Measurement data non-intrusive identification method based on hyper-parameter optimization algorithm

Abstract

The embodiment of the invention provides a measurement data non-intrusive identification method based on a hyper-parameter optimization algorithm. The method comprises the following steps: constructing an original feature set; the feature importance is calculated by using the predicted value variation, the features are sorted, forward feature selection is performed, and the selected features are added to a feature subset; when all the features are added to the feature subset, outputting an optimal feature subset by using a classifier; constructing a CatBoost model, initializing parameters, and determining a search range; the optimal feature subset is used as input, a BOHB hyper-parameter optimization algorithm is used for searching parameters of the CatBoost model in a search range, the optimal hyper-parameter of the model is determined, and a BOHB-CatBoost model is obtained; and inputting data into the model to obtain an identification result. In this way, the optimal value of the hyper-parameter of the CatBoost can be accurately determined, and the model can achieve better performance in an industrial load identification task, so that the identification precision of the industrial load state is improved, and the limitation that the optimal hyper-parameter is difficult to find through a traditional optimization method is overcome.

Description

Technical Field

[0001] The present invention generally relates to the field of power systems, and more particularly, to a non-intrusive identification method of measurement data based on a hyperparameter optimization algorithm. Background Art

[0002] At present, Bayesian hyperband hyperparameter optimization has been widely used in the data reclassification process. For example, in the face of complex and changeable operating conditions of power systems (such as load changes, faults, etc.), adaptive control strategies are crucial. Bayesian optimization can be used to optimize the parameters in adaptive control algorithms, such as the action threshold and time delay parameters in adaptive relay protection devices. By analyzing historical fault data and normal system operation data, the optimal combination of adaptive control parameters under different conditions is found to ensure the safe and reliable operation of the power system.

[0003] In the existing industrial load identification technology, due to the lack of correlation analysis between the industrial load's own characteristics and environmental factors and data imbalance and fluctuation, the feature extraction method has problems with calculation, parameters, high-frequency information and complexity; the CatBoost model hyperparameter optimization efficiency is low. Summary of the invention

[0004] According to an embodiment of the present invention, a non-intrusive identification scheme for measurement data based on a hyperparameter optimization algorithm is provided. This scheme accurately determines the optimal values ​​of the hyperparameters of the CatBoost model, enhances the performance of the model in industrial load identification, enables the model to better cope with data complexity and imbalance, and provides reliable support for industrial load management.

[0005] In a first aspect of the present invention, a non-intrusive identification method for measurement data based on a hyperparameter optimization algorithm is provided. The method comprises:

[0006] Obtain the original samples, perform time domain segmentation on the original samples, and construct the original feature set;

[0007] The feature importance is calculated using the predicted value change, the features in the original feature set are sorted according to the feature importance, forward feature selection is performed, and the selected features are added to the feature subset; when all the features in the original feature set are added to the feature subset, the optimal feature subset is output using the classifier;

[0008] Build a CatBoost model, initialize the parameters of the CatBoost model, and determine the search range;

[0009] Taking the optimal feature subset as input, using the BOHB hyperparameter optimization algorithm, collecting parameters of the CatBoost model within the search range through the collection function of the expected improvement standard, determining the optimal hyperparameters of the CatBoost model, and obtaining the BOHB-CatBoost model;

[0010] The industrial conforming data to be identified is input into the BOHB-CatBoost model to obtain the identification results.

[0011] Furthermore, the time domain segmentation of the original samples to construct the original feature set includes:

[0012] The original samples are divided into three time domain intervals, each of which includes a number of sampling points;

[0013] Several entropy features are extracted for each time domain interval to obtain the original feature set.

