A component reliability evaluation method and system based on big data analysis

Abstract

The present invention discloses a component reliability assessment method and system based on big data analysis, which relates to the technical field of component reliability assessment, including collecting multi-source data by installing multiple sensors, and preprocessing the multi-source data; extracting features and reducing dimensions of the preprocessed multi-source data, and constructing an LSTM model to train the feature data; inputting the multi-source data collected in real time into the trained LSTM model to dynamically assess the reliability of components; introducing an early warning mechanism, performing abnormal detection and early warning according to the reliability results of components, and issuing an early warning signal; formulating an optimization strategy according to the real-time data and the early warning signal, and performing optimization adjustment. The present invention significantly improves the accuracy and real-time performance of component reliability assessment through model training and real-time assessment and early warning mechanism and dynamic optimization, and achieves continuous optimization of system performance.

Description

Technical Field

[0001] The present invention relates to the technical field of component reliability assessment, and in particular to a component reliability assessment method and system based on big data analysis. Background Art

[0002] In recent years, with the rapid development of industrial automation, the Internet of Things, and intelligent manufacturing technologies, the complexity and workload of equipment have continued to increase, and the importance of component reliability assessment has become increasingly prominent. Traditional component reliability assessment methods usually rely on regular maintenance and empirical judgment, which makes it difficult to accurately reflect the health status of components in real time. At the same time, with the increasing complexity of industrial equipment and systems and the diversification of the working environment of components, traditional reliability assessment methods are difficult to cope with the processing needs of multi-source heterogeneous data and cannot effectively capture the changes in component status under complex operating environments.

[0003] Although the existing component reliability assessment methods based on data analysis have improved the limitations of traditional methods to a certain extent, they still have many shortcomings. First, in terms of the collection and processing of multi-source data, the existing technologies generally have problems such as poor data synchronization and inconsistent sampling frequency, which makes it impossible to effectively integrate multi-source data, thus affecting the accuracy of the assessment model. Secondly, when processing complex industrial data, the existing feature extraction and dimensionality reduction methods are often difficult to accurately capture the key features related to component failures, resulting in insufficient prediction accuracy of the model. In addition, in terms of real-time evaluation and early warning mechanisms, the existing technologies mostly use simple threshold judgments, lack the ability to intelligently adjust under different working conditions, and are prone to false alarms or omissions. Summary of the invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a component reliability evaluation method and system based on big data analysis to solve the problems in the prior art of high complexity in multi-source data processing and difficulty in effectively implementing real-time reliability evaluation and dynamic optimization.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, an embodiment of the present invention provides a component reliability evaluation method based on big data analysis, which comprises:

[0008] Collect multi-source data by installing multiple sensors, and pre-process the multi-source data;

[0009] Perform feature extraction and dimensionality reduction on the preprocessed multi-source data, and build an LSTM model to train the feature data;

[0010] Input multi-source data collected in real time into the trained LSTM model to dynamically evaluate the reliability of components;

[0011] Introduce an early warning mechanism to detect and warn of abnormalities based on the reliability results of components and send out early warning signals;

[0012] Develop optimization strategies and implement optimization adjustments based on real-time data and early warning signals.

[0013] As a preferred solution of the component reliability evaluation method based on big data analysis described in the present invention, wherein: the multiple sensors include a temperature sensor, a stress sensor, a vibration sensor and a current and voltage sensor;

[0014] The multi-source data includes environmental data, electrical parameters, mechanical stress data and operational data;

[0015] The preprocessing of multi-source data includes outlier detection and removal, noise filtering, data interpolation and completion, and data normalization.

[0016] As a preferred solution of the component reliability evaluation method based on big data analysis described in the present invention, wherein: the feature extraction and dimensionality reduction processing of the pre-processed multi-source data, and the construction of the LSTM model to train the feature data include the following steps:

[0017] Extract statistical features and time series features from multi-source data to form a high-dimensional feature set;

[0018] The extracted high-dimensional feature set is input into the principal component analysis algorithm, the covariance matrix of the feature matrix is ​​calculated, the eigenvalue decomposition is performed, and the corresponding principal components are selected to form the feature set after dimensionality reduction. The expression is:

[0019]

