Sensor life prediction method based on fusion algorithm and related device

Abstract

The invention belongs to the technical field of sensor life prediction, and discloses a sensor life prediction method based on a fusion algorithm and a related device, and the method comprises the steps: obtaining the corresponding sensitivity data of a sensor under a working condition, and obtaining the sensitivity reduction data caused by voltage, current and temperature after preprocessing; based on sensitivity reduction data caused by the voltage and the current, a model of sensor aging caused by the voltage and the current is constructed; based on sensitivity drop data caused by temperature, combining with algorithm fusion to obtain a sensor life consumption model caused by temperature; and combining the sensor life consumption model caused by the temperature and the sensor aging model caused by the voltage and the current to obtain a composite prediction model, and predicting the life of the sensor. For sensor sensitivity reduction caused by voltage, current and temperature, an extreme gradient boosting tree and a long-short term memory algorithm are respectively adopted for modeling. An optimal parameter combination is obtained through a grid search algorithm, and high precision of the model is ensured.

Description

Technical Field

[0001] The present invention belongs to the technical field of sensor life prediction, and in particular relates to a sensor life prediction method based on a fusion algorithm and a related device. Background Art

[0002] Intelligent sensor life prediction technology refers to obtaining basic prediction data through accelerated aging experiments on target sensors, establishing prediction models according to the characteristics of experimental characteristic variables, and obtaining the optimal prediction sensitivity reduction capability under the sensor life scenario after algorithm fusion of the optimal prediction models under independent scenarios. The sensor detection method in the traditional power system consumes a lot of manpower and material resources and is inefficient, and it is difficult to discover the hidden faults of the mechanism.

[0003] The traditional sensor detection method in the power system consumes a lot of manpower and material resources and is inefficient, and it is difficult to discover the hidden faults of the mechanism.

[0004] The learning rate in deep learning is a crucial hyperparameter that determines the speed and direction of the model's weight parameter updates during training. It controls how much we adjust the network's weights to match the gradient loss. The lower the value, the slower the descent along the gradient. Current technology usually directly selects a small learning rate when using a neural network algorithm to ensure that no local minimum points are missed, but it will also take a long time to converge. Summary of the invention

[0005] The purpose of the present invention is to provide a sensor life prediction method and related devices based on a fusion algorithm to solve the problems of large consumption of manpower and material resources in the prior art and too long convergence time of the neural network algorithm.

[0006] To achieve the above object, the present invention adopts the following technical solutions: In a first aspect, the present invention provides a sensor life prediction method based on a fusion algorithm, comprising: Obtain the corresponding sensitivity data of the sensor under working conditions, and obtain the 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 sensor life consumption model caused by temperature is obtained by combining algorithm fusion; 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, and the sensor life is predicted based on the composite prediction model.

[0007] Furthermore, the obtaining of the sensitivity data corresponding to the sensor under the working condition, and obtaining the sensitivity reduction data caused by the voltage, current and temperature respectively, includes: The magnetic resistance curve of the sensor is obtained experimentally, and the sensitivity data of the sensor under complex working conditions is obtained through linear fitting, and the abnormal data of experimental errors are removed; The sensitivity data of the sensor under complex working conditions is upgraded and reduced in dimension, all array dimensions are unified into three-dimensional arrays, and the sensitivity reduction data caused by voltage, current and temperature are marked to obtain the sensitivity reduction data caused by voltage, current and temperature respectively; The sensitivity reduction data caused by voltage, current and temperature are divided into test set and validation set respectively, and the random sampling data set division is completed by random seeds.

[0008] Furthermore, the construction of a sensor aging model caused by voltage and current based on the sensitivity reduction data caused by voltage and current includes: According to the experimental data of sensor sensitivity decrease caused by voltage and current, the optimal 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 sensor aging caused by voltage and current.

[0009] Furthermore, the sensitivity reduction data caused by temperature is combined with an algorithm to obtain a sensor life consumption model caused by temperature, including: According to the experimental data of the decrease in sensor sensitivity caused by temperature, the long short-term memory algorithm was selected for estimation. When performing neural network processing, the algorithm fusion was used to construct the sensor life consumption model caused by temperature in the face of the learning rate parameters of the neural network.

