Meteorological equipment remaining service life prediction method based on similarity

By constructing the DT-LSTM network model and linear regression model, combining the cosine similarity theorem, matching the health factor curve, the problem of low prediction accuracy of residual service life of meteorological equipment is solved, and higher prediction accuracy is achieved.

CN118964869AInactive Publication Date: 2024-11-15NORTHWESTERN POLYTECHNICAL UNIV

Patent Information

Application Number
CN202410961326.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-17
Publication Date
2024-11-15
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The prior art has low accuracy in predicting the residual service life of meteorological equipment, especially when the data is insufficient or the parameter range is too wide, the model performance is limited.

Method used

The residual service life prediction method of meteorological equipment based on similarity is adopted, and the DT-LSTM network model is constructed by obtaining historical data, multi-dimensional health factor values ​​are output, and the health factor curve is trained using a linear regression model, and the target health factor curve is matched with the cosine similarity theorem. Calculate the remaining service life of the meteorological equipment.

Benefits of technology

The accuracy of the residual service life of meteorological equipment is improved, the accuracy of prediction is improved, and the problem of low accuracy in the prior art is solved.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure BDA0004950367800000121
    Figure BDA0004950367800000121
  • Figure FDA0004950367710000031
    Figure FDA0004950367710000031
  • Figure FDA0004950367710000032
    Figure FDA0004950367710000032
Patent Text Reader

Abstract

The invention discloses a similarity-based residual service life prediction method for meteorological equipment. The method comprises the steps that historical data of meteorological equipment are acquired, the historical data are data of a plurality of multi-dimensional time sequences of the meteorological equipment from operation starting to failure, and the data of each multi-dimensional time sequence comprises failure time; the method comprises the following steps: constructing a DT-LSTM network model, and initializing parameters of the DT-LSTM network model; historical data are input into the initialized DT-LSTM network model, multi-dimensional health factor values are output, each-dimensional health factor value comprises a plurality of health factor values, and the technical problem that in the prior art, prediction precision of the remaining service life of meteorological equipment is low is solved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of prediction and health management, and in particular to a method for predicting the remaining useful life of meteorological equipment based on similarity. Background Art

[0002] At present, most of the remaining useful life prediction models of meteorological equipment are constructed in the following two ways: the equipment mechanism-based approach and the data-driven approach. The equipment mechanism-based approach usually requires a deep understanding of the physical principles and behaviors of the system. This can be a challenge for complex systems and requires specialized domain knowledge and professional skills. At the same time, there are high requirements for a large amount of training data and a comprehensive range of parameters. If the data is insufficient or the parameter range is too wide, the performance of the model may be limited. With a data-driven approach, there is no need to simulate and model the mechanism of the meteorological equipment.

[0003] The full name of the precipitation meter equipment is the precipitation particle three-dimensional imaging measuring instrument. The precipitation meter system consists of an optical unit, an imaging unit, an acquisition and control unit, a data processing unit, and a whole machine protection unit. Among them, the imaging unit is based on the IMX304 model CMOS image sensor, and its function is to quickly expose the precipitation particles passing through the sampling area. According to the working principle of the precipitation meter, whether the equipment can work normally is closely related to the working status of the precipitation meter pulse light source and the CMOS image sensor. Therefore, analyzing the fault health status of the precipitation meter pulse light source and the CMOS image sensor can become a direct basis for judging the fault health status of the precipitation meter equipment. In summary, studying the model of anomaly detection of meteorological equipment has important research significance. Summary of the invention

[0004] The embodiment of the present invention provides a method for predicting the remaining useful life of meteorological equipment based on similarity, so as to at least solve the technical problem of low prediction accuracy of the remaining useful life of meteorological equipment in the prior art.