[0020] Where C is the covariance matrix, n is the number of samples, x i represents the i-th sample in the data set, is the sample mean vector, for The transposed vector of , T represents the transposed matrix;

[0021] Input the reduced feature sequence into the LSTM model;

[0022] Use multiple layers of stacked LSTM units to update the state of each LSTM unit;

[0023] Connect the hidden state of the LSTM layer output, map it to the output dimension through the fully connected layer, and output the failure time prediction value and reliability index of the component;

[0024] Select mean square error as the loss function and use Adam optimizer to update the parameters of the LSTM model;

[0025] Divide the preprocessed multi-source data into training set and validation set;

[0026] The training set data is input into the LSTM model for multiple rounds of iterative training. After each round of training, the training loss is calculated and the LSTM model parameters are updated until the loss function converges and the preset number of training rounds is reached.

[0027] As a preferred solution of the component reliability evaluation method based on big data analysis described in the present invention, the multi-source data collected in real time is input into the trained LSTM model to dynamically evaluate the reliability of components, including the following steps:

[0028] The multi-source data collected in real time is input into the trained LSTM model. The LSTM model predicts the failure probability of the component in the next time step based on the multi-source data input in real time. Assuming that the current time is t, the failure probability P fail The calculation formula of (t) is as follows,

[0029] P fail (t) = 1-R(t)

[0030]

[0031] Among them, P fail (t) represents the failure probability of the component at time t, R(t) represents the reliability of the component at time t, τ is the integral variable, d is the differential symbol, λ(t) represents the failure rate of the component at time t, and f(t) represents the failure probability density of the component at time t;

[0032] Based on the output of the LSTM model, the health status of components is evaluated in real time, and reliability reports are output regularly;

[0033] By analyzing the changing trends of reliability indicators over time, the potential failure risks of components can be identified.

[0034] As a preferred solution of the component reliability evaluation method based on big data analysis described in the present invention, the introduction of the early warning mechanism, performing abnormality detection and early warning according to the reliability results of the components, and issuing an early warning signal, includes the following steps:

[0035] Set the failure probability threshold P warn And the failure rate warning threshold λ warn , when the real-time failure probability P is monitored failWhen the failure rate λ(t) exceeds the set threshold, an early warning signal is automatically generated;

[0036] The warning signal is transmitted to the operator through various channels. After receiving the warning signal, the operator will confirm the effectiveness of the warning at the first time and take corresponding response measures according to the warning information;

[0037] Set response time windows based on component importance and failure risk;

[0038] When the warning signal is triggered, the operator takes corresponding response measures within the specified time. If there is no response within the time window, the warning will be triggered again and emergency measures will be automatically initiated.

[0039] As a preferred solution of the component reliability evaluation method based on big data analysis described in the present invention, wherein: the optimization strategy is formulated according to the real-time data and early warning signals, and the optimization adjustment is performed, including the following steps:

[0040] Check the historical data before the warning is triggered and analyze the change process of abnormal parameters;

[0041] Through data association analysis, the relationship between abnormal parameters and other parameters is found, and the mathematical model between abnormal parameters and their influencing factors is established using multiple regression analysis;

[0042] Based on the analysis results, identify the failure mode;

[0043] Based on the cause analysis results, the LSTM model is used for simulation to predict the change in the probability of equipment failure under each strategy and evaluate the effectiveness of different optimization strategies.

[0044] According to the simulation results, select the best optimization strategy and determine the specific implementation plan of the optimization strategy;

[0045] For parameters that can be automatically controlled, perform optimization adjustments according to the optimization strategy;

[0046] For optimization measures that require human intervention, generate operation instructions and notify operators to execute them.

[0047] In a second aspect, the present invention provides a component reliability evaluation system based on big data analysis, comprising:

[0048] The data acquisition module is responsible for collecting multi-source data by installing multiple sensors;

[0049] Data preprocessing module, responsible for preprocessing data;

[0050] The feature extraction and dimensionality reduction module is responsible for feature extraction and dimensionality reduction of preprocessed multi-source data;

[0051] The model training and evaluation module is responsible for building an LSTM model to train feature data, inputting multi-source data collected in real time into the trained LSTM model, and dynamically evaluating the reliability of components;

[0052] The anomaly detection and early warning module is responsible for introducing an early warning mechanism, performing anomaly detection and early warning according to the reliability results of components, and issuing early warning signals;

[0053] The optimization strategy module is responsible for formulating optimization strategies and executing optimization adjustments based on real-time data and early warning signals.