[0010] Furthermore, the use of algorithm fusion to construct a sensor life consumption model caused by temperature specifically includes: First, select the hyperparameter combination of learning rate, number of grid divisions, initializer, activation function, number of iterations, number of batch calculations, and loss function, and set the initial array for the hyperparameters; Through random seed sampling, we traverse the hyperparameter combinations, use all hyperparameters as the eigenvalues ​​of the reverse engineering model, and construct the model using the long short-term memory method; we use the root mean square error as the predicted value of the reverse engineering model to build the reverse engineering model; For the 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 probabilistic 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 hyperparameter combination of the number of grid divisions, initializer, activation function, number of iterations, number of batch calculations, loss function and the root mean square error are used as feature values, and the learning rate is used as the predicted value to build a model. The learning rate and other hyperparameter combination values ​​are estimated by the extreme gradient boosting tree model. The optimal hyperparameter combination of the scenario is input into the long short-term memory method for model training to obtain the sensor life consumption model caused by temperature under algorithm fusion.

[0011] Furthermore, 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, and the sensor life prediction is performed according to the composite prediction model, including: The sensor operating condition data containing voltage, current and temperature information is input into the composite prediction model. The composite prediction model passes the data to the model of sensor aging caused by voltage and current and the model of sensor life consumption caused by temperature respectively. The two models make predictions respectively and give sensor life prediction results based on their respective operating conditions. The composite prediction model receives the prediction results of the two models, integrates them by weighted average and linear combination, and after integration, the composite prediction model outputs a comprehensive sensor life prediction result.

[0012] In a second aspect, the present invention provides a sensor life prediction system based on a fusion algorithm, comprising: The data acquisition module is used to obtain the sensitivity data of the sensor under working conditions, and obtain the sensitivity reduction data caused by voltage, current and temperature respectively after preprocessing; The model building module is used to build a model of sensor aging caused by voltage and current based on the sensitivity reduction data caused by voltage and current; based on the sensitivity reduction data caused by temperature, the sensor life consumption model caused by temperature is obtained by combining algorithm fusion; The estimation output module is used to combine the sensor life consumption model caused by temperature and the sensor aging model caused by voltage and current to obtain a composite estimation model to predict the sensor life.

[0013] Furthermore, the data acquisition module is specifically used for: The magnetic resistance curve of the sensor is obtained experimentally, and the sensitivity data of the sensor under complex working conditions is obtained through linear fitting, and the abnormal data of experimental errors are removed; The sensitivity data of the sensor under complex working conditions are upgraded and reduced in dimension, all array dimensions are unified into three-dimensional arrays, and the sensitivity reduction data caused by voltage, current and temperature are marked to obtain the sensitivity reduction data caused by voltage, current and temperature.

[0014] Furthermore, the model building module is specifically used to: According to the experimental data of sensor sensitivity decrease caused by voltage and current, the optimal 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 sensor aging caused by voltage and current.

[0015] Furthermore, the model building module is specifically used to: According to the experimental data of the decrease in sensor sensitivity caused by temperature, the long short-term memory algorithm was selected for estimation. When performing neural network processing, the algorithm fusion was used to construct the sensor life consumption model caused by temperature in the face of the learning rate parameters of the neural network.

[0016] Furthermore, the model building module is specifically used to: Select the hyperparameter combination of learning rate, number of grid divisions, initializer, activation function, number of iterations, number of batch calculations, and loss function, and set the initial array for the hyperparameters; Through random seed sampling, we traverse the hyperparameter combinations, use all hyperparameters as the eigenvalues ​​of the reverse engineering model, and construct the model using the long short-term memory method; we use the root mean square error as the predicted value of the reverse engineering model to build the reverse engineering model; For the 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 probabilistic 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 hyperparameter combination of the number of grid divisions, initializer, activation function, number of iterations, number of batch calculations, loss function and the root mean square error are used as feature values, and the learning rate is used as the predicted value to build a model. The learning rate and other hyperparameter combination values ​​are estimated by the extreme gradient boosting tree model. The optimal hyperparameter combination of the scenario is input into the long short-term memory method for model training to obtain the sensor life consumption model caused by temperature under algorithm fusion.

[0017] Furthermore, the estimated output module is specifically used for: The sensor working condition data including voltage, current and temperature information is input into the composite prediction model. The composite prediction model transmits the data to the sensor aging model caused by voltage and current and the sensor life consumption model caused by temperature. The two models make predictions respectively and give the sensor life prediction results based on their respective working conditions. The composite prediction model receives the prediction results of the two models and integrates them by weighted average or linear combination. After integration, the composite prediction model outputs a comprehensive sensor life prediction result.