[0014] Furthermore, the entropy feature includes:

[0015] Power spectrum entropy, calculated as:

[0016]

[0017] Fuzzy entropy, the calculation formula is:

[0018] F fe =lnΨ m (r)-lnΨ m+1 (r)

[0019] Singular spectral entropy, calculated as:

[0020]

[0021] Permutation entropy, calculated as:

[0022]

[0023] Energy entropy, calculated as:

[0024]

[0025] Sample entropy, calculated as:

[0026]

[0027] Approximate entropy, calculated as:

[0028] F ae =Ψ m (r)-Ψ m+1 (r)

[0029] Envelope entropy, calculated as:

[0030]

[0031] Among them, F pse is the power spectrum entropy; N is the total number of samples; n is the sample index value; P n is the probability density value of the nth sampling point; log2P n To measure the uncertainty of information or the degree of confusion of information content; F fe is the fuzzy entropy; m (r) is the number of similar point pairs in the time series under a given embedding dimension m and threshold r; Ψ m+1 (r) is the number of similar point pairs in the time series under a given embedding dimension m+1 and threshold r; F sse is the singular spectral entropy; m is the embedding dimension; F pe is the permutation entropy; m! is the factorial of the embedding dimension; F ee is the energy entropy; x(n) is the amplitude of the nth sampling point; F se is the sample entropy; F ae is the approximate entropy; F en is the envelope entropy.

[0032] Furthermore, the calculation of feature importance using the predicted value change includes:

[0033] F={f1,f2,f2,......f n}

[0034] P i =β i F j

[0035] Among them, F is the input feature set; f n is an element in the input feature set; P i is the feature importance; F j is a specific feature selected from the input feature set; β i is a numerical factor.

[0036] Further, the method of using a classifier to output an optimal feature subset includes:

[0037] If features are added to the feature subset, the classifier is used to calculate the classification accuracy of the feature subset after the features are added, and the feature subset with the highest classification accuracy is taken as the optimal feature subset.

[0038] Furthermore, the parameters of the CatBoost model are collected through the collection function that is expected to improve the standard, and the optimal hyperparameters of the CatBoost model are determined, including:

[0039] For the optimal feature subset, establishing a proxy model;

[0040] In the proxy model, objective function values ​​are selected in sequence through Gaussian decomposition to update hyperparameters of the proxy model;

[0041] The optimal feature subset is traversed to obtain the optimal hyperparameter after the expected improvement.

[0042] In a second aspect of the present invention, a non-intrusive identification device for measurement data based on a hyperparameter optimization algorithm is provided. The device comprises:

[0043] The acquisition module is used to obtain the original samples, perform time domain segmentation on the original samples, and construct the original feature set;

[0044] A feature selection module is used to calculate feature importance using the predicted value change, sort the features in the original feature set according to the feature importance, perform forward feature selection, and add the selected features to the feature subset; when all features in the original feature set are added to the feature subset, use the classifier to output the optimal feature subset;

[0045] CatBoost building module, used to build CatBoost model, initialize the parameters of CatBoost model, and determine the search range;

[0046] A CatBoost optimization module, which is used to take the optimal feature subset as input, use the BOHB hyperparameter optimization algorithm to collect parameters of the CatBoost model within the search range through a collection function of an expected improvement standard, determine the optimal hyperparameters of the CatBoost model, and obtain a BOHB-CatBoost model;

[0047] The recognition module is used to input the industrial conformity data to be recognized into the BOHB-CatBoost model to obtain the recognition result.

[0048] In a third aspect of the present invention, an electronic device is provided. The electronic device has at least one processor; and a memory connected to the at least one processor in communication; the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can perform the method of the first aspect of the present invention.

[0049] In a fourth aspect of the present invention, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to cause the computer to execute the method of the first aspect of the present invention.

[0050] It should be understood that the contents described in the summary of the invention are not intended to limit the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] The above and other features, advantages and aspects of the embodiments of the present invention will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. In the accompanying drawings, the same or similar reference numerals represent the same or similar elements, wherein:

[0052] Figure 1 A flow chart of a non-intrusive identification method of measurement data based on a hyperparameter optimization algorithm according to an embodiment of the present invention is shown;

[0053] Figure 2 A block diagram of a non-intrusive identification device for measurement data based on a hyperparameter optimization algorithm according to an embodiment of the present invention is shown;

[0054] Figure 3 A block diagram of an exemplary electronic device capable of implementing embodiments of the present invention is shown;

