A sensor life prediction method based on a fusion algorithm and related devices

Abstract

The application belongs to the technical field of sensor life prediction, and discloses a sensor life prediction method based on a fusion algorithm and related devices, which comprises the following steps: obtaining the sensitivity data of a sensor under working conditions, and obtaining the sensitivity reduction data caused by voltage, current and temperature respectively after preprocessing; constructing a model of sensor aging caused by voltage and current based on the sensitivity reduction data caused by voltage and current; obtaining a sensor life consumption model caused by temperature by combining algorithm fusion based on the sensitivity reduction data caused by temperature; merging the sensor life consumption model caused by temperature and the model of sensor aging caused by voltage and current to obtain a composite estimation model, and performing sensor life prediction. The application adopts extreme gradient boosting tree and long short-term memory algorithm for modeling respectively in view of the sensitivity reduction of the sensor caused by voltage, current and temperature. The best parameter combination is obtained through the grid search algorithm to ensure the high precision of the model.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of sensor life prediction, and particularly relates to a sensor life prediction method based on a fusion algorithm and a related device. BACKGROUND

[0002] The intelligent sensor life estimation technology refers to obtaining estimation basic data through an accelerated aging experiment on a target sensor, respectively establishing estimation models according to characteristics of experimental characteristic variables, and obtaining optimal prediction sensitivity decline capability of the sensor in a service life scenario after algorithm fusion of optimal estimation models in independent scenarios.

[0003] The sensor detection method in the traditional power system needs to consume a large amount of manpower and material resources and is low in efficiency, and it is difficult to find hidden faults of mechanisms.

[0004] The learning rate in deep learning is a crucial hyperparameter, which determines the speed and direction of updating the weight parameters of the model in the training process. It controls how much we adjust the weights of the network to meet the gradient loss. The lower the value, the slower the gradient descent. The current technology usually directly selects a small learning rate when using a neural network algorithm to ensure that no local minimum point is missed, but it will take a long time to converge. SUMMARY

[0005] The application aims to provide a sensor life prediction method based on a fusion algorithm and a related device to solve the problems of large consumption of manpower and material resources and long convergence time of neural network algorithms in the prior art.

[0006] To achieve the above-mentioned purpose, the application adopts the following technical solutions:

[0007] In a first aspect, the application provides a sensor life prediction method based on a fusion algorithm, comprising:

[0008] Obtaining sensitivity data of the sensor corresponding to working conditions to obtain sensitivity decline data caused by voltage, current and temperature respectively;

[0009] Based on the sensitivity decline data caused by voltage and current, a model of sensor aging caused by voltage and current is constructed; based on the sensitivity decline data caused by temperature, a sensor life consumption model caused by temperature is obtained by algorithm fusion;

[0010] The sensor life consumption model caused by temperature and the model of sensor aging caused by voltage and current are combined to obtain a composite estimation model, and sensor life prediction is performed according to the composite estimation model.

[0011] Further, the acquisition sensor corresponds to the sensitivity data under the working condition, and the voltage, current and temperature respectively cause the sensitivity reduction data, including:

[0012] The magnetoresistance curve of the sensor is obtained by linear fitting, and the abnormal data of the experimental error is removed;

[0013] The sensitivity data of the sensor under complex working conditions is operated by dimensionality reduction, all array dimensions are unified to three arrays, and the sensitivity reduction data caused by voltage, current and temperature is marked, respectively, to obtain the sensitivity reduction data caused by voltage, current and temperature;

[0014] The sensitivity reduction data caused by voltage, current and temperature is divided into test set and validation set, and the data set is divided by random seed.

[0015] Further, the sensitivity reduction data based on voltage and current is used to construct a model of sensor aging caused by voltage and current, including:

[0016] The experimental data of the sensor sensitivity reduction caused by voltage and current is obtained by grid search algorithm to obtain the best parameter combination for estimation, and the optimal hyperparameter combination is input into the extreme gradient boosting tree algorithm to construct the model of sensor aging caused by voltage and current.

[0017] Further, the sensitivity reduction data based on temperature is combined with algorithm fusion to obtain a sensor life consumption model caused by temperature, including:

[0018] The experimental data of the sensor sensitivity reduction caused by temperature is selected for estimation, and the long short-term memory algorithm is used for neural network processing, and the learning rate parameter of the neural network is used to construct the sensor life consumption model caused by temperature.