[0018] In a third aspect, the present invention 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.

[0019] In a fourth aspect, the present invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of a sensor life prediction method based on a fusion algorithm.

[0020] Compared with the prior art, the present invention has the following technical effects: The present invention uses extreme gradient boosting tree (XGBoost) and long short-term memory algorithm (LSTM) to model the decrease in sensor sensitivity caused by voltage, current and temperature. The grid search algorithm is used to obtain the best parameter combination to ensure the high accuracy of the model.

[0021] For temperature data with strong time dependence and complexity, algorithm fusion (including reverse engineering model, simulated annealing algorithm and XGBoost) is used to optimize hyperparameters such as learning rate, which further improves the prediction accuracy of the model.

[0022] The models under complex working conditions of temperature, 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.

[0023] The present invention adopts an intelligent fusion algorithm to realize the automation and intelligence of sensor life prediction, greatly reduces manual intervention and subjective judgment, and improves prediction efficiency.

[0024] By predicting the life of the sensor, preventive maintenance can be performed in advance to avoid system downtime or accidents caused by sudden sensor failure, thereby reducing maintenance costs and potential economic losses.

[0025] The present invention can predict future potential risks based on the working environment of the sensor, which helps to timely discover and deal with possible safety hazards and enhance the safety of the system.

[0026] Accurate life prediction provides strong data support for decision-making, helps to formulate more scientific and reasonable maintenance strategies and management plans, and improves the reliability and stability of the system.

[0027] In summary, the technical effects of the present invention are mainly reflected in improving the accuracy of sensor life prediction, improving prediction efficiency and reducing costs, enhancing system safety and reliability, as well as technical innovation and application prospects. These technical effects together constitute the core value and competitive advantage of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 It is a flow chart of the present invention.

[0029] Figure 2 It is the algorithm fusion logic block diagram of the present invention; Figure 3 This is a diagram showing the actual prediction capability of the present invention; Figure 4 It is a schematic diagram of the model fitting ability of the present invention; DETAILED DESCRIPTION The present invention is further described below in conjunction with the accompanying drawings: Example 1, please refer to Figure 1 The present invention provides a sensor life prediction method based on a fusion algorithm, comprising: Obtain the corresponding sensitivity data of the sensor under working conditions, and obtain the sensitivity reduction data caused by voltage, current and temperature respectively after preprocessing; 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; The sensor life consumption model caused by temperature and the sensor aging model caused by voltage and current are combined to obtain a composite estimation model for sensor life prediction.

[0030] Continuing to analyze the technical effects described in the above paragraphs, we can conduct in-depth discussions from the following aspects: The present invention 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 an accurate data basis for subsequent model construction. This refined data processing method helps to capture subtle changes in sensor aging and improve the accuracy of prediction.

[0031] The present invention constructs corresponding sensor aging models for the two main influencing factors, voltage and current, and temperature. This targeted modeling method can more accurately reflect the impact of different factors on sensor life and avoid the prediction errors caused by general processing in traditional methods.

[0032] When dealing with the sensor life consumption caused by temperature, the present invention adopts the strategy of algorithm fusion. 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 to improve the prediction performance and robustness of the model.

[0033] In the process of algorithm fusion, the present invention 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 method of hyperparameter optimization helps to avoid the problem of overfitting or underfitting of the model, making the prediction results more reliable.

[0034] The present invention 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 combination method can comprehensively consider the impact of multiple factors on sensor life, thereby improving the comprehensiveness and accuracy of the prediction.

[0035] The composite prediction model can better adapt to complex and changing working conditions and provide a more reliable basis for predicting the life of the sensor. Whether in an environment with large voltage and current fluctuations or in a scenario with drastic temperature changes, the composite prediction model can give accurate prediction results.

[0036] It realizes data-driven accurate prediction and improves the accuracy and reliability of sensor life prediction; The prediction performance and robustness of the model are improved through algorithm fusion and hyperparameter optimization; A composite prediction model was constructed, which comprehensively considered the impact of multiple factors on sensor life and improved the comprehensiveness of the prediction; It adapts to complex and changeable working conditions and provides strong support for the maintenance and management of sensors.

[0037] These technical effects together constitute the core value and competitive advantage of the present invention, and provide new ideas and methods for the development of the field of sensor life prediction.

