Test equipment life analysis method and system based on big data

Abstract

The present invention relates to the field of data processing, and in particular to a test equipment life analysis method and system based on big data. The method comprises: collecting operating equipment data, and setting a search range and a window range for each data point, calculating the similarity between each data point and a reference point according to the difference between the data point in the window range and the data point in the window range of a reference point, and further calculating the final similarity between the data point and the corresponding reference point according to the difference between the data point in the window range of the data point and the data point in the window range of the corresponding reference point, using the final similarity as the filtering weight of the data point, performing non-local mean filtering on all data, and predicting the life of the equipment according to the filtered data, thereby improving the accuracy of the filtering result and being able to accurately predict the life of the test equipment.

Description

Technical Field

[0001] The present invention relates to the field of data processing, and in particular to a test equipment life analysis method and system based on big data. Background Art

[0002] The life analysis of test equipment is a complex and critical task. It not only requires the evaluation of the durability and reliability of the equipment under specific working conditions, but also requires an in-depth understanding of its performance degradation process. At present, life prediction is mostly based on the state monitoring method, that is, by real-time tracking of the parameters of the equipment in operation, and comparing with the known standard or reference equipment data, the life status of the current equipment is estimated. For example, the patent document with the authorization announcement number CN103645052B discloses a remote online state monitoring and life assessment method for the gearbox of a wind turbine. The real-time operating parameters, temperature data and vibration data of each wind turbine are collected in real time; by analyzing the characteristic values ​​of the signal in the time domain and frequency domain, the vibration sensor is diagnosed to have a fault; when diagnosing the fault of the gearbox of a wind turbine, different filtering methods are used for different algorithms to extract the corresponding features; the fault diagnosis will give the presence or absence of the fault and the degree of the fault; the residual life of the faulty gearbox is estimated by the gray theory model, and an autoregressive model is established for the residual of its prediction result to improve the prediction accuracy and realize the real-time estimation of the residual life.

[0003] In the process of life analysis of test equipment, high-quality monitoring data is the cornerstone of prediction accuracy, but due to the environmental factors faced by the equipment, such as electromagnetic interference and vibration, data noise will be introduced, which will reduce the data quality. The long-term operation of the equipment is also accompanied by aging and wear, which will cause performance degradation and parameter fluctuation, further increasing the uncertainty of the data. Therefore, the denoising process before prediction is crucial to purify the data and ensure the accuracy of the prediction. Another example is the patent application document with publication number CN117874445A, which discloses a real-time online monitoring data analysis enzyme production monitoring method, including: collecting temperature data at different positions of a fermentation tank, by the search window size in the adaptive non-local mean filtering algorithm, improving the effect of data denoising, obtaining accurate and reliable temperature data, thereby improving the effect of enzyme production monitoring.

[0004] The non-local mean filtering algorithm adopted in the above scheme is based on the local similarity of data for denoising, which can effectively retain the trend of data changes. However, due to its high reliance on the local similarity of data, the filtering process is easily affected by local outliers or specific fluctuations in the data, which may lead to low accuracy of the filtering results, affect the quality of the collected data, and thus reduce the accuracy of the life analysis results of the test equipment. Summary of the invention

[0005] In order to solve the problem that when using the non-local mean filtering algorithm to denoise parameter data, it is easily affected by local outliers or specific fluctuations in the data, which reduces the accuracy of the life analysis results of the test equipment, the present invention proposes a test equipment life analysis method and system based on big data. The specific technical scheme is as follows:

[0006] In one aspect, the present invention provides a test equipment life analysis method based on big data, comprising:

[0007] Collecting running equipment data, screening data points, and setting a search range and a window range for the data points, using other data points within the search range as reference points for the data points, and using data points corresponding to the reference points in pre-collected reference equipment data as corresponding reference points; the window range is smaller than the search range;

[0008] Calculate the similarity between the data point and the reference point based on the difference between the window range of the data point and the window range of the reference point, and calculate the similarity between the data point and the corresponding reference point based on the difference between the window range of the data point and the window range of the corresponding reference point, wherein the similarity is negatively correlated with the difference;

[0009] Based on the similarity between the data point and the reference point, and the similarity between the data point and the corresponding reference point, the final similarity between the data point and the reference point is calculated, the final similarity is used as the filtering weight of the data point, all data are subjected to non-local mean filtering, and the life of the equipment is predicted based on the filtered data.