[0019] Further, the algorithm fusion is used to construct the sensor life consumption model caused by temperature, including:

[0020] First, select the learning rate, grid division number, initializer, activation function, iteration number, batch calculation number and loss function hyperparameter combination, and set the initial array for the hyperparameters;

[0021] The hyperparameter combination is sampled by random seed, all hyperparameters are used as the characteristic value of the reverse engineering model, the long short-term memory method is used for model construction, the root mean square error is used as the prediction value of the reverse engineering model, and the reverse engineering model is constructed;

[0022] For the large amount of data generated by reverse engineering model, the model data is introduced into the simulated annealing algorithm to obtain the super parameter of the probability optimal solution and the batch reverse optimal array with the minimum root mean square error;

[0023] The reverse optimal array is input into the extreme gradient boosting tree model, at this time, the grid division number, the initializer, the activation function, the iteration number, the batch calculation number, the loss function super parameter combination and the root mean square error are taken as characteristic values, the learning rate is taken as a prediction value to construct a model, and the learning rate and other super parameter combination values are obtained by the extreme gradient boosting tree model prediction;

[0024] The scene optimal super parameter combination is input into the long short term memory method, model training is performed, and a temperature-induced sensor life consumption model under algorithm fusion is obtained.

[0025] Further, the temperature-induced sensor life consumption model and the model of voltage and current-induced sensor aging are combined to obtain a composite prediction model, and sensor life prediction is performed according to the composite prediction model, including:

[0026] Sensor working condition data containing voltage, current and temperature information are input into the composite prediction model, the composite prediction model transmits the data to the model of voltage and current-induced sensor aging and the temperature-induced sensor life consumption model respectively, the two models are predicted respectively, and the sensor life prediction results based on respective working conditions are given; the composite prediction model receives the prediction results of the two models, integrates them by weighted average and linear combination, and outputs the comprehensive sensor life prediction result after integration.

[0027] In a second aspect, the present application provides a sensor life prediction system based on fusion algorithm, comprising:

[0028] A data acquisition module is configured to obtain sensitivity data of a sensor under working conditions, and to obtain voltage and current-induced sensitivity degradation data and temperature-induced sensitivity degradation data after preprocessing.

[0029] A model construction module is configured to construct a model of voltage and current-induced sensor aging based on voltage and current-induced sensitivity degradation data, and to obtain a temperature-induced sensor life consumption model by combining algorithm fusion based on temperature-induced sensitivity degradation data.

[0030] An estimation output module is configured to combine a temperature-induced sensor life consumption model and a model of voltage and current-induced sensor aging to obtain a composite prediction model, and to perform sensor life prediction.

[0031] Further, the data acquisition module is specifically configured to:

[0032] The magnetoresistance curve of the sensor is obtained through experiments, and the corresponding sensitivity data of the sensor under complex working conditions is obtained through linear fitting to remove abnormal data caused by experimental errors;

[0033] The corresponding sensitivity data of the sensor under complex working conditions is subjected to dimensionality reduction and dimensionality reduction operations, all array dimensions are unified into three-dimensional arrays, and the sensitivity reduction data caused by voltage, current and temperature are labeled respectively to obtain the sensitivity reduction data caused by voltage, current and temperature.

[0034] Further, the model construction module is specifically configured to:

[0035] For the experimental data of the sensor sensitivity reduction caused by voltage and current, the best parameter combination is obtained through the grid search algorithm for estimation, and the optimal hyperparameter combination is input into the extreme gradient boosting tree algorithm to construct a model of the sensor aging caused by voltage and current.

[0036] Further, the model construction module is specifically configured to:

[0037] For the experimental data of the sensor sensitivity reduction caused by temperature, the long short-term memory algorithm is selected for estimation, and when the neural network is processed, the learning rate parameter of the neural network is fused to construct a sensor life consumption model caused by temperature.

[0038] Further, the model construction module is specifically configured to:

[0039] Select the hyperparameter combination of learning rate, grid division number, initializer, activation function, iteration number, batch calculation number and loss function, and set the initial array for the hyperparameters;

[0040] The hyperparameter combination is traversed by random seed sampling, all hyperparameters are taken as the eigenvalues of the reverse engineering model, and the long short-term memory method is used for model construction; the root mean square error is taken as the predicted value of the reverse engineering model, and the reverse engineering model is constructed;

[0041] For a large amount of data generated by the reverse engineering model, the model data is introduced into the simulated annealing algorithm to obtain the hyperparameters of the probability optimal solution and the batch reverse optimal array with the minimum root mean square error;

[0042] The reverse optimal array is input into the extreme gradient boosting tree model, at this time the hyperparameter combination of grid division number, initializer, activation function, iteration number, batch calculation number and loss function and the root mean square error are taken as eigenvalues, and the learning rate is taken as a predicted value to construct a model, and the learning rate and other hyperparameter combination values are obtained through the extreme gradient boosting tree model estimation;

[0043] The scene optimal super parameter combination is input into the long short-term memory method for model training to obtain a temperature-induced sensor life consumption model under algorithm fusion.