[0005] According to one aspect of an embodiment of the present invention, a method for predicting the remaining useful life of meteorological equipment based on similarity is provided. The method may include: obtaining historical data of meteorological equipment, wherein the historical data is data of multiple multidimensional time series of meteorological equipment from the start of operation to failure, and each multidimensional time series data includes failure time; constructing a DT-LSTM network model and initializing the parameters of the DT-LSTM network model; inputting the historical data into the initialized DT-LSTM network model and outputting multidimensional health factor values, wherein each dimensional health factor value includes multiple health factor values; training the LR model using the multidimensional health factor values ​​to obtain multiple health factor curves corresponding to the multidimensional health factors; processing the multiple health factor curves by an exponential smoothing method and storing them in a health factor curve library, wherein the health factor curve library contains multiple health factor curves; obtain target data of meteorological equipment, wherein the target data is the current online data of the meteorological equipment, and the target data includes time length; input the target data into the DT-LSTM network model and the LR model to obtain the target health factor curve corresponding to the target data; use the cosine similarity theorem to match the target health factor curve with each health factor curve in the health factor curve library to obtain the target curve with the highest similarity to the target health factor in the health factor curve library, wherein there is a lag time in the matching process between the target health factor curve and each health factor curve in the health factor curve library; after obtaining the target curve, the remaining service life of the meteorological equipment is obtained based on the failure time, lag time and time length.

[0006] Optionally, constructing a DT-LSTM network model includes: constructing an encoder of a DT-STMAE network model based on a multi-layer standard LSTM; constructing a decoder of a DT-STMAE network model based on a DT-LSTM and a fully connected neural network; and constructing a DT-LSTM network model based on the encoder and decoder of the DT-STMAE network model.

[0007] Optionally, historical data is input into the initialized DT-LSTM network model to output a multidimensional health factor value, including: using the encoder of the DT-STMAE network model to encode the historical data to obtain multiple one-dimensional vectors; and inputting the multiple one-dimensional vectors into the decoder of the DT-STMAE network model to obtain a multidimensional health factor value.

[0008] Optionally, the LR model is trained using multiple health factor values ​​to obtain multiple health factor curves corresponding to the multi-dimensional health factors, including: normalizing the health factor value of each dimension to obtain the target health factor value of each dimension corresponding to the health factor value of each dimension; based on the target health factor value of each dimension, obtaining multiple health factor curves corresponding to the target health factor value of each dimension.

[0009] Optionally, the process of calculating the similarity between each health factor curve in the health factor curve library and the target health factor curve is: Among them, HI' is the target health factor curve, HI r is each health factor curve, τ is the difference between HI' and HI r , λ is the relaxation factor that controls the similarity between given distances, Sim r,t For HI' and HI r Similarity at lag time τ, cos(HI',HI r ,τ) is HI' and HI r The cosine distance between them.

[0010] Optionally, HI' and HI r The expression of the cosine distance between two points is: Among them, the value range of τ is 0~τ max , τ max is the maximum lag time, k is the time, the value range of k is 1 to T', T' is the time length of the target health factor curve, h' k is the health factor value at each moment under each health factor curve, h τ ' +k The health factor value of the target health factor curve at each moment of the lag time.

[0011] Optionally, based on the failure time, hysteresis time and time length, the expression of the remaining useful life of the meteorological equipment is obtained: RUL = T r -T'-τ(3), where RUL is the remaining useful life of the meteorological equipment, T r is the failure time, T' is the time length, and τ is the lag time.

[0012] Beneficial effects of the present invention:

[0013] The present invention proposes a method for predicting the remaining service life of meteorological equipment based on similarity. First, a DT-STMAE model is proposed, which maps multiple historical multidimensional time series into multidimensional health factor values, and maps the multidimensional health factor values ​​into multiple health factor curves. Then, the current one-dimensional time series is mapped into a one-dimensional health factor value, and the one-dimensional health factor value is mapped into a health factor curve. Finally, a health factor value curve matching method based on similarity is used to match multiple historical health factor curves with a current health factor curve, so as to obtain the remaining service life of the current meteorological equipment. The technical problem of low accuracy in predicting the remaining service life of meteorological equipment in the prior art is solved, and the technical effect of improving the accuracy of predicting the remaining service life of meteorological equipment is achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0015] Figure 1 is a flow chart of a method for predicting the remaining useful life of meteorological equipment based on similarity according to an embodiment of the present invention;

[0016] Figure 2 is a schematic diagram of a DT-LSTM network model according to an embodiment of the present invention;

[0017] Figure 3 It is a schematic diagram of the remaining useful life prediction result of a meteorological device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0018] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only embodiments of a part of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0019] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the terms used in this way can be interchangeable where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0020] Example 1

[0021] According to an embodiment of the present invention, a method for predicting the remaining useful life of meteorological equipment based on similarity is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system comprising at least one set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in an order different from that shown here.