[0010] The above scheme significantly improves the accuracy and reliability of the test equipment life prediction through a sophisticated data processing process. First, by collecting and screening comprehensive data of the operating equipment and the reference equipment, a solid foundation is laid for the analysis. The set data point search range and window strategy allow the difference between the data point and the reference point to be effectively identified in the local similarity calculation, and the similarity between the data point and the reference point is obtained based on the difference. In addition, the comprehensiveness of the reference equipment data is combined to provide a better comparison benchmark, evaluate the similarity between the data point and the corresponding reference point in the reference equipment, enhance the breadth and generalization of the data point similarity evaluation, overcome the limitations of a single device data, and obtain the final similarity between the data point and the reference point. On this basis, non-local mean filtering is performed with the final similarity as the weight, achieving more accurate denoising, obtaining accurate filtering result data, and accurately predicting the life of the test equipment based on the accurate filtering results. This series of steps jointly ensures the efficiency and reliability of equipment life prediction.

[0011] Preferably, the final similarity between the data point and the reference point satisfies the following relationship:

[0012]

[0013] In the formula, For the data point and its The final similarity of the reference points, For the data point and its The similarity of the reference points, For the The value of the data point, For the The value of the reference point, For the Data points and The similarity of corresponding reference points, For the The value corresponding to the reference point, is an absolute value symbol; the reference point corresponds one-to-one to the corresponding reference point.

[0014] The above scheme comprehensively considers the similarities between the target data points and multiple reference points in different dimensions (including the corresponding points with the direct reference points and the reference device data set), and provides a more comprehensive and detailed similarity measurement in a weighted average manner, thereby improving the accuracy and specificity of the denoising process.

[0015] Preferably, the similarity between the data point and the corresponding reference point satisfies the following relationship:

[0016] ;

[0017] In the formula, For the said Data points and The similarity of corresponding reference points, For the The local current difference of each data point, For the The local current difference corresponding to the reference point, For the The value of the data point, For the The method for obtaining the local current difference of the corresponding reference point is the same as the method for obtaining the local current difference of the data point.

[0018] The above scheme integrates the similarity metrics of current difference and numerical difference. This method effectively improves the accuracy of data analysis with large changes in electrical characteristics and physical behaviors, especially in the links where current fluctuations are significant, ensuring the accuracy of similarity calculation.

[0019] Preferably, predicting the life of the equipment based on the filtered data includes:

[0020] By comparing the filtered data of the operating device with the data of the reference device, the stage where the operating conditions of the two are similar is determined, and the life of the operating device is predicted based on the remaining life of the reference device in the stage of similar operating conditions.

[0021] Preferably, the method for obtaining the local current difference of the data point is:

[0022] The collected currents corresponding to all data points within the window range of the data points are obtained as the local current sequence of the data points, and the mean of the first-order differences of the local current sequence is used as the local current difference of the data points.

[0023] Preferably, the similarity between the data point and the reference point satisfies the following relationship:

[0024] ;

[0025] In the formula, For the data point and its The similarity of the reference points, For the said The noise performance of the reference point, For the The window of data points The value of the data point, For the The reference point within the window The value of the data point, For the The window of data points The collection time of data points, For the The reference point within the window The collection time of data points, is the natural exponential function, is the length of the window range, the height of the window range is 1, is the absolute value symbol.

[0026] The above solution effectively improves the sophistication and accuracy of similarity evaluation by introducing detailed comparison and time alignment of all data points in the window.

[0027] Preferably, setting the search range and window range of the data points includes:

[0028] Centered on the data point, set the length to The window is used as the search range of the data point, with the data point as the center and the length is set to , as the window range of the data points, where and are all preset values ​​and are odd numbers. .