[0044] Further, the estimated output module is specifically configured to:

[0045] The sensor working condition data containing voltage, current and temperature information are input into the composite estimation model, the composite estimation model transmits the data to a voltage and current induced sensor aging model and a temperature induced sensor life consumption model respectively, the two models are predicted respectively, and a sensor life prediction result based on respective working conditions is given;

[0046] The composite estimation model receives the prediction results of the two models, integrates the prediction results by weighted average or linear combination, and outputs a comprehensive sensor life prediction result after integration.

[0047] In a third aspect, the present application provides a computer device, comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of a sensor life prediction method based on a fusion algorithm when executing the computer program.

[0048] In a fourth aspect, the present application provides a computer readable storage medium, wherein the computer readable storage medium stores a computer program, and the computer program implements the steps of a sensor life prediction method based on a fusion algorithm when executed by a processor.

[0049] Compared with the prior art, the present application has the following technical effects:

[0050] The present application adopts extreme gradient boosting tree (XGBoost) and long short-term memory algorithm (LSTM) to model the sensor sensitivity decline caused by voltage, current and temperature respectively. The best parameter combination is obtained by grid search algorithm to ensure high precision of the model.

[0051] For temperature data with strong time dependence and complexity, the learning rate and other super parameters are optimized by algorithm fusion (including reverse engineering model, simulated annealing algorithm and XGBoost) to further improve the prediction accuracy of the model.

[0052] The models under complex working conditions of temperature and voltage and current are combined to form a composite estimation model, which can more comprehensively reflect the life consumption of the sensor in the actual working environment.

[0053] The present application adopts intelligent fusion algorithm to realize the automation and intelligentization of sensor life prediction, greatly reduces the manual intervention and subjective judgment, and improves the prediction efficiency.

[0054] By predicting the life of the sensor, preventive maintenance can be performed in advance to avoid system downtime or accidents caused by sudden failure of the sensor, thereby reducing maintenance costs and potential economic losses.

[0055] The present application can predict potential risks in the future based on the working environment of the sensor, which helps to discover and handle possible safety hazards in time, enhancing the safety of the system.

[0056] Accurate life prediction provides strong data support for decision-making, helping to develop more scientific and reasonable maintenance strategies and management plans, and improving the reliability and stability of the system.

[0057] In summary, the technical effects of the present application mainly include improving the accuracy of sensor life prediction, improving prediction efficiency and reducing cost, enhancing the safety and reliability of the system, and technical innovation and application prospect. These technical effects together constitute the core value and competitive advantage of the present application. BRIEF DESCRIPTION OF DRAWINGS

[0058] Figure 1 The flowchart of the present application.

[0059] Figure 2 The algorithm fusion logic block diagram of the present application;

[0060] Figure 3 The actual prediction capability effect diagram of the present application;

[0061] Figure 4 The model fitting capability diagram of the present application; DETAILED DESCRIPTION

[0062] The present application is further described below in conjunction with the drawings:

[0063] Example 1, please refer to Figure 1 The present application provides a sensor life prediction method based on fusion algorithm, comprising:

[0064] Obtain the sensitivity data of the sensor under the working condition, and obtain the sensitivity reduction data caused by voltage, current and temperature respectively after preprocessing;

[0065] Based on the sensitivity reduction data caused by voltage and current, a model of sensor aging caused by voltage and current is constructed; based on the sensitivity reduction data caused by temperature, a sensor life consumption model caused by temperature is obtained by combining algorithm fusion;

[0066] The sensor life consumption model caused by temperature and the sensor aging model caused by voltage and current are combined to obtain a composite prediction model for sensor life prediction.

[0067] Continuing to analyze the technical effects described in the above paragraphs, we can delve into the following aspects:

[0068] The present application obtains the sensitivity data of the sensor under actual working conditions, pre-processes it, separates the influence of voltage, current and temperature on sensitivity, and provides accurate data basis for subsequent model construction. This fine data processing method helps to capture the subtle changes of sensor aging and improve the accuracy of prediction.