[0055] Among them, 300 is an electronic device, 301 is a computing unit, 302 is a ROM, 303 is a RAM, 304 is a bus, 305 is an I / O interface, 306 is an input unit, 307 is an output unit, 308 is a storage unit, and 309 is a communication unit. DETAILED DESCRIPTION

[0056] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0057] In addition, the term "and / or" in this article is only a description of the association relationship between the associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0058] In the present invention, a probabilistic proxy model and an acquisition function are used to more efficiently search the hyperparameter space and quickly find a better hyperparameter combination; thereby, the optimal value of the hyperparameters of the classification boosting tree model (CatBoost) can be determined more accurately, so that the model can achieve better performance in the industrial load identification task, thereby improving the identification accuracy of the industrial load state and overcoming the limitation of traditional optimization methods that it is difficult to find the optimal hyperparameters.

[0059] Figure 1 A flow chart of a non-intrusive identification method of measurement data based on a hyperparameter optimization algorithm according to an embodiment of the present invention is shown.

[0060] Specifically, non-intrusive identification is a technical method to identify and judge the relevant characteristics, parameters or states of the target system without interfering with its normal operation or structural integrity. In the engineering field, for example, the analysis of load characteristics in the power system does not require the installation of additional sensors inside the load. It only collects measurable data such as voltage and current at the system port, and uses advanced signal processing and data analysis algorithms such as wavelet analysis and principal component analysis to extract feature information, thereby accurately identifying parameters such as load type and power. This method is widely used in various fields such as electrical and mechanical, which can reduce costs and impact on equipment.

[0061] The method includes:

[0062] S101, obtaining original samples, performing time domain segmentation on the original samples, and constructing an original feature set.

[0063] Specifically, the original samples are divided into three time domain intervals, each of which includes a number of sampling points; a number of entropy features are extracted from each time domain interval to obtain an original feature set.

[0064] By extracting features from different time segments, the feature expression capability is enhanced, which can more comprehensively describe data changes; at the same time, it can also improve the adaptability of the model and capture various trends in time series; in addition, it can reduce data complexity and decompose complex time domain data into simple sub-problems.

[0065] Among them, dividing into three time domain intervals means: first, each sample is divided into three intervals with 20 sampling points in the front and back (40 sampling points as a whole), then, 27 time domain features containing 8 entropy features are extracted from different intervals, and finally, an original feature set is constructed from the original data according to the 8 entropies.

[0066] In this embodiment, the entropy features include:

[0067] Power spectrum entropy, calculated as:

[0068]

[0069] Fuzzy entropy, the calculation formula is:

[0070] F fe =lnΨ m (r)-lnΨ m+1 (r)

[0071] Singular spectral entropy, calculated as:

[0072]

[0073] Permutation entropy, calculated as:

[0074]

[0075] Energy entropy, calculated as:

[0076]

[0077] Sample entropy, calculated as:

[0078]

[0079] Approximate entropy, calculated as:

[0080] F ae =Ψ m (r)-Ψ m+1 (r)

[0081] Envelope entropy, calculated as:

[0082]

[0083] Among them, F pse is the power spectrum entropy; N is the total number of samples; n is the sample index value; P n is the probability density value of the nth sampling point; log2P n To measure the uncertainty of information or the degree of confusion of information content; F fe is the fuzzy entropy; m (r) is the number of similar point pairs in the time series under a given embedding dimension m and threshold r, where r is a threshold used to determine whether two points are similar; Ψ m+1 (r) is the number of similar point pairs in the time series under a given embedding dimension m+1 and threshold r; F sse is the singular spectral entropy; m is the embedding dimension; F pe is the permutation entropy; m! is the factorial of the embedding dimension; F ee is the energy entropy; x(n) is the amplitude of the nth sampling point; F se is the sample entropy; F ae is the approximate entropy; F en is the envelope entropy.

[0084] The original feature set constructed by the above entropy enriches the feature dimension and can more comprehensively characterize data information. In addition, different entropies capture data characteristics from multiple angles, which can enhance the model's ability to understand complex data and help improve the model's accuracy, stability, and generalization ability.