[0029] Preferably, the method for screening data points is:

[0030] Among all the data points, calculate the noise performance of each data point:

[0031] ;

[0032] In the formula, For the The noise level of each data point is For the The value of the data point, For the The value of the data point, For the The value of the data point, is the linear normalization function, is the absolute value symbol;

[0033] A noise performance level threshold is preset, and data points having a noise performance level greater than the noise performance level threshold are screened out.

[0034] The above scheme quantifies the degree of noise performance of each data point by calculating the difference between the adjacent data points before and after each data point, uses normalization to ensure the uniformity of the evaluation, sets a threshold to remove high-noise data points, effectively filters out abnormal interference, improves the purity of the data set, and provides a reliable data basis for subsequent similarity analysis and life prediction.

[0035] On the other hand, the present invention also provides a test equipment life analysis system based on big data, the analysis system includes a memory and a processor, the memory stores a computer program, and the processor executes the computer program to implement the steps described in the above method.

[0036] The present invention has the following effects:

[0037] The present invention significantly enhances the local similarity evaluation of data points in the operating parameters of experimental equipment through an innovative comprehensive analysis method, thereby optimizing the filtering process and ensuring the high accuracy of the life prediction of the operating equipment. This method breaks through the similarity of the local performance of the current equipment data points in its own parameters and the corresponding parameters of the reference equipment, constructs a more comprehensive similarity evaluation system, effectively suppresses the random noise introduced by sensor failure and external interference, and improves the purity and signal-to-noise ratio of the filter. In addition, compared with the local similarity filtering that relies only on a single equipment parameter, this strategy compensates for the shortcomings of filtering under specific working conditions by incorporating the complementary information of the reference equipment, improves the accuracy and generalization ability of the filtering, and makes the equipment life prediction more reliable. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0039] Figure 1 It is a schematic flow chart of the method of the present invention;

[0040] Figure 2 It is a schematic diagram of reference points and corresponding reference points of the present invention. DETAILED DESCRIPTION

[0041] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0042] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0043] Reference Figure 1 , a test equipment life analysis method and system based on big data, including steps S1-S5, as follows:

[0044] S1: Collect operating equipment data and reference equipment data.

[0045] To accurately analyze the life of equipment, the data collected from the running equipment must fully cover all indicators that affect its performance, health status and potential failures, including but not limited to the following key data types:

[0046] Temperature data: the temperature of key parts of the equipment during operation, reflecting the load, thermal management efficiency and material aging; vibration data: the vibration level of the component, indicating imbalance, asymmetry, wear or loose balance state; current and voltage data: the current and voltage of electrical equipment such as motors, drives, power supplies, etc., to evaluate efficiency and load; pressure data: for hydraulic and pneumatic systems, pressure data reflects working conditions and seals and leaks; speed and working cycle: operating speed, start-up and shutdown cycles, thermal cycle stress analysis; wear residue: wear amount, clearance, thickness of key components, such as bearings and gears; environmental data: temperature and humidity, dust level, corrosive environmental conditions; maintenance history, repair frequency, failure history. Through comprehensive analysis of these data, combined with equipment characteristics, operating environment conditions, and historical failure modes, a life prediction model can be established, and preventive maintenance measures can be taken to extend the service life of the equipment.

[0047] It can be seen that there are many parameters that need to be collected during the operation of the test equipment, but this embodiment only takes temperature data as an example for explanation, and the remaining parameter data can also be processed in the same way. Specifically, this step takes temperature data as an example, and collects temperature data during the operation of the test equipment by installing a sensor, with a collection frequency of once per second, and selects the temperature data of the reference equipment from the first use to the time when it cannot be used in the database of the equipment operation, that is, the reference equipment needs to have a complete service life, and the life analysis of the currently running test equipment is performed once a day in units of days.