[0022] Figure 1 is a flow chart of a method for predicting the remaining useful life of meteorological equipment based on similarity according to an embodiment of the present invention. Figure 1 As shown, the method may include the following steps:

[0023] Step S101, obtaining historical data of meteorological equipment, wherein the historical data is data of multiple multidimensional time series from the start of operation to failure of the meteorological equipment, and the data of each multidimensional time series includes the failure time.

[0024] In the technical solution provided in step S101 of the present invention, it is assumed that a multivariate time series represents the historical data from the rth operation to failure, where T r is the expiration time, represents the simultaneous readings from m sensors at time i. Using a time window of fixed length l, along Y r Slide, step size is 1, there will be T t -l+1 windows (subsequences), marked as in Each subsequence is transformed into a one-dimensional value z through the DT-STMAE model i+l-1 , the time series of the entire life cycle of meteorological equipment Can be converted to a univariate time series storing weather equipment degradation information Among them, historical data is also training data.

[0025] Step S102, constructing a DT-LSTM network model and initializing the parameters of the DT-LSTM network model.

[0026] In the technical solution provided in the above step S102 of the present invention, a network model is constructed according to a traditional LSTM network, and the parameters of the DT-LSTM network model are initialized.

[0027] Step S103, inputting historical data into the initialized DT-LSTM network model, and outputting multi-dimensional health factor values, wherein each dimensional health factor value includes multiple health factor values.

[0028] In the technical solution provided in the above step S103 of the present invention, historical data is processed by the initialized DT-LSTM network model to obtain multi-dimensional health factor values ​​corresponding to the historical data.

[0029] Step S104, using the multi-dimensional health factor values ​​to train the LR model to obtain multiple health factor curves corresponding to the multi-dimensional health factors.

[0030] In the technical solution provided in the above step S104 of the present invention, the LR model is trained using multi-dimensional health factor values ​​to obtain multiple health factor curves corresponding to the multi-dimensional health factors.

[0031] Step S105, after processing the multiple health factor curves by an exponential smoothing method, the multiple health factor curves are stored in a health factor curve library, wherein the health factor curve library contains multiple health factor curves.

[0032] In the technical solution provided in the above step S105 of the present invention, each health factor curve is processed according to the exponential smoothing method and stored in a health factor curve library to obtain a health factor curve library, wherein the health factor curve library contains multiple health factor curves.

[0033] Step S106, obtaining target data of the meteorological equipment, wherein the target data is the current online data of the meteorological equipment, and the target data includes a time length.

[0034] In the technical solution provided in the above step S105 of the present invention, target data of the meteorological equipment is obtained, wherein the target data is also test data.

[0035] Step S107, input the target data into the DT-LSTM network model and the LR model to obtain the target health factor curve corresponding to the target data.

[0036] In the technical solution provided in step S107 of the present invention, the DT-LSTM network model processes the target data to obtain a one-dimensional health factor value, and inputs the one-dimensional health factor value into the LR model to obtain a target health factor curve corresponding to the target data, wherein the target health factor curve is obtained by HI'={h1',h'2,...,h' T}, where h1',h'2,...,h' T For each health factor in the target health factor curve.

[0037] Step S108, using the cosine similarity theorem, the target health factor curve is matched with each health factor curve in the health factor curve library to obtain the target curve with the highest similarity to the target health factor in the health factor curve library, wherein there is a lag time in the matching process between the target health factor curve and each health factor curve in the health factor curve library.

[0038] In the technical solution provided in the above step S108 of the present invention, the cosine similarity theorem is used to perform similarity matching between the target health factor curve and each health factor curve in the health factor curve library to obtain the target curve with the highest similarity to the target health factor in the health factor curve library.