[0054] In a third aspect, an embodiment of the present invention provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the component reliability assessment method based on big data analysis as described in the first aspect of the present invention is implemented.

[0055] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the component reliability assessment method based on big data analysis as described in the first aspect of the present invention is implemented.

[0056] The beneficial effects of the present invention are as follows: by installing multiple sensors to collect multi-source data and preprocessing the data, real-time collection of multi-source data of components is achieved, and the reliability and accuracy of the data are improved; by performing feature extraction and dimensionality reduction on the preprocessed data and building a model to train the feature data, effective processing of complex and high-dimensional data is achieved, and the prediction accuracy and stability of the model are significantly improved; through the input of real-time data and the dynamic evaluation of the model, real-time monitoring of the status of components is achieved, potential fault trends can be quickly identified, and the failure risk of components can be predicted in time; by introducing an early warning mechanism, abnormality detection and early warning are performed according to the reliability results of the components, and early prediction of potential faults of components is achieved, which effectively reduces the risks and losses caused by sudden failures; by performing optimization adjustments, the system can actively reduce the failure risk of components, extend the operating life of the equipment, improve the reliability of the equipment and reduce maintenance costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0058] Figure 1This is a flow chart of the component reliability assessment method based on big data analysis in Example 1.

[0059] Figure 2 This is a warning schematic diagram of the component reliability assessment method based on big data analysis in Example 1. DETAILED DESCRIPTION

[0060] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0061] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0062] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0063] Example 1, reference Figure 1 and Figure 2 , which is the first embodiment of the present invention, and provides a component reliability evaluation method based on big data analysis, comprising the following steps:

[0064] S1. Collect multi-source data by installing multiple sensors and pre-process the multi-source data;

[0065] Select a platinum resistance temperature sensor (such as PT1000) with high accuracy and good stability, and install it in key parts of the components, such as the chip surface and the packaging layer. Calibrate the temperature sensor in a constant temperature bath to ensure that its linearity is within the allowable error range (±0.1°C);

[0066] Use resistance strain gauges, attached to the structural weak points of components to monitor the mechanical stress during operation, and use standard weights to load the strain gauges and calibrate their strain coefficients;

[0067] Use a MEMS accelerometer (such as ADXL345) fixed on the housing of the component to monitor the vibration in real time. Install the accelerometer on a standard vibration table to perform frequency response testing and calibrate its sensitivity.

[0068] Select high-precision Hall effect sensors, install them on the circuit board, monitor the changes in operating voltage and current, and calibrate the Hall effect sensors with standard signal sources to ensure their measurement accuracy;

[0069] Collect environmental data including temperature, humidity and pressure, which directly affect the working conditions of components; collect electrical parameters including voltage and current, which directly reflect the working status of the circuit; collect mechanical stress data including structural stress and vibration, which reflect the influence of the external mechanical environment; collect operation data such as switching frequency, working mode, etc., which reflect the usage of components;

[0070] The data collected by various sensors are input into the isolation forest model, and the data points with scores exceeding the set threshold (such as 0.7) are marked and eliminated;

[0071] Applying Kalman filters to vibration and electrical data reduces noise by dynamically adjusting the error between model predictions and measurements;

[0072] Detect the null values ​​(NaN) in the data set, count the missing data of each type, and use linear interpolation to fill in the missing data. The linear interpolation method assumes that the change between data points is linear, and interpolates based on the straight line between two known data points.

[0073] All data are normalized and mapped to the interval of 0 and 1 to eliminate the dimensional differences between different physical quantities.

[0074] S2, extract features and reduce dimension of preprocessed multi-source data, build LSTM model to train feature data, including the following steps:

[0075] Extract the maximum value, minimum value, average value and standard deviation from the temperature data as temperature features, extract the peak value, root mean square value and frequency distribution features from the vibration data, and extract the mean value and maximum stress amplitude of stress;

[0076] Apply sliding window technology to time series data (such as temperature and vibration) to extract local features within each window and form a new feature set. The size of the sliding window is determined according to the sampling frequency and change rate of the data.