[0048] Temperature data also plays an important role in equipment life prediction because temperature has a direct and significant impact on equipment performance and material degradation. Although other data are also involved in equipment status monitoring, the direct relevance of temperature, its reaction to material and performance degradation, and its easy monitoring characteristics make it an indispensable indicator in life prediction. The analysis of temperature data combined with other comprehensive parameters can provide a comprehensive life prediction perspective and optimize maintenance strategies. Almost all mechanical equipment and electronic equipment can be used to predict life based on temperature data during operation, as long as heat energy is involved in heat exchange and temperature rise, because temperature directly affects the physical and chemical properties of metals and insulating materials, such as strength, hardness, ductility, creep embrittlement, etc. , excessive temperature accelerates aging and is the key factor of failure; temperature changes cause thermal expansion and contraction, generate internal stress, and material fatigue under long-term action, leading to cracks, deformation, loose layers, and shortened life; temperature changes periodically during equipment operation, such as start-up and shutdown, load, form thermal cycles, accelerate thermal fatigue, and affect life; the design of the equipment takes into account the ambient temperature range, and the out-of-range working temperature warning directly reflects the environmental adaptability and predicts the life under extreme conditions; monitoring convenience: temperature sensors are popular, easy to install, real-time monitoring, good data continuity, easy to obtain, conducive to long-term analysis, and convenient for model building. Abnormal temperature rise is often a precursor to failure and an early indicator. It reacts earlier than other data, intervenes in time, avoids losses, and has low maintenance costs.

[0049] Here are some common, typical categories of equipment whose lifetime is closely tied to temperature and can be predicted by monitoring temperature:

[0050] Motors and drive equipment, including: motors, generators, transformers, operating temperature reflects load, efficiency, overheating management, high temperature accelerates aging of insulation materials. Mechanical equipment, including: bearings, gearboxes, pumps, compressors, transmission systems, temperature reflects lubrication conditions, friction, wear, overheating causes parts damage. Electronic equipment, including: servers, routers, data centers, power supplies, thermal management is key, high temperature and poor heat dissipation, chip, circuit degradation, reduced life. Heating and cooling systems, including: boilers, air conditioners, heat pumps, heat exchangers, temperature directly reflects efficiency, thermal fluid performance, overheating affects material durability. Energy equipment, including: solar panels, fuel cells, temperature affects efficiency, overheating management, heat island effect, life prediction. Chemical and refining equipment, including: reactors, pipelines, furnaces. Automotive parts, including engines, brake systems, batteries, thermal management, temperature reflects efficiency, loss, heat accumulation affects life.

[0051] By using temperature data, combined with the equipment's operation, environmental conditions, historical fault records, and maintenance cycles, and using statistical models, machine learning and other technologies, we can more accurately predict the remaining life of these devices, optimize maintenance plans, reduce failure rates, and extend the service life.

[0052] S2: Calculate the similarity between each data point in the operating equipment data and each reference point in its search range.

[0053] During the operation of the experimental equipment, it is susceptible to factors such as electromagnetic interference and equipment aging, which leads to noise in the collected data and affects the accuracy of the life analysis. Therefore, the non-local mean filtering technology can be used to filter and denoise the collected data to eliminate noise interference and ensure the accuracy of the life analysis results. The non-local mean filtering is based on the local similarity of the data, so this step requires analyzing the similarity between the data points first.

[0054] In summary, the optimization strategy adopted in this embodiment is: screening data points, using the local similarity of data points in the running equipment and the corresponding similarity in the reference equipment parameters for comprehensive evaluation, accurately determining the final similarity of the data points, and performing targeted filtering based on the final similarity, thereby purifying the data, providing high-quality life analysis for the experimental equipment, and ensuring the accuracy of the prediction.

[0055] Specifically, the following steps are included:

[0056] S21: based on the noise performance of the collected data points, screen out the data points for subsequent processing, and the screened out data points are the data points that need to be filtered in a targeted manner.

[0057] Considering that filtering all data one by one is time-consuming, and because the noise characteristics of each data are different and the filtering requirements are different, it is critical to prioritize the data for targeted filtering. In the collected data, taking any data point as an example, the noise performance of the data point is reflected by the cumulative sum of the data change difference between the data point and the adjacent data points before and after. For example, the data point Noise performance of data points It is expressed as:

[0058]

[0059] In the formula, For the The value of the data point, i.e. The data points collected at all times, For the The value of the data point, For the The value of the data point, is a linear normalization function to normalize the result to the range of 0-1, is the absolute value symbol.