[0039] Step S109, after obtaining the target curve, the remaining service life of the meteorological equipment is obtained based on the failure time, the lag time and the time length.

[0040] In the technical solution provided in the above step S109 of the present invention, after obtaining the target curve, the remaining service life of the meteorological equipment is obtained by calculating the failure time of the curve in the health factor curve library with the highest similarity to the target curve, and the time length and lag time of the target curve.

[0041] The above method of this embodiment is further introduced below.

[0042] As an optional implementation method, step S102, constructing a DT-LSTM network model, including: constructing an encoder of a DT-STMAE network model based on a multi-layer standard LSTM; constructing a decoder of a DT-STMAE network model based on DT-LSTM and a fully connected neural network; constructing a DT-LSTM network model based on the encoder and decoder of the DT-STMAE network model.

[0043] In this embodiment, according to an embodiment of the present invention, a DT-LSTM network model is provided. Figure 2 is a schematic diagram of a DT-LSTM network model according to an embodiment of the present invention, such as Figure 2 As shown, Figure 2 In this example, encoder is the encoding layer, decoder is the decoding layer, and a DT-LSTM model is constructed. The parameters of the model are initialized, and the LSTM parameters are calculated in the forward propagation. In the backward propagation, the L2 regularization method is used to avoid overfitting, and the Adam method is used to update the weight matrix and bias vector of the LSTM.

[0044] As an optional implementation method, step S103, inputs historical data into the initialized DT-LSTM network model, and outputs a multidimensional health factor value, including: using the encoder of the DT-STMAE network model to encode the historical data to obtain multiple one-dimensional vectors; inputting multiple one-dimensional vectors into the decoder of the DT-STMAE network model to obtain a multidimensional health factor value.

[0045] In this embodiment, the encoder layer converts the input multidimensional time series X={X1, X2, ..., X l}, where each point X i (i=1,2,...,t) is an m-dimensional vector. The DT-STMAE model first uses the LSTM encoder to learn a vector format of fixed length l from the input time series, and then reconstructs it at the decoder layer; t, the hidden layer state h at time t-1 t-1 and cell state c t-1 To predict the hidden layer state h at time t t and cell state c t , propagates forward step by step, and obtains the final hidden layer state h of the encoder layer E , cell status c E and the final output state y E , where y E Construct a context vector sequence; the fully connected layer converts y E Mapped into a one-dimensional variable, the specific formula is h = f (W·y E +b), where W is the weight vector, b is the offset vector, and h is the final output value; at the decoder layer, the one-dimensional variable is reconstructed to obtain the context vector, and its formula is h'=f(W'h+b'), and h' is the final reconstructed context vector.

[0046] As an optional implementation method, step S104 uses multiple health factor values ​​to train the LR model to obtain multiple health factor curves corresponding to the multi-dimensional health factors, including: normalizing the health factor value of each dimension to obtain the target health factor value of each dimension corresponding to the health factor value of each dimension; based on the target health factor value of each dimension, obtaining multiple health factor curves corresponding to the target health factor value of each dimension.

[0047] In this embodiment, before normalizing the health factor value of each dimension, the reconstruction error at time t in the DT-STMAE network model is use Calculate, where ||·||2 is the vector's two-norm operator, and the reconstruction error is the original HI value; where Y t represents the reconstructed time series, X t represents the input time series. Since the original HI value obtained according to DT-STMAE is not necessarily within [0,1], it is normalized to the range of [0,1] to unify the standard. Finally, the normalized HI value is used as the target HI value. The normalization equation is: in, is the maximum value of the health factor of each dimension, Health factor value for each dimension, The exponential smoothing method is used to process the HI curve and store it in the HI curve library. Finally, the HI curve library ∑ = {HI r}(r=1,2,...,R).

[0048] As an optional embodiment, in step S108, the process of calculating the similarity between each health factor curve in the health factor curve library and the target health factor curve is as follows: Among them, HI' is the target health factor curve, HI r is each health factor curve, τ is the difference between HI' and HI r , λ is the relaxation factor that controls the similarity between given distances, Sim r,t For HI' and HI r Similarity at lag time τ, cos(HI',HI r ,τ) is HI' and HI r The cosine distance between them.