[0085] S102, using the predicted value change to calculate the feature importance, sorting the features in the original feature set according to the feature importance, performing forward feature selection, and adding the selected features to the feature subset; when all the features in the original feature set are added to the feature subset, using the classifier to output the optimal feature subset.

[0086] Specifically, sorting the features in the original feature set according to feature importance is to arrange the features in the feature set in descending order according to feature importance.

[0087] In this embodiment, the calculation of feature importance using the predicted value change includes:

[0088] F={f1,f2,f2,......f n}

[0089] P i =β i F j

[0090] P i+1 =β i+1 F j

[0091] P i＝0 ≠P i ≠P i+1

[0092] Among them, F is the input feature set; f n is an element in the input feature set; P i is the feature importance; F j is a specific feature selected from the input feature set; β i is a numerical factor; P i+1 is the predicted value when the numerical factor changes; β i+1 The modified numerical factor.

[0093] Specifically, for each feature added, the classifier is used to calculate the classification accuracy under the feature subset; this process is repeated, and the remaining features are gradually added until all features are added to the feature set.

[0094] Sorting features by calculating feature importance can quickly identify key features and concentrate resources for in-depth analysis. It can also remove unimportant features, reduce data dimensions and computational complexity. It helps to understand the intrinsic relationship of data, optimize model construction, and improve model training efficiency and prediction accuracy.

[0095] In this embodiment, the step of outputting the optimal feature subset using a classifier includes:

[0096] If features are added to the feature subset, the classifier is used to calculate the classification accuracy of the feature subset after the features are added, and the feature subset with the highest classification accuracy is taken as the optimal feature subset.

[0097] Specifically, each time a feature is added, the classification accuracy under the current feature subset is recorded and compared with the previous results. Then, the optimal feature subset is determined based on the highest recognition accuracy of the classifier.

[0098] Using the classifier to output the optimal feature subset reduces the data dimension, reduces computing costs and storage requirements; it can remove irrelevant and redundant features, increase model training speed, enhance model interpretability, reduce noise interference, and improve classification accuracy.

[0099] S103, constructing a CatBoost model, initializing parameters of the CatBoost model, and determining a search range.

[0100] Specifically, the CatBoost ensemble algorithm is a special type of decision tree gradient boosting because it can handle categorical, ordered features, and the overfitting of the model is handled by the Bayesian estimator. The search range is determined using the Bayesian Optimization (BO) algorithm. The core of the BO algorithm framework is the acquisition function and the probabilistic proxy model, which establishes the objective function through the probabilistic proxy model, and uses the maximized acquisition function to select the next evaluation point, and continuously updates the probabilistic proxy model, preliminarily screens hyperparameters, and combines the HyperBand algorithm to allocate budgets for multiple configurations and evaluate performance, eliminate poor configurations, and repeat this process to further optimize hyperparameters, and finally determine the optimal hyperparameters of the classification boosting tree model (CatBoost) model.

[0101] S104, taking the optimal feature subset as input, using the BOHB (Bayesian optimizationHyperband) hyperparameter optimization algorithm, collecting the parameters of the CatBoost model within the search range through the collection function of the expected improvement standard, determining the optimal hyperparameters of the CatBoost model, and obtaining the BOHB-CatBoost model.

[0102] Among them, the BOHB hyperparameter optimization algorithm is a combination of the Bayesian optimization algorithm (BO) and the Hyperband (HB) algorithm. The Bayesian Hyperband, which combines the two, can efficiently search in the hyperparameter space, find the best hyperparameter combination with limited resources, and accelerate the model training and optimization process. Bayesian optimization uses probabilistic models (such as Gaussian processes) to estimate the relationship between hyperparameters and objective functions, and guides the search direction through acquisition functions. Hyperband quickly screens hyperparameter combinations through resource allocation and early stopping mechanisms.

[0103] Specifically, the detailed process of constructing the BOHB-CatBoost classifier is as follows:

[0104] (1) Initialization is required to clarify the specific optimization direction, upper and lower limits of the value, parameter type, and maximum number of iterations.