[0060] Preset noise level threshold , which is an empirical value. Data points with a noise performance level greater than the noise performance level threshold are screened out. The screened out data points have a large noise performance and need to be subsequently filtered. Data with a noise performance level less than or equal to the noise performance level threshold will not be subsequently filtered.

[0061] S22: Set the search range and window range for each data point, select any data point, take the data point as the center, and set the length The window is used as the search range for the data point, with the data point as the center and the length is set to , as the window range of the data point, where and are all preset values ​​and are odd numbers. .

[0062] In this embodiment, it is set is 15, is 5, and the height of the window is 1, that is, each data point and its left and right 7 data points, a total of 15 data points are used as the search range of the data point, and each data point and its left and right 2 data points, a total of 5 data points are used as the window range of the data point. According to this method, the search range and window range of each data point are obtained.

[0063] S23: Calculate the similarity between each data point in the operating device data and each reference point within its search range.

[0064] Taking any data point as an example, each data point within the search range of the data point is used as a reference point of the data point, and then the similarity between the data point in the window range of the data point and the data point in the window range of the reference point is calculated, and the difference is negatively correlated with the similarity. The difference includes a numerical difference and a collection time difference, as shown in the first The similarity between a data point and the rth reference point in the search range is calculated according to the following formula:

[0065]

[0066] In the formula, For the data point and its The similarity of the reference points, For the said The noise performance of the reference point, For the The window of data points The value of the data point, For the The reference point within the window The value of the data point, For the The window of data points The collection time of data points, For the The reference point within the window The collection time of data points, is the natural exponential function, is the length of the window range, is the absolute value symbol.

[0067] The similarity is achieved by obtaining the window range of the current data point and the window range of the reference point, comparing the numerical differences between the data points in the two window ranges, as well as the differences in the collection times of the data points, and accumulating the differences to calculate the mean. The smaller the mean, the smaller the difference between the current data point and the reference point, that is, the higher the similarity between the current data point and the reference point.

[0068] like Figure 2 As shown in is the data point in the running equipment data, for A reference point, this step is to calculate and The similarities between.

[0069] S3: Calculate the similarity between each data point in the operating equipment data and the corresponding reference point.

[0070] The definition of the corresponding reference point is: in the currently collected running equipment data, the reference point of each data point in the previously collected reference equipment data is used as the corresponding reference point, and the correspondence is based on the collection time, such as Figure 2 As shown in for The corresponding reference point of and The similarities between.

[0071] In the process of life analysis of test equipment, if we only rely on the data collected during the operation of the test equipment itself, perform local similarity analysis, and filter based on the local similarity of the data, it may amplify the impact of accidental noise and cause errors in the filtering results. Therefore, this embodiment combines the operating data of the current test equipment with the reference equipment data to provide a more robust reference benchmark, effectively improve the ability to suppress outliers and noise, enhance the stability of filtering, and obtain accurate filtering results.

[0072] The similarity between each data point in the running device data and the corresponding reference point is specifically calculated based on the difference between the data point in the window range of the data point and the data point in the window range of the corresponding reference point, and the calculation formula is:

[0073]

[0074] In the formula, since the reference point and the corresponding reference point correspond one to one, then Represents the serial number of the reference point, and also represents the serial number of the corresponding reference point. For the said Data points and The similarity of corresponding reference points, For the The local current difference of each data point, For the The local current difference corresponding to the reference point, For the The value of the data point, For the The method for obtaining the local current difference corresponding to the reference point is the same as the method for obtaining the local current difference corresponding to the data point.