[0049] In this embodiment, the similarity between the target health factor curve and each health factor curve is calculated according to formula (1).

[0050] As an optional embodiment, HI' and HI r The expression of the cosine distance between two points is: Among them, the value range of τ is 0~τ max , τ max is the maximum lag time, k is the time, the value range of k is 1 to T', T' is the time length of the target health factor curve, h' k is the health factor value at each moment under each health factor curve, h τ ' +k The health factor value of the target health factor curve at each moment of the lag time.

[0051] In this embodiment, the cosine distance between the labeled health factor curve and each health factor curve can be obtained according to formula (2).

[0052] As an optional embodiment, in step S109, based on the failure time, the lag time and the time length, the expression of the remaining service life of the meteorological equipment is obtained: RUL = T r -T'-τ(3), where RUL is the remaining useful life of the meteorological equipment, T r is the failure time, T' is the time length, and τ is the lag time.

[0053] In this embodiment, the remaining service life of the meteorological equipment is obtained according to formula (3). According to an embodiment of the present invention, a prediction result of the remaining service life of the meteorological equipment is provided. Figure 3 is a schematic diagram of a prediction result of the remaining useful life of a meteorological device according to an embodiment of the present invention. Figure 3 As shown, in Figure 3 In the figure, blue represents the training curve and red represents the test curve. The embodiment of the present invention uses the same evaluation index to compare the proposed DT-STMAE with other algorithms. The comparison results are shown in Table 1.

[0054] Table 1 Comparison of performance indicators of remaining life prediction methods

[0055]

[0056] As can be seen from Table 1, the method proposed in the embodiment of the present invention has good performance in the three indicators. Compared with several other algorithms, it can still have higher accuracy and lower RMSE value on the basis of lower S score. Its accuracy is improved by 10.6% compared with the BiLSTM-DED algorithm, 5.0% compared with DNN-LSTM, and 14.5% compared with BiLSTM-ED. Its RMSE value is reduced by 46.2% compared with the BiLSTM-DED algorithm, 25.5% compared with DNN-LSTM, and 40.2% compared with BiLSTM-ED.

[0057] In an embodiment of the present invention, historical data of meteorological equipment is obtained, wherein the historical data is data of multiple multidimensional time series of meteorological equipment from the start of operation to failure, and the data of each multidimensional time series includes the failure time; a DT-LSTM network model is constructed, and the parameters of the DT-LSTM network model are initialized; the historical data is input into the initialized DT-LSTM network model, and a multidimensional health factor value is output, wherein each dimensional health factor value includes multiple health factor values; the LR model is trained using the multidimensional health factor value to obtain multiple health factor curves corresponding to the multidimensional health factor; the multiple health factor curves are processed by an exponential smoothing method and stored in a health factor curve library, wherein the health factor curve library contains multiple health factor curves; the target data of the meteorological equipment is obtained, wherein the target number The data is the current online data of the meteorological equipment, and the target data includes the time length; the target data is input into the DT-LSTM network model and the LR model to obtain the target health factor curve corresponding to the target data; the cosine similarity theorem is used to match the target health factor curve with each health factor curve in the health factor curve library to obtain the target curve with the highest similarity to the target health factor in the health factor curve library, wherein there is a lag time in the matching process between the target health factor curve and each health factor curve in the health factor curve library; after obtaining the target curve, the remaining service life of the meteorological equipment is obtained based on the failure time, lag time and time length, which solves the technical problem of low accuracy in the existing technology for predicting the remaining service life of the meteorological equipment and achieves the technical effect of improving the accuracy of predicting the remaining service life of the meteorological equipment.