[0069] For the two main influencing factors of voltage and current and temperature, the present application respectively constructs the corresponding sensor aging model. This targeted modeling method can more accurately reflect the influence of different factors on the service life of the sensor, avoiding the prediction error caused by general processing in traditional methods.

[0070] In dealing with the sensor life consumption caused by temperature, the present application adopts an algorithm fusion strategy. By combining multiple algorithms (such as long short-term memory algorithm, reverse engineering model, simulated annealing algorithm, etc.), the advantages of each algorithm are fully utilized, improving the prediction performance and robustness of the model.

[0071] In the algorithm fusion process, the present application also focuses on the optimization of hyperparameters. By intelligently searching and selecting the best hyperparameter combination, the prediction accuracy and generalization ability of the model are further improved. This hyperparameter optimization method helps to avoid the problems of model overfitting or underfitting, making the prediction result more reliable.

[0072] The present application combines the sensor life consumption model caused by temperature and the sensor aging model caused by voltage and current to obtain a composite prediction model. This merging method can consider the influence of multiple factors on the service life of the sensor, improving the comprehensiveness and accuracy of prediction.

[0073] The composite prediction model can better adapt to complex and variable working conditions, providing a more reliable basis for the life prediction of the sensor. Whether in an environment with large voltage and current fluctuations or in a scene with drastic temperature changes, the composite prediction model can give accurate prediction results.

[0074] It realizes data-driven accurate prediction, improves the accuracy and reliability of sensor life prediction;

[0075] Through algorithm fusion and hyperparameter optimization, the prediction performance and robustness of the model are improved;

[0076] A composite prediction model is constructed, which considers the influence of multiple factors on the service life of the sensor, improving the comprehensiveness of prediction;

[0077] It adapts to complex and variable working conditions, providing strong support for the maintenance and management of the sensor.

[0078] These technical effects together constitute the core value and competitive advantage of the present application, providing a new approach and method for the development of sensor life prediction.

[0079] Embodiment 2, please refer to Figure 2 The present application provides a sensor life prediction method based on fusion algorithm, specifically comprising:

[0080] I. Obtain experimental data and perform data preprocessing:

[0081] 1. Build a sensor sensitivity test experimental environment.

[0082] 2. Obtain the magnetoresistance curve of the sensor through the experiment, obtain the corresponding sensitivity data of the sensor under complex working conditions through linear fitting, and remove abnormal data due to experimental errors.

[0083] 3. Perform dimensionality reduction and dimensionality reduction operations on the experimental data, unify all array dimensions to three-dimensional arrays, and mark the sensitivity reduction data caused by voltage, current and temperature respectively to facilitate subsequent algorithm fusion operations.

[0084] 4. Divide the test set and validation set of the experimental data, and complete the random sampling of the data set division through the random seed.

[0085] Please refer to Figure 3 The actual effect diagram of the present application, the experimental data obtained under different working conditions of a randomly selected sensor sample is used as the test data set, and the experimental data samples of the other 21 TMR sensor samples under different working conditions are used as the training set for model training to obtain the model prediction ability in the actual scene.

[0086] Referring to FIG. 4, the fitting ability diagram of the prediction model of the present application, the fitting ability of the training set and the experimental set is used to intuitively display the model prediction ability.

[0087] II. Algorithm fusion:

[0088] 1. For the experimental data part caused by the decrease of sensor sensitivity caused by voltage and current, the data change amplitude is small and the trend is linear. The best parameter combination is obtained through the grid search algorithm for prediction, and the optimal hyperparameter combination is input into the eXtreme Gradient Boost Tree (XGBoost) algorithm to construct a model for predicting the sensor aging caused by voltage and current.

[0089] 2. For the experimental data caused by temperature-induced sensor sensitivity decline with complex problems and strong time dependence, the long short term memory algorithm (Long Short Term Memory Neural Network, LSTM) is selected for prediction. When processing the neural network, the learning rate parameter, which is complex in parameter combination of the neural network and has an important influence on the prediction accuracy, is solved by algorithm fusion. The main algorithm fusion process is:

[0090] i. First, select the learning rate, grid division number, initializer, activation function, iteration number, batch calculation number and loss function hyperparameter combination, and set the initial array for the hyperparameters.

[0091] ii. The hyperparameter combination is traversed by random seed sampling, all hyperparameters are taken as the eigenvalues of the reverse engineering model, and the long short term memory method is used to construct the model. The root mean square error is taken as the predicted value of the reverse engineering model, and the reverse engineering model is constructed.

[0092] iii. For a large amount of data generated by the reverse engineering model, the model data is introduced into the simulated annealing algorithm to obtain the hyperparameters of the probability optimal solution and the batch reverse optimal array with the smallest root mean square error.