[0105] (2) Start the loop: At each stage of the “successive halving” algorithm, the top 1 / eta parameters with the best performance will be selected and run with more computing resources.

[0106] (3) Repeat the above process for the loop and exit the loop when the number of iterations reaches the maximum.

[0107] (4) Determine the optimal values ​​of the model’s hyperparameters and output the constructed BOHB-CatBoost classifier.

[0108] In this embodiment, the parameters of the CatBoost model are collected by collecting functions that are expected to improve the standard, and the optimal hyperparameters of the CatBoost model are determined, including:

[0109] For the optimal feature subset, establishing a proxy model;

[0110] In the proxy model, objective function values ​​are selected in sequence through Gaussian decomposition to update hyperparameters of the proxy model;

[0111] The optimal feature subset is traversed to obtain the optimal hyperparameter after the expected improvement.

[0112] The above method can efficiently explore the parameter space, quickly locate high-quality parameter combinations, reduce search time and computing resource consumption; at the same time, it can improve model performance and enhance generalization capabilities.

[0113] As a specific embodiment of the above method, the acquisition function of the desired improvement standard includes:

[0114]

[0115] Among them, EI is the acquisition function; is the optimal hyperparameter value, that is, the optimal hyperparameter of the CatBoost model; y * is the threshold; f(x'|δ) is the random forest model at x', where x' is the current hyperparameter, δ is the variable of the acquisition function EI; d is the differential operator; p(f|D) is the predicted distribution.

[0116]

[0117] Where l(m) is the probability that the random variable y is less than the threshold y under given condition m and data set D. * g(m) is the probability that the random variable y is greater than or equal to the threshold y given the condition m and the data set D. * probability.

[0118] y * =min{y'0,y'1,y'2,...,y' n ,}

[0119] Among them, the random variable y={y0,y1,y2,...,y n};y n is any random variable in the threshold.

[0120] D={(x0,y0),(x1,y1),...,(x n-1 ,y n-1 )}

[0121] Where D is the data set; (x n-1 ,y n-1 ) is the random variable factor in the data set.

[0122] In this embodiment, BO uses the Tree Parzen Estimator (TPE) method. TPE models the density of its input configuration using a kernel density estimator. The ratio of needs to be optimized for the acquisition function EI. The specific calculation formula is as follows:

[0123] l(m)=p(y <y * |m,D)

[0124] g(m)=p(y≥y * |m,D)

[0125] Using the expected improvement standard acquisition function to obtain the best hyperparameters can improve the efficiency of model optimization, accurately explore the hyperparameter space, avoid local optimality, and speed up the search for the optimal configuration; at the same time, it can save computing resources and time costs, reduce the number of experiments, and efficiently use existing results. In addition, it can also improve the interpretability and stability of the model. Its selection process is interpretable, and the performance improvement trend is stable, which makes it easier to grasp the relationship between model hyperparameters and performance and enhance the ability to control model effects.

[0126] S105, inputting the industrial conformity data to be identified into the BOHB-CatBoost model to obtain the identification result.

[0127] According to the embodiments of the present invention, the following beneficial effects are achieved:

[0128] 1. The BO algorithm is adopted to improve the acquisition function and synthesize sample data by calling the optimal hyperparameter value of EI, which can balance the data and solve the data imbalance problem caused by the unequal cumulative data volume of samples of different categories in industrial load data. The classification performance of the model on the unbalanced data set is improved. Compared with the original data, the accuracy, precision and recall rate are significantly improved, which effectively addresses the problem of difficulty in mining internal rules and low accuracy of industrial equipment identification caused by the imbalance of positive and negative sample data.

[0129] 2. The parameters of the CatBoost model are collected using the Bayesian hyperband hyperparameter optimization algorithm, the optimal values ​​of the hyperparameters are determined, and the BOHB-CatBoost classifier is constructed, which effectively improves the efficiency of hyperparameter optimization. The CatBoost model optimized by the Bayesian hyperband is superior to other methods in terms of accuracy and stability, significantly improving the accuracy of industrial load identification, and has good effects on industrial load identification in small sample scenarios and unbalanced scenarios.