[0077] The statistical features and time series features extracted from the multi-source data form a high-dimensional feature set;

[0078] The extracted high-dimensional feature set is input into the principal component analysis algorithm, the covariance matrix of the feature matrix is ​​calculated, the eigenvalue decomposition is performed, and the principal components with a cumulative contribution rate of more than 90% are selected to form a feature set after dimensionality reduction. The expression is:

[0079]

[0080] Where C is the covariance matrix, n is the number of samples, x i represents the i-th sample in the data set, is the sample mean vector, for The transposed vector of , T represents the transposed matrix;

[0081] Input the reduced feature sequence into the LSTM model;

[0082] Using multiple layers of stacked LSTM units, the number of hidden units in each layer is h, and there are 3 layers in total. The state update formula of each LSTM unit is as follows:

[0083] i t =σ(W i ·[h t-1 ,x t ]+b i )

[0084] f t =σ(W f ·[h t-1 ,x t ]+b f )

[0085] o t =σ(W o ·[h t-1 ,x t ]+b o )

[0086] c t =f t ⊙c t-1 +i t ⊙tanh(W c ·[h t-1 ,x t ]+b c )

[0087] h t =o t ⊙tanh(c t )

[0088] Among them, σ is the Sigmoid activation function, ⊙ is the element-level multiplication, i t 、f t , o t are the input gate, forget gate and output gate vectors respectively, c t is the cell state, h t is the hidden state, W i ,W f ,W o ,Wc are the weight matrices of the input gate, forget gate, output gate, and candidate memory unit, respectively, and b i ,b f ,b o ,b c They are the bias vectors of the input gate, forget gate, output gate, and candidate memory unit respectively;

[0089] Connect the hidden state of the LSTM layer output, map it to the output dimension through the fully connected layer, and output the failure time prediction value and reliability index of the component;

[0090] The mean square error is selected as the loss function to measure the difference between the predicted value of the LSTM model and the true value, and the Adam optimizer is used to update the parameters of the LSTM model;

[0091] The preprocessed multi-source data is divided into a training set and a validation set in a ratio of 8:2;

[0092] Input the training set data into the LSTM model and perform multiple rounds of iterative training. After each round of training, calculate the training loss and update the LSTM model parameters until the loss function converges or reaches the preset number of training rounds (such as 100 rounds).

[0093] During the training process, an early stopping mechanism is used. When the validation set loss does not decrease within a certain number of rounds (such as 10 rounds), the training is terminated early to prevent overfitting.

[0094] After each round of training, the LSTM model is evaluated using the validation set, the validation loss is calculated, and the optimal model parameters are recorded;

[0095] The hyperparameters of the LSTM model were tuned through grid search and Bayesian optimization, and the parameter combination with the lowest validation loss was selected as the configuration of the final LSTM model.

[0096] S3, input the multi-source data collected in real time into the trained LSTM model, and dynamically evaluate the reliability of components based on the input real-time data, including the following steps:

[0097] Multi-source data collected in real time is transmitted to edge computing devices through a stable communication network for preliminary processing and caching. The cached data is regularly uploaded to the cloud server for real-time analysis by the model.

[0098] The multi-source data collected in real time is input into the trained LSTM model. The LSTM model predicts the failure probability of the component in the next time step based on the real-time input data. Assuming that the current time is t, the failure probability P fail (t), the calculation formula is as follows,

[0099] P fail(t) = 1-R(t)

[0100] Among them, P fail (t) represents the failure probability of the component at time t, R(t) is the reliability of the component at time t, that is, the probability that the component still works normally at time t;

[0101] The output of the LSTM model can be directly given as P fail (t);

[0102] The reliability R(t) can be calculated by the failure time prediction value output by the model and the current time t. The calculation formula is as follows:

[0103]

[0104] Among them, R(t) represents the reliability of the component at time t, λ(t) is the failure rate function, which is the instantaneous failure rate of the component at time t, τ is the integral variable, which represents the cumulative failure rate from time 0 to the current time t, and d is the differential symbol;

[0105] In dynamic evaluation, the LSTM model updates λ(t) in real time, calculates the updated R(t), and continuously evaluates the health status of components through the reliability function;

[0106] The failure rate λ(t) can be calculated by the relationship between the failure probability density function f(t) and the reliability R(t), the formula is,