[0038] Example 2, please refer to Figure 2 The present invention provides a sensor life prediction method based on a fusion algorithm, which specifically includes: 1. Obtain experimental data and perform data preprocessing: 1. Build a sensor sensitivity test experimental environment.

[0039] 2. Experimentally obtain the magnetic resistance curve of the sensor, obtain the corresponding sensitivity data of the sensor under complex working conditions through linear fitting, and remove abnormal data due to experimental errors.

[0040] 3. Perform dimension-up and dimension-down operations on the experimental data, unify all array dimensions into three-dimensional arrays, and mark the sensitivity reduction data caused by voltage, current, and temperature, respectively, to facilitate subsequent algorithm fusion operations.

[0041] 4. Divide the experimental data into test sets and validation sets, and complete the random sampling data set division through random seeds.

[0042] See also Figure 3 , the estimated actual effect diagram of the present invention, randomly selects the experimental data obtained under different working conditions of one sensor sample as the test data set, and uses the experimental data samples of the other 21 TMR sensor samples under different working conditions as the training set for model training to obtain the model prediction ability in the actual scenario.

[0043] See FIG. 4 , which is a schematic diagram of the estimated model fitting ability of the present invention, which intuitively demonstrates the model estimation ability through the fitting ability of the training set and the experimental set.

[0044] 2. Algorithm Fusion: 1. For the experimental data of sensor sensitivity reduction caused by voltage and current, the data change amplitude is small and the trend is quasi-linear. 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 (eXtremeGradientBoost Tree, XGBoost) to build a model of sensor aging caused by voltage and current.

[0045] 2. For the experimental data of sensor sensitivity reduction caused by temperature, which is complex and time-dependent, the Long Short Term Memory Neural Network (LSTM) algorithm is selected for estimation. When performing neural network processing, the learning rate parameter of the neural network has a complex parameter combination and has an important influence on the estimation accuracy. The present invention solves it by algorithm fusion. The main algorithm fusion process is: i. First, select the hyperparameter combination of learning rate, number of grid divisions, initializer, activation function, number of iterations, number of batch calculations, and loss function, and set the initial array for the hyperparameters.

[0046] ii. Through random seed sampling, we traverse the hyperparameter combinations, take all hyperparameters as the eigenvalues ​​of the reverse engineering model, and build the model using the long short-term memory method. We take the root mean square error as the predicted value of the reverse engineering model and build the reverse engineering model.

[0047] iii. For the 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 probabilistic optimal solution and the batch reverse optimal array with the minimum root mean square error.

[0048] iv. Input the reverse optimal array into the extreme gradient boosting tree model. At this time, the hyperparameter combination of the number of grid divisions, initializer, activation function, number of iterations, number of batch calculations, loss function (excluding the learning rate) and the root mean square error are used as feature values, and the learning rate is used as the predicted value to build the model. The learning rate and other hyperparameter combination values ​​are estimated by the extreme gradient boosting tree model.

[0049] v. Input the optimal hyperparameter combination of the scenario into the long short-term memory method for model training to obtain the sensor life consumption model caused by temperature under algorithm fusion.

[0050] 3. Algorithm fusion under complex working conditions: The models under complex working conditions of temperature, voltage and current are merged, and the composite estimation model is completed by identifying the algorithm models that enter the data under different working conditions.

[0051] 3. Sensor life estimation: 1. Evaluate the model from multiple directions through mean square error, root mean square error, mean absolute error, and determination coefficient to verify the model's fitting and prediction capabilities.

[0052] 2. The present invention uses actual experimental data to increase the model prediction ability assessment method in actual scenarios. By randomly sampling and selecting a sample number of experimental data to input into the model, the predicted experimental data on the sampling node is intuitively compared with the actual experimental data to verify the model prediction accuracy.

[0053] Different from the traditional detection method, a prediction method for early warning sensor aging failure is proposed. The sensitivity of the sensor under the influence of temperature, voltage and current is obtained through accelerated aging experiments, and the life estimation is carried out based on the standard. Accelerated aging experiments are used to simulate complex working conditions for sensors in practical applications, and the life data threshold of temperature is explored through experiments, which builds a more effective and accurate data framework for intelligent estimation.