[0130] It should be noted that, for the above-mentioned method embodiments, for the sake of simplicity, they are all described as a series of action combinations, but those skilled in the art should know that the present invention is not limited by the described order of actions, because according to the present invention, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily required by the present invention.

[0131] The above is an introduction to a method embodiment. The following is a further explanation of the scheme of the present invention through a device embodiment having the same inventive concept as the method in the aforementioned embodiment.

[0132] like Figure 2 As shown, the device 200 includes:

[0133] An acquisition module 210 is used to acquire original samples, perform time domain segmentation on the original samples, and construct an original feature set;

[0134] The feature selection module 220 is used to calculate the feature importance by using the predicted value change, sort the features in the original feature set according to the feature importance, perform forward feature selection, and add the selected features to the feature subset; when all the features in the original feature set are added to the feature subset, output the optimal feature subset by using the classifier;

[0135] CatBoost construction module 230, used to construct a CatBoost model, initialize parameters of the CatBoost model, and determine a search range;

[0136] A CatBoost optimization module 240, configured to use the optimal feature subset as input, collect parameters of the CatBoost model within the search range by using a collection function of a desired improvement standard using a BOHB hyperparameter optimization algorithm, determine the optimal hyperparameters of the CatBoost model, and obtain a BOHB-CatBoost model;

[0137] The identification module 250 is used to input the industrial conformity data to be identified into the BOHB-CatBoost model to obtain the identification result.

[0138] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the described module can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0139] In the technical solution of the present invention, the acquisition, storage and application of user personal information involved are in compliance with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0140] According to an embodiment of the present invention, the present invention further provides an electronic device and a readable storage medium.

[0141] Figure 3 A schematic block diagram of an electronic device 300 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0142] The electronic device 300 includes a computing unit 301, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 302 or a computer program loaded from a storage unit 308 into a random access memory (RAM) 303. In the RAM 303, various programs and data required for the operation of the electronic device 300 can also be stored. The computing unit 301, the ROM 302, and the RAM 303 are connected to each other via a bus 304. An input / output (I / O) interface 305 is also connected to the bus 304.

[0143] A number of components in the electronic device 300 are connected to the I / O interface 305, including: an input unit 306, such as a keyboard, a mouse, etc.; an output unit 307, such as various types of displays, speakers, etc.; a storage unit 308, such as a disk, an optical disk, etc.; and a communication unit 309, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 309 allows the electronic device 300 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0144] The computing unit 301 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 301 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 301 performs the various methods and processes described above, such as methods S101 to S105. For example, in some embodiments, methods S101 to S105 may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit 308. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 300 via the ROM 302 and / or the communication unit 309. When the computer program is loaded into the RAM 303 and executed by the computing unit 301, one or more steps of the methods S101 to S105 described above may be performed. Alternatively, in other embodiments, the computing unit 301 may be configured to execute methods S101 to S105 in any other appropriate manner (eg, by means of firmware).

[0145] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0146] The program code for implementing the method of the present invention can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code can be executed entirely on the machine, partially on the machine, partially on the machine as a stand-alone software package and partially on a remote machine, or entirely on a remote machine or server.

[0147] In the context of the present invention, a machine-readable medium may be a tangible medium that may contain or store a program for use by an instruction execution system, device or equipment or used in combination with an instruction execution system, device or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0148] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0149] The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0150] A computer system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The relationship of client and server is generated by computer programs running on respective computers and having a client-server relationship with each other. The server may be a cloud server, a server of a distributed system, or a server combined with a blockchain.

[0151] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0152] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A non-intrusive identification method for measurement data based on a hyperparameter optimization algorithm, characterized in that: include: Obtain the original samples, perform time domain segmentation on the original samples, and construct the original feature set; The feature importance is calculated using the predicted value change, the features in the original feature set are sorted according to the feature importance, forward feature selection is performed, and the selected features are added to the feature subset; when all the features in the original feature set are added to the feature subset, the optimal feature subset is output using the classifier; Build a CatBoost model, initialize the parameters of the CatBoost model, and determine the search range; Taking the optimal feature subset as input, using the BOHB hyperparameter optimization algorithm, collecting parameters of the CatBoost model within the search range through the collection function of the expected improvement standard, determining the optimal hyperparameters of the CatBoost model, and obtaining the BOHB-CatBoost model; The industrial conforming data to be identified is input into the BOHB-CatBoost model to obtain the identification results.