[0075] Wherein, the method for obtaining the local current difference of the data point is:

[0076] The collected currents corresponding to all data points within the window range of the data point are obtained as the local current sequence of the data point, and the mean of the first-order difference of the local current sequence is taken as the local current difference of the data point, for example , satisfying the following relationship:

[0077]

[0078] In the formula, For the The local current difference of each data point, For the The total number of data points within a window of data points, is the serial number of the data point within the window range, For the The local range of the data point The collected current corresponding to each data point (the current of the device corresponding to the collection time of the data point), is the local range of the tth data point The collected current corresponding to each data point.

[0079] During the analysis process, taking any data point in the operating data of the current device as an example, the corresponding data point is identified in the reference device data, and a one-to-one correspondence is made based on the acquisition time. Then, the local current difference of each data point in the operating data of the current device and the local current difference of the reference point of the data point and the corresponding reference point in the data of the reference device are obtained. Based on the similarity between the data point and the reference point, although the data point of the operating data of the current device and the data point of the reference device data come from different devices, the comparison of current parameters is universal, especially in situations where the equipment types and specifications are similar, the current change trends can be effectively compared with each other. Therefore, the current change is used as a measurement standard to evaluate the similarity between the data points in the current operating device data and the data points in the reference device data.

[0080] In the specific implementation, first, taking a data point in the current device operation data as an example, the adjacent data points of the data point within the window range are analyzed, and the current changes corresponding to these data points within the window range are recorded, which are used as the local current difference of the data point within the window range. A similar strategy is used to calculate the local current change value corresponding to a reference point of the data point in the reference device.

[0081] The local current difference of the data point in the currently operating device and the local current difference of the reference point of the data point in the reference device, as well as the difference between the value of the data point in the currently operating device and the value of the reference point of the data point in the reference device are obtained.

[0082] If the difference between the two local currents is smaller and the difference between the two data is smaller, the data point and the reference point have high similarity. This strategy effectively improves the accuracy of the similarity evaluation of data points between different devices, thereby optimizing the accuracy of the filtering analysis.

[0083] S4: Determine the final similarity of each data point in the operating device data with each reference point within its search range.

[0084] When analyzing each data point in the currently running device data, the weighted similarity between the data point and the reference point, as well as the weighted similarity between the data point and the corresponding reference point in the reference device (where the reference point is located) is considered to synthesize the similarity between the data point and the reference point. The weight is determined based on the numerical difference. The larger the difference, the lower the similarity, and the weight decreases accordingly, ensuring the accuracy of the evaluation.

[0085] The final similarity between the data point and the reference point is calculated based on the similarity between the data point and the reference point and the similarity between the data point and the corresponding reference point. The calculation formula is:

[0086]

[0087] In the formula, For the data point and its The final similarity of the reference points, For the data point and its The similarity of the reference points, For the The value of the data point, For the The value of the reference point, For the Data points and The similarity of corresponding reference points, is the value of the rth corresponding reference point, is an absolute value symbol. The reference point and the corresponding reference point have a one-to-one correspondence, so it can be used It serves as the serial number of the reference point and can also serve as the serial number of the corresponding reference point.

[0088] S5: Using the final similarity as the filtering weight of the data point, performing non-local mean filtering on all the data, and predicting the life of the equipment based on the filtered data.

[0089] In order to accurately predict the remaining life of the experimental equipment, the following optimization steps are taken for data processing and analysis:

[0090] First, the final similarity between the data point and the reference point is used as the weight basis of the data point in the non-local mean filtering process, and the non-local mean filtering operation is performed on all the data points (screened out) to obtain the denoised data points;

[0091] Then, using the filtered data, we enter the life prediction stage. By observing the performance of the filtered data of the operating equipment and the historical reference equipment under similar working conditions, assuming that the trends of the two are consistent, the remaining life of the operating equipment can be compared to the life estimation of the reference equipment after this stage. To this end, similarity measurement techniques such as the Pearson correlation coefficient or dynamic time warping are used to quantify the matching degree of temperature parameters between the current equipment and the reference equipment under similar working conditions, enhance the reliability of the prediction, and obtain accurate life prediction results.