[0054] Embodiment 3, a tunneling magnetoresistive sensor life prediction method based on an intelligent fusion algorithm, comprises the following steps: Step 1: When processing data under complex working conditions, the algorithm model can better consider the complex working conditions of temperature, voltage and current, and alternately test and simulate the sensor aging in the real experimental environment. 200 accelerated aging test data were randomly selected to verify data under 22 different complex working conditions.

[0055] Step 2: Use a fusion algorithm based on the extreme gradient boosting tree algorithm and the long short-term memory neural network algorithm to train the model.

[0056] Step 3: In order to better improve the credibility of the invention, the experimental data obtained from one sensor sample under different working conditions were randomly selected as the test data set, and the experimental data samples of the other 21 sensor samples under different working conditions were used as the training set for model training, and the reliability of the test sample data was verified by randomly sampling points. The machine learning prediction model can predict the aging trend of sensor life under different complex conditions with high accuracy.

[0057] Step 4: Compare the mainstream algorithms, where SA+XGB+LSTM (Simulated Annealing + eXtremeGradient Boosting + Long Short Term Memory) is a machine learning algorithm fusion framework that combines the simulated annealing algorithm with the extreme boosting algorithm and the long short-term memory method. When processing data close to the actual situation to simulate complex working conditions, the algorithm fusion model's estimation accuracy is more than 60% lower than other mainstream regression estimation algorithms, and the estimation accuracy is improved by 3%. It has a significant advantage in life estimation accuracy in the current sensor accelerated aging experiment.

[0058] In yet another embodiment of the present invention, 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 includes: The data acquisition module is used to obtain the sensitivity data of the sensor under working conditions, and obtain the sensitivity reduction data caused by voltage, current and temperature respectively after preprocessing; The model building module is used to build a model of sensor aging caused by voltage and current based on the sensitivity reduction data caused by voltage and current; based on the sensitivity reduction data caused by temperature, the sensor life consumption model caused by temperature is obtained by combining algorithm fusion; The estimation output module is used to combine the sensor life consumption model caused by temperature and the sensor aging model caused by voltage and current to obtain a composite estimation model to predict the sensor life.

[0059] Furthermore, the data acquisition module is specifically used for: The magnetic resistance curve of the sensor is obtained experimentally, and the sensitivity data of the sensor under complex working conditions is obtained through linear fitting, and the abnormal data of experimental errors are removed; The sensitivity data of the sensor under complex working conditions are upgraded and reduced in dimension, all array dimensions are unified into three-dimensional arrays, and the sensitivity reduction data caused by voltage, current and temperature are marked to obtain the sensitivity reduction data caused by voltage, current and temperature.

[0060] Furthermore, the model building module is specifically used to: According to the experimental data of sensor sensitivity decrease caused by voltage and current, the optimal 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 sensor aging caused by voltage and current.

[0061] Furthermore, the model building module is specifically used to: According to the experimental data of the decrease in sensor sensitivity caused by temperature, the long short-term memory algorithm was selected for estimation. When performing neural network processing, the algorithm fusion was used to construct the sensor life consumption model caused by temperature in the face of the learning rate parameters of the neural network.

[0062] Furthermore, the model building module is specifically used to: Select the hyperparameter combination of learning rate, number of grid divisions, initializer, activation function, number of iterations, number of batch calculations, and loss function, and set the initial array for the hyperparameters; Through random seed sampling, we traverse the hyperparameter combinations, use all hyperparameters as the eigenvalues ​​of the reverse engineering model, and construct the model using the long short-term memory method; we use the root mean square error as the predicted value of the reverse engineering model to build the reverse engineering model; For the 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 probabilistic 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 hyperparameter combination of the number of grid divisions, initializer, activation function, number of iterations, number of batch calculations, loss function and the root mean square error are used as feature values, and the learning rate is used as the predicted value to build a model. The learning rate and other hyperparameter combination values ​​are estimated by the extreme gradient boosting tree model. The optimal hyperparameter combination of the scenario is input into the long short-term memory method for model training to obtain the sensor life consumption model caused by temperature under algorithm fusion.

[0063] Furthermore, the estimated output module is specifically used for: The sensor working condition data including voltage, current and temperature information is input into the composite prediction model. The composite prediction model transmits the data to the sensor aging model caused by voltage and current and the sensor life consumption model caused by temperature. The two models make predictions respectively and give the sensor life prediction results based on their respective working conditions. The composite prediction model receives the prediction results of the two models and integrates them by weighted average or linear combination. After integration, the composite prediction model outputs a comprehensive sensor life prediction result.