2. The method according to claim 1, characterized in that The time domain segmentation of the original sample to construct the original feature set includes: The original samples are divided into three time domain intervals, each of which includes a number of sampling points; Several entropy features are extracted for each time domain interval to obtain the original feature set.

3. The method according to claim 2, characterized in that The entropy features include: Power spectrum entropy, calculated as: Fuzzy entropy, the calculation formula is: F fe =lnΨ m (r)-lnΨ m+1 (r) Singular spectral entropy, calculated as: Permutation entropy, calculated as: Energy entropy, calculated as: Sample entropy, calculated as: Approximate entropy, calculated as: F ae =Ψ m (r)-Ψ m+1 (r) Envelope entropy, calculated as: Among them, F pse is the power spectrum entropy; N is the total number of samples; n is the sample index value; P n is the probability density value of the nth sampling point; log2P n To measure the uncertainty of information or the degree of confusion of information content; F fe is the fuzzy entropy; m (r) is the number of similar point pairs in the time series under a given embedding dimension m and threshold r; Ψ m+1 (r) is the number of similar point pairs in the time series under a given embedding dimension m+1 and threshold r; F sse is the singular spectral entropy; m is the embedding dimension; F pe is the permutation entropy; m! is the factorial of the embedding dimension; F ee is the energy entropy; x(n) is the amplitude of the nth sampling point; F se is the sample entropy; F ae is the approximate entropy; F en is the envelope entropy.

4. The method according to claim 1, characterized in that: The calculating of feature importance by using the predicted value change includes: F={f1,f2,f2,......f n } P i =b i F j Among them, F is the input feature set; f n is an element in the input feature set; P i is the feature importance; F j is a specific feature selected from the input feature set; β i is a numerical factor.

5. The method according to claim 1, characterized in that The method of using a classifier to output an optimal feature subset includes: If features are added to the feature subset, the classifier is used to calculate the classification accuracy of the feature subset after the features are added, and the feature subset with the highest classification accuracy is taken as the optimal feature subset.

6. The method according to claim 1, characterized in that The method collects the parameters of the CatBoost model by using the collection function of the expected improvement standard to determine the optimal hyperparameters of the CatBoost model, including: Establishing a proxy model for the optimal feature subset; In the proxy model, objective function values ​​are selected in sequence through Gaussian decomposition to update hyperparameters of the proxy model; The optimal feature subset is traversed to obtain the optimal hyperparameter after the expected improvement.

7. A non-intrusive identification device for measurement data based on a hyperparameter optimization algorithm, characterized in that: include: The acquisition module is used to obtain the original samples, perform time domain segmentation on the original samples, and construct the original feature set; A feature selection module is used to calculate feature importance using the predicted value change, sort the features in the original feature set according to the feature importance, perform forward feature selection, and add the selected features to the feature subset; when all features in the original feature set are added to the feature subset, use the classifier to output the optimal feature subset; CatBoost building module, used to build CatBoost model, initialize the parameters of CatBoost model, and determine the search range; A CatBoost optimization module, which is used to take the optimal feature subset as input, use the BOHB hyperparameter optimization algorithm to collect parameters of the CatBoost model within the search range through a collection function of an expected improvement standard, determine the optimal hyperparameters of the CatBoost model, and obtain a BOHB-CatBoost model; The recognition module is used to input the industrial conformity data to be recognized into the BOHB-CatBoost model to obtain the recognition result.

8. An electronic device comprising at least one processor; and a memory connected in communication with the at least one processor; characterized in that: The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 6.

9. A non-transitory computer-readable storage medium storing computer instructions, characterized in that: The computer instructions are used to cause the computer to execute the method according to any one of claims 1-6.

Images