[0058] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0059] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0060] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0061] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed over multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0062] In addition, each functional unit in each embodiment of the present invention may be integrated into a first processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0063] The above are only preferred embodiments of the present invention. It should be pointed out that, for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for predicting the remaining useful life of meteorological equipment based on similarity, characterized in that: include: Acquire historical data of meteorological equipment, wherein the historical data is data of multiple multidimensional time series from the start of operation to failure of the meteorological equipment, and the data of each multidimensional time series includes failure time; Constructing a DT-LSTM network model and initializing parameters of the DT-LSTM network model; Inputting the historical data into the initialized DT-LSTM network model, and outputting a multi-dimensional health factor value, wherein each dimensional health factor value includes multiple health factor values; Using the multidimensional health factor values ​​to train the LR model, a plurality of health factor curves corresponding to the multidimensional health factors are obtained; After processing the multiple health factor curves by an exponential smoothing method, the multiple health factor curves are stored in a health factor curve library, wherein the health factor curve library contains multiple health factor curves; Acquire target data of a meteorological device, wherein the target data is current online data of the meteorological device, and the target data includes a time length; Inputting the target data into the DT-LSTM network model and the LR model to obtain a target health factor curve corresponding to the target data; The target health factor curve is matched with each health factor curve in the health factor curve library by using the cosine similarity theorem to obtain the target curve with the highest similarity to the target health factor in the health factor curve library, wherein there is a lag time in the matching process between the target health factor curve and each health factor curve in the health factor curve library; After the target curve is obtained, the remaining service life of the meteorological equipment is obtained based on the failure time, hysteresis time and time length.

2. The method according to claim 1, characterized in that The DT-LSTM network model is constructed, including: Based on multi-layer standard LSTM, the encoder of the DT-STMAE network model is constructed; Based on DT-LSTM and fully connected neural network, the decoder of DT-STMAE network model is constructed; Based on the encoder and decoder of the DT-STMAE network model, the DT-LSTM network model is constructed.

3. The method according to claim 1, characterized in that: The historical data is input into the initialized DT-LSTM network model to output multi-dimensional health factor values, including: Encoding the historical data using the encoder of the DT-STMAE network model to obtain multiple one-dimensional vectors; The multiple one-dimensional vectors are input into the decoder of the DT-STMAE network model to obtain a multi-dimensional health factor value.

4. The method according to claim 1, characterized in that: The method of training the LR model using the multiple health factor values ​​to obtain multiple health factor curves corresponding to the multi-dimensional health factors includes: Normalizing the health factor values ​​of each dimension to obtain the target health factor value of each dimension corresponding to the health factor value of each dimension; Based on the target health factor value of each dimension, a plurality of health factor curves corresponding to the target health factor value of each dimension are obtained.

5. The method according to claim 1, characterized in that The process of calculating the similarity between each health factor curve in the health factor curve library and the target health factor curve is as follows: Wherein, HI' is the target health factor curve, HI r is each health factor curve, τ is the relationship between the HI' and the HI r , λ is the relaxation factor that controls the similarity between given distances, Sim r,t For the HI' and the HI r Similarity at lag time τ, cos(HI',HI r ,τ) is the difference between the HI' and the HI r The cosine distance between .

6. The method according to claim 5, characterized in that The HI' and the HI r The expression of the cosine distance between two points is: Among them, the value range of τ is 0~τ max , τ max is the maximum lag time, k is the time, the value range of k is 1 to T', T' is the time length of the target health factor curve, h' k is the health factor value at each moment under each health factor curve, h τ ' +k The health factor value of the target health factor curve at each moment of the lag time.

7. The method according to claim 1, characterized in that The expression for obtaining the remaining service life of the meteorological equipment based on the failure time, lag time and time length is: RUL=T r -T'-τ (3) Where RUL is the remaining useful life of the meteorological equipment, T r is the failure time, T' is the time length, and τ is the lag time.

8. A processor, characterized in that: The processor is used to run a program, wherein the program executes the method according to any one of claims 1 to 7 when running.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the method according to any one of claims 1 to 7.

Citation Information

Patent Citations

  • Equipment residual life prediction method and device, electronic equipment and storage medium

    CN116911149A

  • System and method for predicting remaining useful life of a machine component

    US20230376398A1

Cited By

  • Prediction method and system for earth surface time sequence deformation trend of permafrost region

    CN120562692A