[0064] The division of modules in the embodiments of the present invention is schematic and is only a logical function division. There may be other division methods in actual implementation. In addition, each functional module in each embodiment of the present invention may be integrated into one processor, or may exist physically separately, or two or more modules may be integrated into one module. The above-mentioned integrated modules may be implemented in the form of hardware or in the form of software functional modules.

[0065] In another embodiment of the present invention, a computer device is provided, the computer device includes a processor and a memory, the memory is used to store a computer program, the computer program includes program instructions, and the processor is used to execute the program instructions stored in the computer storage medium. The processor may be a central processing unit (CPU), or 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 devices, discrete hardware components, etc., which are the computing core and control core of the terminal, which are suitable for implementing one or more instructions, and are specifically suitable for loading and executing one or more instructions in the computer storage medium to implement the corresponding method flow or corresponding function; the processor described in the embodiment of the present invention can be used for the operation of a sensor life prediction method based on a fusion algorithm.

[0066] In another embodiment of the present invention, the present invention further provides a storage medium, specifically a computer-readable storage medium (Memory), which is a memory device in a computer device for storing programs and data. It is understandable that the computer-readable storage medium here can include both built-in storage media in a computer device and, of course, extended storage media supported by the computer device. The computer-readable storage medium provides a storage space, which stores the operating system of the terminal. In addition, one or more instructions suitable for being loaded and executed by a 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 here can be a high-speed RAM memory or a non-volatile memory, such as at least one disk memory. The processor can load and execute one or more instructions stored in the computer-readable storage medium to implement the corresponding steps of a sensor life prediction method based on a fusion algorithm in the above embodiment.

[0067] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0068] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0069] These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0070] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0071] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A sensor life prediction method based on fusion algorithm, characterized in that: include: Obtain the corresponding sensitivity data of the sensor under working conditions, and obtain the 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 sensor life consumption model caused by temperature is obtained by combining algorithm fusion; 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, and the sensor life is predicted based on the composite prediction model.

2. The sensor life prediction method based on fusion algorithm according to claim 1 is characterized in that: The step of obtaining the sensitivity data corresponding to the sensor under the working condition and obtaining the sensitivity reduction data caused by the voltage, current and temperature respectively includes: The magnetic resistance curve of the sensor is obtained experimentally, and the sensitivity data of the sensor under complex working conditions is obtained through linear fitting, and the abnormal data of experimental errors are removed; The sensitivity data of the sensor under complex working conditions are upgraded and reduced in dimension, all array dimensions are unified into three-dimensional arrays, and the sensitivity reduction data caused by voltage, current and temperature are marked to obtain the sensitivity reduction data caused by voltage, current and temperature.

3. The sensor life prediction method based on fusion algorithm according to claim 1 is characterized in that: The method of constructing a sensor aging model caused by voltage and current based on the sensitivity reduction data caused by voltage and current includes: According to the experimental data of sensor sensitivity decrease caused by voltage and current, the optimal 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 sensor aging caused by voltage and current.

4. The sensor life prediction method based on fusion algorithm according to claim 1 is characterized in that: The sensitivity reduction data caused by temperature is combined with an algorithm to obtain a sensor life consumption model caused by temperature, including: According to the experimental data of the decrease in sensor sensitivity caused by temperature, the long short-term memory algorithm was selected for estimation. When performing neural network processing, the learning rate parameters of the neural network were used to use algorithm fusion to construct a temperature-induced sensor life consumption model.

5. The sensor life prediction method based on fusion algorithm according to claim 4 is characterized in that: The algorithm fusion is used to construct a sensor life consumption model caused by temperature, including: Select the hyperparameter combination of learning rate, number of grid divisions, initializer, activation function, number of iterations, number of batch calculations, and loss function, and set the initial array for the hyperparameters; Through random seed sampling, we traverse the hyperparameter combinations, use all hyperparameters as the eigenvalues ​​of the reverse engineering model, and construct the model using the long short-term memory method; we use the root mean square error as the predicted value of the reverse engineering model to build the reverse engineering model; For the 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 probabilistic 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 hyperparameter combination of the number of grid divisions, initializer, activation function, number of iterations, number of batch calculations, loss function and the root mean square error are used as feature values, and the learning rate is used as the predicted value to build a model. The learning rate and other hyperparameter combination values ​​are estimated by the extreme gradient boosting tree model. The optimal hyperparameter combination of the scenario is input into the long short-term memory method for model training to obtain the sensor life consumption model caused by temperature under algorithm fusion.