[0092] Finally, based on the quantified similarity results and the life data of the reference equipment after similar working conditions, the machine learning model is used to accurately predict the remaining life of the current equipment. This process emphasizes that although the focus is on the denoising step in the data preprocessing stage, there are no strict restrictions on the selection of similarity measurement methods and machine learning models, as long as the prediction goals can be achieved. This process ensures that each step from data preprocessing to life prediction is logically clear and smooth, and also provides a scientific basis for the maintenance and management of experimental equipment.

[0093] The present invention also provides a test equipment life analysis system based on big data, the analysis system includes a memory and a processor, the memory stores a computer program, and the processor executes the computer program to implement the steps of the life analysis method.

[0094] In the description of this specification, "plurality" or "several" means at least two, such as two, three or more, etc., unless otherwise clearly and specifically defined.

[0095] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

Claims

1. A test equipment life analysis method based on big data, characterized in that: include: Collect running equipment data, filter data points, and set the search range and window range of the data points, including: taking the data point as the center, setting the length to The window is used as the search range of the data point, with the data point as the center and the length is set to , as the window range of the data points, where and are all preset values ​​and are odd numbers. ; The height of the windows is 1; The other data points within the search range are used as reference points of the data point, and the data points corresponding to the reference points in the pre-collected reference device data are used as corresponding reference points; the window range is smaller than the search range; the similarity between the data point and the reference point is calculated based on the difference between the window range of the data point and the window range of the reference point, and the similarity between the data point and the corresponding reference point is calculated based on the difference between the window range of the data point and the window range of the corresponding reference point, satisfying the following relationship: ； In the formula, For the Data points and The similarity of corresponding reference points, For the The local current difference of each data point, For the The local current difference corresponding to the reference point, For the The value of the data point, For the A value of a corresponding reference point, wherein the method for obtaining the local current difference of the corresponding reference point is the same as the method for obtaining the local current difference of the data point; Based on the similarity between the data point and the reference point, and the similarity between the data point and the corresponding reference point, a final similarity between the data point and the reference point is calculated, the final similarity is used as a filtering weight for the data point, non-local mean filtering is performed on all data, and the life of the equipment is predicted based on the filtered data; The similarity between the data point and the reference point satisfies the following relationship: ； In the formula, For the data point and its The similarity of the reference points, For the The noise performance of the reference point, For the The window of data points The value of the data point, For the The reference point within the window The value of the data point, For the The window of data points The collection time of data points, For the The reference point within the window The collection time of data points, is the natural exponential function, is the length of the window range, is the absolute value symbol; The final similarity between the data point and the reference point satisfies the following relationship: ； In the formula, For the data point and its The final similarity of the reference points, For the data point and its The similarity of the reference points, For the The value of the data point, For the The value of the reference point, For the Data points and The similarity of corresponding reference points, For the The value corresponding to the reference point, is an absolute value symbol; the reference point corresponds one-to-one to the corresponding reference point.

2. The test equipment life analysis method based on big data according to claim 1 is characterized in that: The predicting of the life of the equipment according to the filtered data includes: By comparing the filtered data of the operating device with the data of the reference device, the stage where the operating conditions of the two are similar is determined, and the life of the operating device is predicted based on the remaining life of the reference device in the stage of similar operating conditions.

3. The test equipment life analysis method based on big data according to claim 1 is characterized in that: The method for obtaining the local current difference of the data point is: The collected currents corresponding to all data points within the window range of the data points are obtained as the local current sequence of the data points, and the mean of the first-order differences of the local current sequence is used as the local current difference of the data points.

4. The test equipment life analysis method based on big data according to claim 1 is characterized in that: The method for filtering data points is: Among all the data points, calculate the noise performance of each data point: ； In the formula, For the The noise level of each data point is For the The value of the data point, For the The value of the data point, For the The value of the data point, is the linear normalization function, is the absolute value symbol; A noise performance level threshold is preset, and data points having a noise performance level greater than the noise performance level threshold are screened out.

5. A test equipment life analysis system based on big data, characterized in that: The analysis system comprises a memory and a processor, wherein a computer program is stored in the memory, and the processor executes the computer program to implement the steps of the life analysis method according to any one of claims 1 to 4.

Images