[0107]

[0108] Among them, λ(t) represents the failure rate of components at time t, f(t) is the failure probability density function, which represents the probability density of component failure at time t;

[0109] The failure probability density function f(t) is the negative value of the derivative of the reliability function R(t) with respect to time t, expressed as,

[0110]

[0111] The mean time to failure (MTTF) is the integral of the reliability function R(t) over time, and the calculation formula is:

[0112]

[0113] Among them, MTTF is the expected life of components under normal working conditions, which means the average time from initial state to failure state;

[0114] Since R(t) is usually a nonlinear function, numerical integration methods (such as trapezoidal method and Simpson method) are used to approximate MTTF in actual calculation.

[0115] Based on the prediction results of the model, the health status of components is evaluated in real time according to the current failure probability, remaining life prediction and failure rate;

[0116] Output reports regularly, including current failure probability, failure rate, and remaining life prediction, to help maintenance personnel understand the health status of components in a timely manner and take corresponding measures;

[0117] Identify potential failure risks of components by analyzing the changing trends of reliability indicators over time;

[0118] Trend analysis methods include: moving average: smoothing time series, eliminating short-term fluctuations, and highlighting long-term trends; weighted moving average: giving higher weights to recent data and capturing trend changes more sensitively; trend decomposition: decomposing time series data into trend, seasonality, and residual parts to identify long-term trends.

[0119] S4. Introduce an early warning mechanism to perform abnormal detection and early warning based on the reliability results of components and send out early warning signals, including the following steps:

[0120] Set the failure probability threshold P warn , when the real-time failure probability P fail (t) When this threshold is exceeded, an early warning signal will be triggered;

[0121] P warn The specific value depends on the importance and safety requirements of the components. For critical components, it is recommended to set a lower threshold (such as 0.01), while for non-critical components, a slightly higher threshold (such as 0.05) can be set;

[0122] The failure rate λ(t) reflects the failure risk of components at the current moment. A failure rate warning threshold λ is set. warn , when λ(t) exceeds this threshold, an early warning signal is triggered immediately;

[0123] When the real-time failure probability P is monitored fail When the failure rate λ(t) exceeds the set threshold, an early warning signal is automatically generated, which contains the current time t, the indicator exceeding the threshold (such as P fail (t) and λ(t)) and current reliability indicators (such as remaining life prediction and failure probability curve);

[0124] The early warning signal is transmitted to the operators and related systems through a variety of channels, including: Visual interface: the early warning information is displayed in real time on the interface of the monitoring system, and the early warning status is marked with a striking color (such as red); SMS / email notification: the early warning information is sent to the mobile phone or email of the relevant personnel to ensure that the early warning information can be delivered in time; Automatic control system interface: the early warning signal can be directly transmitted to the automatic control system to trigger the corresponding emergency response measures;

[0125] After receiving the warning signal, the operator needs to confirm the effectiveness of the warning as soon as possible and take corresponding response measures according to the warning information. Common response measures include reducing the workload: reducing the operating voltage or current of components to slow down their aging rate; adjusting the operation mode: switching to the backup operation mode or reducing the operation frequency to reduce the pressure on components; planned shutdown and maintenance: arranging shutdown and maintenance without affecting the operation of the overall system, and replacing components that may fail in time;

[0126] Set response time windows based on component importance and failure risk;

[0127] When the warning signal is triggered, the operator must take corresponding response measures within the specified time (such as 10 minutes or 1 hour). If there is no response within the time window, the warning will be triggered again and emergency measures will be automatically initiated.

[0128] S5. Develop optimization strategies based on real-time data and early warning signals, and implement optimization adjustments, including the following steps:

[0129] Check the historical data before the warning is triggered and analyze the change process of abnormal parameters. For example, analyze the temperature change trend over the past period of time to find out the root cause of the temperature increase;

[0130] Through data association analysis, find out the relationship between abnormal parameters and other parameters. For example, through correlation analysis, it is found that the temperature increase may be related to the increase of equipment load. Using multiple regression analysis, a mathematical model between abnormal parameters and their influencing factors is established to find out the main influencing factors.

[0131] Based on the analysis results, the failure mode is identified. For example, if a specific failure frequency appears in the vibration spectrum, it indicates that the equipment may have a bearing damage problem.