[0093] iv. The reverse optimal array is input into the extreme gradient boosting tree model, at this time the hyperparameter combination (remove learning rate) of grid division number, initializer, activation function, iteration number, batch calculation number and loss function and root mean square error are taken as eigenvalues, and learning rate is taken as predicted value to construct model, and learning rate and other hyperparameter combination values are obtained by extreme gradient boosting tree model prediction.

[0094] v. The scene optimal hyperparameter combination is input into the long short term memory method to train the model and obtain the temperature-induced sensor life consumption model under algorithm fusion.

[0095] 3. Algorithm fusion under complex working conditions:

[0096] The models under complex working conditions of temperature and voltage and current are combined, and the composite prediction model is completed by identifying different working conditions data entering the algorithm model.

[0097] III. Sensor life prediction:

[0098] 1. Model evaluation is performed from multiple directions by mean square error, root mean square error, mean absolute error and determination coefficient to verify the fitting and prediction ability of the model.

[0099] 2. The application uses actual experimental data to increase the model prediction ability evaluation method in the actual scene, and inputs the experimental data of a sample number selected by random sampling into the model to visually compare the predicted experimental data on the sampling node with the real experimental data, and verify the model prediction accuracy.

[0100] Unlike traditional detection methods, a prediction method for early warning of sensor aging failure is proposed. Through accelerated aging experiments, the sensitivity of the sensor under the influence of temperature and voltage and current is taken as a standard to estimate the service life. Through accelerated aging experiments, complex working conditions for the sensor in actual application are simulated, and the threshold value of the service life data of temperature is explored through experiments, and a more effective and accurate data framework is built for intelligent prediction.

[0101] Embodiment 3, a tunneling magnetoresistance sensor life prediction method based on intelligent fusion algorithm, comprising the following steps:

[0102] Step 1: When processing data under complex working conditions, the algorithm model can better consider the alternating test of the sensor aging condition under the real experimental environment when the complex working conditions of temperature and voltage and current are applied. 200 pieces of accelerated aging experimental data are randomly selected as verification data under 22 different complex working conditions.

[0103] Step 2: A fusion algorithm based on an extreme gradient boosting tree algorithm and a long short-term memory neural network algorithm is used to train the model.

[0104] Step 3: In order to better improve the credibility of the invention, the experimental data obtained under different working conditions of a randomly selected sensor sample are used as the test data set, and the experimental data samples of the other 21 sensor samples under different working conditions are used as the training set for model training, and the reliability is verified by randomly sampling the test sample data. The machine learning prediction model can accurately predict the sensor life aging trend under different complex conditions.

[0105] Step 4: Compare the mainstream algorithm, wherein SA+XGB+LSTM (Simulated Annealing + eXtreme Gradient Boosting + Long Short Term Memory) is a machine learning algorithm fusion framework that simulates annealing algorithm, extreme boosting algorithm and long short-term memory method. When processing data close to the actual situation to obtain simulated complex working condition data, the prediction accuracy of the algorithm fusion model is reduced by more than 60% compared with other mainstream regression prediction algorithms, and the prediction accuracy is improved by 3%. In the current sensor accelerated aging experiment, it has obvious advantages in life prediction accuracy.

[0106] In still another embodiment of the present application, a sensor life prediction system based on a fusion algorithm is provided, which can be used to implement the above-mentioned sensor life prediction method based on a fusion algorithm. Specifically, the system comprises:

[0107] A data acquisition module is configured to acquire sensitivity data of the sensor under working conditions, and obtain voltage-current and temperature-induced sensitivity degradation data after preprocessing.

[0108] A model construction module is configured to construct a voltage-current-induced sensor aging model based on voltage-current-induced sensitivity degradation data, and obtain a temperature-induced sensor life consumption model by algorithm fusion based on temperature-induced sensitivity degradation data.

[0109] A prediction output module is configured to combine the temperature-induced sensor life consumption model and the voltage-current-induced sensor aging model to obtain a composite prediction model, and perform sensor life prediction.

[0110] Further, the data acquisition module is specifically configured to:

[0111] Obtain a sensor magnetoresistance curve through experiments, and obtain the corresponding sensitivity data of the sensor under complex working conditions through linear fitting, and remove abnormal data with experimental errors.

[0112] Perform dimensionality reduction and dimensionality reduction operations on the corresponding sensitivity data of the sensor under complex working conditions, unify all array dimensions to three arrays, and mark the voltage-current and temperature-induced sensitivity degradation data to obtain the voltage-current and temperature-induced sensitivity degradation data.