6. The sensor life prediction method based on fusion algorithm according to claim 1 is characterized in that: 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, and the sensor life prediction is performed according to the composite prediction model, including: The sensor working condition data including voltage, current and temperature information is input into the composite prediction model. The composite prediction model transmits the data to the sensor aging model caused by voltage and current and the sensor life consumption model caused by temperature. The two models make predictions respectively and give the sensor life prediction results based on their respective working conditions. The composite prediction model receives the prediction results of the two models and integrates them by weighted average or linear combination. After integration, the composite prediction model outputs a comprehensive sensor life prediction result.

7. A sensor life prediction system based on fusion algorithm, characterized in that: include: The data acquisition module is used to obtain the sensitivity data of the sensor under working conditions, and obtain the sensitivity reduction data caused by voltage, current and temperature respectively after preprocessing; The model building module is used to build a model of sensor aging caused by voltage and current based on the sensitivity reduction data caused by voltage and current; based on the sensitivity reduction data caused by temperature, the sensor life consumption model caused by temperature is obtained by combining algorithm fusion; The estimation output module is used to combine the sensor life consumption model caused by temperature and the sensor aging model caused by voltage and current to obtain a composite estimation model to predict the sensor life.

8. The sensor life prediction system based on fusion algorithm according to claim 7 is characterized in that: The data acquisition module is specifically used for: The magnetic resistance curve of the sensor is obtained experimentally, and the sensitivity data of the sensor under complex working conditions is obtained through linear fitting, and the abnormal data of experimental errors are removed; The sensitivity data of the sensor under complex working conditions are upgraded and reduced in dimension, all array dimensions are unified into three-dimensional arrays, and the sensitivity reduction data caused by voltage, current and temperature are marked to obtain the sensitivity reduction data caused by voltage, current and temperature.

9. The sensor life prediction system based on fusion algorithm according to claim 7, characterized in that: The model building module is specifically used for: According to the experimental data of sensor sensitivity decrease caused by voltage and current, the optimal 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 sensor aging caused by voltage and current.

10. The sensor life prediction system based on fusion algorithm according to claim 7, characterized in that: The model building module is specifically used for: According to the experimental data of the decrease in sensor sensitivity caused by temperature, the long short-term memory algorithm was selected for estimation. When performing neural network processing, the learning rate parameters of the neural network were used to use algorithm fusion to construct a temperature-induced sensor life consumption model.

11. The sensor life prediction system based on fusion algorithm according to claim 10, characterized in that: The model building module is specifically used for: Select the hyperparameter combination of learning rate, number of grid divisions, initializer, activation function, number of iterations, number of batch calculations, and loss function, and set the initial array for the hyperparameters; Through random seed sampling, we traverse the hyperparameter combinations, use all hyperparameters as the eigenvalues ​​of the reverse engineering model, and construct the model using the long short-term memory method; we use the root mean square error as the predicted value of the reverse engineering model to build the reverse engineering model; For the 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 probabilistic 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 hyperparameter combination of the number of grid divisions, initializer, activation function, number of iterations, number of batch calculations, loss function and the root mean square error are used as feature values, and the learning rate is used as the predicted value to build a model. The learning rate and other hyperparameter combination values ​​are estimated by the extreme gradient boosting tree model. The optimal hyperparameter combination of the scenario is input into the long short-term memory method for model training to obtain the sensor life consumption model caused by temperature under algorithm fusion.

12. The sensor life prediction system based on fusion algorithm according to claim 7, characterized in that: The estimated output module is specifically used for: The sensor working condition data including voltage, current and temperature information is input into the composite prediction model. The composite prediction model transmits the data to the sensor aging model caused by voltage and current and the sensor life consumption model caused by temperature. The two models make predictions respectively and give the sensor life prediction results based on their respective working conditions. The composite prediction model receives the prediction results of the two models and integrates them by weighted average or linear combination. After integration, the composite prediction model outputs a comprehensive sensor life prediction result.

13. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the sensor life prediction method based on the fusion algorithm as described in any one of claims 1 to 6 are implemented.

14. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of a sensor life prediction method based on a fusion algorithm as claimed in any one of claims 1 to 6 are implemented.

Images