[0132] Based on the cause analysis results, the effectiveness of different optimization strategies is evaluated. For example, the impact of temperature optimization, load optimization, and stress optimization on the probability of equipment failure is evaluated. The evaluation of optimization strategies can be simulated through the LSTM model to predict the change in the probability of equipment failure under each strategy.

[0133] Based on the simulation results, select the best optimization strategy. For example, if the LSTM model predicts that temperature optimization can significantly reduce the probability of failure, then select the temperature optimization strategy.

[0134] For parameters that can be automatically controlled, automatically adjust relevant operating parameters according to the optimization strategy. For example, automatically reduce the cooling temperature setting value by 2°C, or reduce the equipment load by 10%. The adjustment process is automatically executed by the PLC or SCADA system to ensure the accuracy of the adjustment; monitor the adjusted equipment status in real time, especially the key parameters related to the optimization strategy, for example, monitor the temperature change after temperature optimization to ensure that the temperature returns to the safe range, and observe whether the vibration and load of the equipment are stable;

[0135] For optimization measures that require human intervention, detailed operation steps are generated according to the optimization strategy. For example, if the installation position of the equipment needs to be adjusted to reduce mechanical stress, a specific operation guide is generated to explain how to adjust, the adjustment angle and the required tools. It also includes safety precautions to ensure that operators avoid risks during the adjustment process. Operators perform adjustments according to the generated instructions, such as reinstalling equipment or replacing equipment components. During the adjustment process, operators can record the adjustment progress and problems encountered, and provide feedback on the final adjustment results.

[0136] This embodiment also provides a component reliability evaluation system based on big data analysis, including:

[0137] The data acquisition module is responsible for collecting multi-source data by installing multiple sensors; the data preprocessing module is responsible for preprocessing the data; the feature extraction and dimensionality reduction module is responsible for feature extraction and dimensionality reduction of the preprocessed multi-source data; the model training and evaluation module is responsible for building an LSTM model to train the feature data, inputting the real-time collected multi-source data into the trained LSTM model, and dynamically evaluating the reliability of components; the anomaly detection and early warning module is responsible for introducing the early warning mechanism, performing anomaly detection and early warning according to the reliability results of components, and issuing early warning signals; the optimization strategy module is responsible for formulating optimization strategies based on real-time data and early warning signals, and performing optimization adjustments.

[0138] This embodiment also provides a computer device, which is suitable for the component reliability assessment method based on big data analysis, including: a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the component reliability assessment method based on big data analysis proposed in the above embodiment.

[0139] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0140] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the component reliability assessment method based on big data analysis proposed in the above embodiment is implemented; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0141] In summary, the present invention ensures the quality and consistency of the data by installing multiple sensors to collect multi-source data and preprocessing the data, thereby improving the accuracy and reliability of subsequent data analysis; extracts features and performs dimensionality reduction on the preprocessed data, and constructs a model to train the feature data, thereby achieving effective processing of complex and high-dimensional data and significantly improving the prediction accuracy and stability of the model; and realizes real-time monitoring of the status of components through the input of real-time data and the dynamic evaluation of the model. It is able to quickly identify potential fault trends and timely predict the failure risks of components; and by introducing an early warning mechanism, performs abnormality detection and early warning based on the reliability results of components, thereby achieving early prediction of potential component failures, which not only improves the safety of equipment operation, but also effectively reduces the risks and losses caused by sudden failures; and by performing optimization adjustments, the system can actively reduce the failure risks of components, extend the operating life of the equipment, improve the reliability of the equipment, and reduce maintenance costs.

[0142] Example 2, referring to Table 1, is the second example of the present invention. To further verify the technical solution of the present invention, experimental simulation data of a component reliability evaluation method based on big data analysis is provided.

[0143] This embodiment aims to verify the effectiveness and advantages of a component reliability assessment method based on big data analysis. The key components in a certain model of industrial equipment were selected as the research object. During the experiment, temperature sensors, stress sensors, vibration sensors, and current and voltage sensors were installed to collect multi-source data. These sensors are used to collect environmental data (such as temperature and humidity), electrical parameters (such as current and voltage), mechanical stress data (such as stress amplitude and frequency), and operating data (such as operating cycle and load conditions). In the data acquisition stage, the reliability and consistency of the data are guaranteed by performing outlier detection and removal, noise filtering, data interpolation and completion, and data normalization on the data obtained by each sensor.