[0113] Further, the model construction module is specifically configured to:

[0114] For experimental data of voltage-current-induced sensor sensitivity degradation, the best parameter combination is obtained through a grid search algorithm for prediction, and the optimal hyperparameter combination is input into an extreme gradient boosting tree algorithm to construct a voltage-current-induced sensor aging model.

[0115] Further, the model construction module is specifically configured to:

[0116] For experimental data of temperature-induced sensor sensitivity degradation, a long short-term memory algorithm is selected for prediction, and when performing neural network processing, the learning rate parameter of the neural network is constructed into a temperature-induced sensor life consumption model by algorithm fusion.

[0117] Further, the model construction module is specifically configured to:

[0118] Select the hyperparameter combination of learning rate, grid division number, initializer, activation function, iteration number, batch calculation number and loss function, and set the initial array for the hyperparameters;

[0119] By random seed sampling traversing the hyperparameter combination, all hyperparameters are taken as the eigenvalues of the reverse engineering model, and the long short-term memory method is used for model construction; the root mean square error is taken as the predicted value of the reverse engineering model, and the reverse engineering model is constructed;

[0120] For a large amount of data generated by the reverse engineering model, the model data is introduced into the simulated annealing algorithm to obtain the hyperparameters of the optimal solution and the batch reverse optimal array with the minimum root mean square error;

[0121] The reverse optimal array is input into the extreme gradient boosting tree model, at this time the hyperparameter combination of grid division number, initializer, activation function, iteration number, batch calculation number and loss function and the root mean square error are taken as eigenvalues, and the learning rate is taken as a predicted value to construct a model, and the learning rate and other hyperparameter combination values are obtained through the extreme gradient boosting tree model estimation;

[0122] The scene optimal hyperparameter combination is input into the long short-term memory method for model training to obtain the temperature-induced sensor life consumption model under algorithm fusion.

[0123] Further, the estimation output module is specifically used for:

[0124] The sensor working condition data containing voltage, current and temperature information are input into the composite estimation model, the composite estimation model transmits the data to the voltage and current induced sensor aging model and the temperature induced sensor life consumption model respectively, two models are predicted respectively, and the sensor life prediction results based on respective working conditions are given;

[0125] The composite estimation model receives the prediction results of the two models, integrates them through weighted average or linear combination, and outputs the comprehensive sensor life prediction result after integration.

[0126] The division of the modules in the embodiments of the application is illustrative, and is only a logical function division; in actual implementation, another division mode can be used; in addition, the function modules in each embodiment of the application can be integrated in one processor, or can be physically separated, or two or more modules can be integrated in one module. The integrated modules can be realized in the form of hardware or in the form of software function modules.

[0127] In still another embodiment of the present application, a computer device is provided, which comprises a processor and a memory, the memory being configured to store a computer program, the computer program comprising program instructions, and the processor being configured to execute the program instructions stored in the computer storage medium. The processor can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic, discrete hardware components, etc., which are the computing core and control core of the terminal, and are suitable for implementing one or more instructions, and are particularly suitable for loading and executing one or more instructions in the computer storage medium to implement a corresponding method flow or corresponding function; the processor in the embodiments of the present application can be used for the operation of the sensor life prediction method based on the fusion algorithm.

[0128] In still another embodiment of the present application, the present application further provides a storage medium, specifically a computer readable storage medium (Memory), which is a memory device in the computer device, and is configured to store programs and data. It can be understood that the computer readable storage medium herein can include the built-in storage medium in the computer device, and of course can also include the expansion storage medium supported by the computer device. The computer readable storage medium provides a storage space, which stores the operating system of the terminal. Moreover, one or more instructions suitable for being loaded and executed by the processor are also stored in the storage space, and these instructions can be one or more computer programs (including program codes). It should be noted that the computer readable storage medium herein can be a high-speed RAM memory, or a non-volatile memory such as at least one disk memory. One or more instructions stored in the computer readable storage medium can be loaded and executed by the processor to implement the corresponding steps of the sensor life prediction method based on the fusion algorithm in the above embodiments.

[0129] Those skilled in the art will appreciate that embodiments of the application can be devised for a method, a system, or a computer program product. Accordingly, the present application can be embodied in the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application can take the form of a computer program product on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage devices, etc.) embodying computer readable program code.

