the covariance matrix, wherein the eigenvector is used to characterize the principal component vector; S1-2-3-3. Select the square of the singular value of the diagonal matrix as the eigenvalue of the covariance matrix, and the eigenvalue is used to characterize the variance of the principal component vector.

9. A method for controlling dynamic configuration switching in the form of serialized annotations according to claim 5, characterized in that: The step of arranging the eigenvalues ​​in descending order and determining the principal component vector according to a preset variance explanation rate comprises: S1-2-4-1. Arrange the eigenvalues ​​obtained after singular value decomposition in descending order to obtain a sorted eigenvalue sequence; S1-2-4-2, calculating the cumulative variance explanation rate of the eigenvalues ​​in descending order in the eigenvalue sequence until the cumulative variance explanation rate reaches the preset variance explanation rate; S1-2-4-3. Select the eigenvector corresponding to the eigenvalue that satisfies the cumulative variance explanation rate and use it as the principal component vector.

10. A dynamic configuration switching system in the form of serialized annotations, characterized in that: The system is a dynamic configuration switching control method based on the serialized annotation form of any one of claims 1 to 9, and the system includes: A real-time operation status acquisition module, used to acquire the real-time operation status of the industrial equipment; A configuration parameter set output module, used for inputting the real-time operation status into a preset parameter configuration model and outputting a dynamic configuration parameter set; A configuration parameter loading module, used to load a number of configuration parameters according to the dynamic configuration parameter set; wherein the dynamic configuration parameter set is a number of configuration parameters output by a parameter configuration model and represented in a serialized annotation format; The modeling method of the parameter configuration model includes: Inputting the model training samples of the parameter training set into the multi-task learning model to train the parameter configuration model; The steps of training the parameter configuration model include: Receive multi-parameter trend features in model training samples as input to the multi-task learning model, and output a configuration parameter prediction set through forward propagation; Use a multi-task loss function to calculate the loss between the configuration parameter prediction set and the historical configuration parameter set; In the multi-task learning model, back propagation is performed and the model update parameters are iteratively obtained; Use the model to update the parameters for the next forward propagation, output the next configuration parameter prediction set, and repeat the forward propagation and backward propagation until the convergence conditions are met; Export the multi-task learning model that meets the convergence conditions as a parameter configuration model; Accurate prediction of dynamic configuration parameters is achieved through forward propagation and back propagation of multi-task learning models; The multi-task loss function is: Among them, MTL represents multi-task loss, n is the number of model training samples, is the historical configuration parameter set in the i-th model training sample, The configuration parameter prediction set of the i-th model training sample; C represents the total number of categories of the historical configuration parameter set, represents the one-hot encoding of the cth category in the i-th model training sample, represents the predicted probability of the cth category in the i-th model training sample; represents the weight of the mean squared error loss, is the weight of the cross entropy loss.

Images