[0144] After data preprocessing, principal component analysis (PCA) was used to reduce the dimensionality of the extracted high-dimensional feature data, retaining more than 90% of the feature contribution rate and significantly reducing the data dimension. Then, the reduced feature set was input into the LSTM (Long Short-Term Memory) model for training. After multiple rounds of iterative training, the LSTM model finally obtained a high-precision component failure prediction model through the continuous decline of the validation set loss. During the training process, the early stopping mechanism and Bayesian optimization method were used to fine-tune the model's hyperparameters to ensure the generalization ability and prediction accuracy of the model.

[0145] During the real-time data collection stage, the data is transmitted to the edge computing device through a stable communication network for preliminary processing and uploaded to the cloud server regularly. The LSTM model dynamically evaluates the failure probability and reliability of components based on real-time input data. When the failure probability or failure rate exceeds the set threshold, the system automatically triggers a warning signal. Subsequently, the historical data that triggered the warning is analyzed based on the warning signal, possible failure modes are identified, and changes in the probability of equipment failure under different optimization strategies are simulated, and finally the optimal optimization adjustment strategy is selected. After the optimization is implemented, the optimized data is input into the model for incremental training through the data feedback mechanism, and the model parameters are continuously updated to ensure the adaptability and accuracy of the model.

[0146] The details are shown in Table 1 below:

[0147] Table 1 Component reliability evaluation test data table

[0148] Test subjects Temperature(℃) Stress(MPa) Vibration(Hz) Voltage (V) Failure probability (%) Initial state (existing technology) 45 120 60 220 5 Before optimization (existing technology) 50 130 65 225 15 Before optimization (the present invention) 50 130 65 225 10 After optimization (the present invention) 48 125 62 223 3

[0149] It can be intuitively seen from the test data table that the present invention has obvious advantages over the prior art. First, before optimization, the failure probability of the components using the prior art is as high as 15% under the conditions of temperature of 50°C, stress of 130MPa, vibration frequency of 65Hz, and voltage of 225V. Under the same conditions, the method of the present invention can reduce the failure probability to 10% through more accurate feature extraction and dimensionality reduction processing, as well as an optimized LSTM model, which shows that the present invention has higher accuracy and robustness in reliability assessment.

[0150] Further analysis of the optimized data shows that the present invention reduces the failure probability from 10% before optimization to 3% through real-time data input and continuous model updating. This result is due to the early warning mechanism introduced by the present invention and the optimization strategy based on big data analysis. By capturing abnormal changes in real-time data and dynamically adjusting the operating parameters of components, the failure risk of components is significantly reduced. In addition, the continuous update mechanism of the model ensures the long-term adaptability and accuracy of the model, avoiding the degradation of model performance due to changes in the operating status of the equipment.

[0151] 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 preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A component reliability assessment method based on big data analysis, characterized in that: include, Collect multi-source data by installing multiple sensors and pre-process the multi-source data; the multi-source data includes environmental data, electrical parameters, mechanical stress data and operation data; Perform feature extraction and dimensionality reduction on the preprocessed multi-source data, and build an LSTM model to train the feature data; Input multi-source data collected in real time into the trained LSTM model to dynamically evaluate the reliability of components; Introduce an early warning mechanism to detect and warn of abnormalities based on the reliability results of components and send out early warning signals; Formulate optimization strategies and implement optimization adjustments based on real-time data and early warning signals; The method of inputting the real-time collected multi-source data into the trained LSTM model to dynamically evaluate the reliability of components includes the following steps: The multi-source data collected in real time is input into the trained LSTM model. The LSTM model predicts the failure probability of the component in the next time step based on the multi-source data input in real time. Based on the output of the LSTM model, the health status of components is evaluated in real time, and reliability reports are output regularly; Identify potential failure risks of components by analyzing the changing trends of reliability indicators over time; The introduction of the early warning mechanism, performing abnormal detection and early warning according to the reliability results of components and sending out early warning signals, includes the following steps: Set the failure probability threshold P warn And the failure rate warning threshold λ warn , when the real-time failure probability P is monitored fail When the failure rate λ(t) exceeds the set threshold, an early warning signal is automatically generated; The warning signal is transmitted to the operator through various channels. After receiving the warning signal, the operator will confirm the effectiveness of the warning at the first time and take corresponding response measures according to the warning information; Set response time windows based on component importance and failure risk; When the warning signal is triggered, the operator takes corresponding response measures within the specified time. If there is no response within the time window, the warning will be triggered again and emergency measures will be automatically initiated; The output of the LSTM model can be directly given as P fail (t), P fail (t)=1-R(t) Among them, P fail (t) represents the failure probability of the component at time t, R(t) represents the reliability of the component at time t, τ is the integral variable, d is the differential symbol, λ(t) represents the failure rate of the component at time t, f(t) represents the failure probability density of the component at time t, and MTTF is the expected life of the component under normal working conditions; Based on the prediction results of the model, the health status of components is evaluated in real time according to the current failure probability, remaining life prediction and failure rate.