[0130] The present application is described in reference to the flowchart and / or block diagrams of the method, apparatus (system) and computer program product according to embodiments of the application. It will be understood that each block of the flowchart and / or block diagrams, and combinations of blocks in the flowchart and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general purpose computer, special purpose computer, embedded processing device or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart and / or block diagram block or blocks. Figure 1 one or more functions specified in the flowchart and / or block diagram block or blocks. Figure 1 means for carrying out each of the one or more functions specified in the flowchart and / or block diagram block or blocks.

[0131] These computer program instructions can also be stored in a computer- readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instructions which implement the function specified in the flowchart and / or block diagram block or blocks. Figure 1 one or more functions specified in the flowchart and / or block diagram block or blocks. Figure 1 means for carrying out each of the one or more functions specified in the flowchart and / or block diagram block or blocks.

[0132] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart and / or block diagram block or blocks. Figure 1 one or more functions specified in the flowchart and / or block diagram block or blocks. Figure 1 means for carrying out each of the one or more functions specified in the flowchart and / or block diagram block or blocks.

[0133] Finally, it should be noted that the above-mentioned embodiments are merely intended for describing and illustrating, not limiting, the technical solutions of the present application. Although the present application has been described in detail with reference to the above-mentioned embodiments, those skilled in the art should understand that the specific embodiments of the present application can be modified or replaced by equivalents without departing from the spirit and scope of the present application, and any modification or equivalent replacement without departing from the spirit and scope of the present application should be covered in the protection scope of the claims of the present application.

Claims

1. A sensor life prediction method based on a fusion algorithm, characterized by, The method comprises the following steps: obtaining sensitivity data of the sensor corresponding to the working condition to obtain sensitivity reduction data caused by voltage, current and temperature respectively; based on the sensitivity reduction data caused by voltage and current, a model of sensor aging caused by voltage and current is constructed; based on the sensitivity reduction data caused by temperature, a temperature-induced sensor life consumption model is obtained by algorithm fusion; merge the temperature-induced sensor life consumption model and the sensor aging model caused by voltage and current to obtain a composite prediction model, and predict the sensor life according to the composite prediction model; based on the sensitivity reduction data caused by voltage and current, the model of sensor aging caused by voltage and current comprises: for the experimental data of the sensor sensitivity reduction caused by voltage and current, the best parameter combination is obtained by grid search algorithm for prediction, and the optimal hyperparameter combination is input into the extreme gradient boosting tree algorithm to construct the model of sensor aging caused by voltage and current; based on the sensitivity reduction data caused by temperature, the temperature-induced sensor life consumption model is obtained by algorithm fusion, comprising: for the experimental data of the sensor sensitivity reduction caused by temperature, the long short-term memory algorithm is selected for prediction, and when the neural network is processed, the learning rate parameter of the neural network is fused to construct the temperature-induced sensor life consumption model; the algorithm fusion to construct the temperature-induced sensor life consumption model comprises: selecting the learning rate, the grid division number, the initializer, the activation function, the iteration number, the batch calculation number and the loss function hyperparameter combination, and setting the initial array for the hyperparameters; all hyperparameters are taken as the inverse engineering model characteristic value by random seed sampling to traverse the hyperparameter combination, and the long short-term memory method is used for model construction; the root mean square error is taken as the prediction value of the inverse engineering model to construct the inverse engineering model; for a large amount of data generated by the inverse engineering model, the model data is introduced into the simulated annealing algorithm to obtain the hyperparameters of the probability optimal solution and the batch reverse optimal array with the minimum root mean square error; the reverse optimal array is input into the extreme gradient boosting tree model, at this time, the grid division number, the initializer, the activation function, the iteration number, the batch calculation number, the loss function hyperparameter combination and the root mean square error are taken as the characteristic value, and the learning rate is taken as the prediction value to construct the model, and the learning rate and other hyperparameter combination values are obtained by the extreme gradient boosting tree model prediction; the scene optimal hyperparameter combination is input into the long short-term memory method for model training to obtain the temperature-induced sensor life consumption model under algorithm fusion.

2. The sensor life prediction method based on fusion algorithm according to claim 1, characterized in that, the method comprises the following steps: obtaining the sensor's magnetoresistance curve through experiments, and obtaining the sensor's sensitivity data corresponding to the complex working condition by linear fitting to remove the abnormal data of experimental error; the sensitivity data of the sensor corresponding to the complex working condition is operated by dimensionality reduction and dimensionality reduction, all array dimensions are unified to three arrays, and the sensitivity reduction data caused by voltage, current and temperature is labeled to obtain the sensitivity reduction data caused by voltage, current and temperature respectively.