2. The component reliability assessment method based on big data analysis according to claim 1, characterized in that: The plurality of sensors include a temperature sensor, a stress sensor, a vibration sensor, and a current and voltage sensor; The preprocessing of multi-source data includes outlier detection and removal, noise filtering, data interpolation and completion, and data normalization.

3. The component reliability assessment method based on big data analysis according to claim 2, characterized in that: The method of extracting features and reducing dimension of preprocessed multi-source data and building an LSTM model to train the feature data includes the following steps: Extract statistical features and time series features from multi-source data to form a high-dimensional feature set; The extracted high-dimensional feature set is input into the principal component analysis algorithm, the covariance matrix of the feature matrix is ​​calculated, the eigenvalue decomposition is performed, and the corresponding principal components are selected to form the feature set after dimensionality reduction. The expression is: Where C is the covariance matrix, n is the number of samples, x i represents the i-th sample in the data set, is the sample mean vector, for The transposed vector of , T represents the transposed matrix; The reduced feature set is input into the LSTM model, and the state of each LSTM unit is updated using multiple layers of stacked LSTM units. Connect the hidden state of the LSTM layer output, map it to the output dimension through the fully connected layer, and output the failure time prediction value and reliability index of the component; Select mean square error as the loss function and use Adam optimizer to update the parameters of the LSTM model; Divide the preprocessed multi-source data into training set and validation set; The training set data is input into the LSTM model for multiple rounds of iterative training. After each round of training, the training loss is calculated and the LSTM model parameters are updated until the loss function converges and the preset number of training rounds is reached.

4. The component reliability assessment method based on big data analysis according to claim 3, characterized in that: The optimization strategy is formulated based on real-time data and early warning signals, and the optimization adjustment is performed, including the following steps: Check the historical data before the warning is triggered and analyze the change process of abnormal parameters; Through data association analysis, the relationship between abnormal parameters and other parameters is found, and the mathematical model between abnormal parameters and their influencing factors is established using multiple regression analysis; Based on the analysis results, identify the failure mode; Based on the cause analysis results, the LSTM model is used for simulation to predict the change in the probability of equipment failure under each strategy and evaluate the effectiveness of different optimization strategies. According to the simulation results, select the best optimization strategy and determine the specific implementation plan of the optimization strategy; For parameters that can be automatically controlled, perform optimization adjustments according to the optimization strategy; For optimization measures that require human intervention, generate operation instructions and notify operators to execute them.

5. A component reliability evaluation system based on big data analysis, based on the component reliability evaluation method based on big data analysis according to any one of claims 1 to 4, characterized in that: include, The data acquisition module is responsible for collecting multi-source data by installing multiple sensors; Data preprocessing module, responsible for preprocessing data; The feature extraction and dimensionality reduction module is responsible for feature extraction and dimensionality reduction of preprocessed multi-source data; The model training and evaluation module is responsible for building an LSTM model to train feature data, inputting multi-source data collected in real time into the trained LSTM model, and dynamically evaluating the reliability of components; The anomaly detection and early warning module is responsible for introducing an early warning mechanism, performing anomaly detection and early warning according to the reliability results of components, and issuing early warning signals; The optimization strategy module is responsible for formulating optimization strategies and executing optimization adjustments based on real-time data and early warning signals.

Images