3. The sensor life prediction method based on fusion algorithm according to claim 1, characterized in that, The temperature-induced sensor life consumption model and the voltage and current-induced sensor aging model are combined to obtain a composite prediction model, and sensor life prediction is performed according to the composite prediction model, including: The sensor working condition data containing voltage, current and temperature information are input into the composite prediction model, the composite prediction model transmits the data to the voltage and current-induced sensor aging model and the temperature-induced sensor life consumption model respectively, two models are predicted respectively, and the sensor life prediction results based on the respective working conditions are given; The composite prediction model receives the prediction results of the two models, integrates them by weighted average or linear combination, and outputs the comprehensive sensor life prediction result after integration.

4. A sensor life prediction system based on fusion algorithm, characterized by, It includes: The data acquisition module is used for acquiring the sensitivity data of the sensor under the working condition, and the preprocessed voltage and current and temperature-induced sensitivity reduction data are obtained; The model construction module is used for constructing the voltage and current-induced sensor aging model based on the voltage and current-induced sensitivity reduction data, and obtaining the temperature-induced sensor life consumption model by algorithm fusion based on the temperature-induced sensitivity reduction data; The prediction output module is used for combining the temperature-induced sensor life consumption model and the voltage and current-induced sensor aging model to obtain a composite prediction model, and performing sensor life prediction; The model construction module is specifically used for: For the experimental data of the sensor sensitivity reduction caused by voltage and current, the best parameter combination is obtained by grid search algorithm for prediction, and the optimal hyperparameter combination is input into the extreme gradient boosting tree algorithm to construct the voltage and current-induced sensor aging model; The model construction module is specifically used for: For the experimental data of the temperature-induced sensor sensitivity reduction, the long short-term memory algorithm is selected for prediction, and the learning rate parameter of the neural network is constructed into the temperature-induced sensor life consumption model by algorithm fusion when the neural network is processed; The model construction module is specifically used for: Selecting the learning rate, grid division number, initializer, activation function, iteration number, batch calculation number and loss function hyperparameter combination, and setting the initial array for the hyperparameters; Through random seed sampling to traverse the hyperparameter combination, all hyperparameters are taken as the inverse engineering model characteristic value, and the long short-term memory method is used for model construction; the root mean square error is taken as the prediction value of the inverse engineering model, and the inverse engineering model is constructed; For a large amount of data generated by the inverse engineering model, the model data is introduced into the simulated annealing algorithm to obtain the hyperparameters of the probability optimal solution and the batch reverse optimal array with the minimum root mean square error; The reverse optimal array is input into the extreme gradient boosting tree model, at this time the grid division number, initializer, activation function, iteration number, batch calculation number, loss function hyperparameter combination and root mean square error are taken as characteristic values, and the learning rate is taken as a prediction value to construct a model, and the learning rate and other hyperparameter combination values are obtained by the extreme gradient boosting tree model prediction; The scene optimal hyperparameter combination is input into the long short-term memory method for model training, and the temperature-induced sensor life consumption model under algorithm fusion is obtained.

5. The sensor life prediction system based on fusion algorithm according to claim 4, wherein, The data acquisition module is specifically configured to: Obtain the magnetoresistance curve of the sensor through experiments, and obtain the sensitivity data of the sensor under complex working conditions through linear fitting to remove abnormal data caused by experimental errors; Perform dimensionality reduction and dimensionality increase operations on the sensitivity data of the sensor under complex working conditions, unify all array dimensions to three-dimensional arrays, and mark the sensitivity reduction data caused by voltage, current and temperature respectively to obtain the sensitivity reduction data caused by voltage, current and temperature respectively.

6. The sensor life prediction system based on fusion algorithm of claim 4, wherein, The estimated output module is specifically configured to: Input the sensor working condition data containing voltage, current and temperature information into the composite estimation model, and the composite estimation model transmits the data to the model of sensor aging caused by voltage and current and the model of sensor life consumption caused by temperature respectively, two models are predicted respectively, and the sensor life prediction results based on respective working conditions are given; The composite estimation model receives the prediction results of the two models, integrates the prediction results through weighted average or linear combination, and outputs the comprehensive sensor life prediction result after integration.

7. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, The processor executes the computer program to realize the steps of the sensor life prediction method based on the fusion algorithm in any one of claims 1 to 3.

8. A computer-readable storage medium storing a computer program, the computer-readable storage medium comprising: The computer program is executed by the processor to realize the steps of the sensor life prediction method based on the fusion algorithm in any one of claims 1 to 3.

Images