Method for predicting service life of compressor and electronic starter controller

Abstract

The invention discloses a method for predicting the service life of a compressor and an electronic starter controller, and the method comprises the steps: obtaining working condition parameters of the compressor, carrying out the time sequence unification and preprocessing of the working condition parameters, and generating quantitative indexes which comprise a long time sequence sensitive index and a short time sequence fluctuation index; extracting physical features and model automatic features of the long-time-sequence sensitive indexes and the short-time-sequence fluctuation indexes, and obtaining a time sequence sample set; and training and testing the pre-constructed fusion model by using the time sequence sample set to obtain an evaluation result of the fusion model, and performing deployment based on the evaluation result. According to the method, the fusion model is used for performing parallel processing on the long-period aging features and the high-frequency fluctuation features in the compressor, compared with an existing service life prediction mode, the overall precision is higher, comprehensive consideration factors are more comprehensive, meanwhile, the overall precision can be adjusted by adjusting layering parameters through double paths, the stable features are reserved, and the service life prediction accuracy is improved. And the dynamic working condition of the compressor is adapted.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of compressors, in particular to a compressor service life prediction method and an electronic starting controller. BACKGROUND

[0002] The core function of a compressor is to compress gas, increase gas pressure or density, and realize energy transmission. Compressors are widely used in industrial pneumatic systems for compressed air supply, natural gas transportation, gas compression in chemical reactions, new energy vehicle air conditioning, heat pumps, and other fields, and are also the power source for refrigerant circulation in refrigeration equipment (air conditioners and refrigerators).

[0003] Currently, when predicting the service life of a compressor, the average life and remaining life of the compressor are usually calculated based on historical operation data of products of the same type, such as cumulative running time, start-stop frequency, or fault records. Alternatively, extreme conditions are simulated in a laboratory to accelerate compressor aging, and test data are used to extrapolate the service life in actual use scenarios.

[0004] However, the use environment of compressors in different scenarios is different, and there are differences between simulated conditions and actual use scenarios, such as intermittency and load fluctuations in actual conditions, which leads to large errors in the predicted results. The service life cannot be predicted based on the current use conditions of the compressor, and the actual service life cannot be fully reflected. SUMMARY

[0005] To solve the above problems, the present application provides a compressor service life prediction method and an electronic starting controller.

[0006] To achieve the above-mentioned purpose, the present application provides a compressor service life prediction method, characterized in that the compressor service life prediction method comprises: obtaining working condition parameters of a compressor, performing time sequence unification and preprocessing on the working condition parameters, and generating quantitative indicators, wherein the quantitative indicators include long-time sequence sensitive indicators and short-time sequence fluctuation indicators; extracting physical features and model automatic features of the long-time sequence sensitive indicators and the short-time sequence fluctuation indicators, and obtaining a time sequence sample set; training and testing a pre-constructed fusion model using the time sequence sample set, obtaining an evaluation result of the fusion model, and deploying based on the evaluation result.

[0007] According to one aspect of the above technical solution, in the step of obtaining quantitative indicators of a compressor and performing time sequence unification and preprocessing on the quantitative indicators: The working condition parameters of the compressor are collected by different types of sensors with unified timing at different frequencies, the working condition parameters are subjected to data cleaning, timing alignment and standardization processing, and quantitative indexes are generated; Based on the timing of the quantitative indexes, the quantitative indexes include long-time sequence sensitive indexes and short-time sequence fluctuation indexes, the long-time sequence sensitive indexes at least include vibration signals and lubrication parameters of compressor components, and the short-time sequence fluctuation indexes at least include operating parameters and environmental parameters of the compressor.

[0008] According to an aspect of the above technical solution, the step of extracting the physical characteristics and model automatic characteristics of the long-time sequence sensitive indexes and the short-time sequence fluctuation indexes includes: According to the vibration signal, a timing vibration curve is drawn, a first window is used to sample the timing vibration curve, a linear regression fitting is performed on the sampling interval, a timing trend feature of the timing vibration curve is obtained, and the vibration kurtosis and waveform factor of the curve segment in each sliding sampling unit in the first window are calculated, and the week-on-week growth rate and oil film thickness of the lubrication parameter are calculated; The timing curves of the operating parameters and the environmental parameters are sampled using a second window and a third window respectively, and the variation coefficient of the operating parameters and the power load rate correlation and the corrected power of the environmental parameters are calculated according to the segment data in the sliding sampling units of the second window and the third window; Wherein, the length of the sliding sampling unit of any window is equal, and the physical characteristics of the long-time sequence sensitive indexes and the short-time sequence fluctuation indexes obtained by the sliding sampling unit are summarized.

[0009] According to an aspect of the above technical solution, after obtaining the physical characteristics, the model automatic characteristics are extracted, a preset length window and a preset step sliding window are used, the timing axis is taken as the reference, a plurality of timing segment samples are obtained by cutting the continuous quantitative indexes according to the preset window length and the preset step, and the dimension attribute of any timing segment sample matches the window length and the step of the sliding window; According to the fault threshold and the fault shutdown timestamp of each component of the compressor, the residual life label of the timing segment sample is calculated and obtained:

[0010] Wherein, is the residual life label, is the fault shutdown timestamp, is the end timestamp of the timing segment sample; According to the extracted physical characteristics and model automatic characteristics, a timing sample set obtained by training a fusion model is output.

[0011] According to an aspect of the above technical solution, the constructed fusion model comprises an input layer, a feature extraction layer, and an output layer, wherein: The feature extraction layer comprises an LSTM path and a GRU path connected in parallel, the LSTM path and the GRU path are connected to the output layer through a feature fusion layer, and the output layer comprises a full connection layer and a final output layer; The LSTM path comprises a first long-time extraction layer and a second long-time extraction layer arranged in series, the number of neurons of the first long-time extraction layer is 256, and the number of neurons of the second long-time extraction layer is 128; The GRU path comprises a first short-time extraction layer and a second short-time extraction layer arranged in series, the number of neurons of the first short-time extraction layer is 128, and the number of neurons of the second short-time extraction layer is 64; The LSTM path and the GRU path are spliced using the feature fusion layer to output a fusion feature vector.

[0012] According to an aspect of the above technical solution, the fusion feature vector is non-linearly mapped using a full connection layer to obtain an intermediate feature, and the intermediate feature is linearly mapped through a final output layer to obtain a service life prediction value:

[0013] wherein, is the intermediate feature, is the fusion feature vector, is a weight matrix of the full connection layer, is a bias vector of the full connection layer, is an activation function;

[0014] wherein, is the service life prediction value, is a weight matrix of the final output layer, is a bias vector of the final output layer.

[0015] According to an aspect of the above technical solution, the step of using the time series sample set to train and test the pre-constructed fusion model, obtaining an evaluation result of the fusion model, and deploying based on the evaluation result comprises: The time series sample set is divided into a training set, a validation set, and a test set according to a preset proportion, and the fusion model is trained, wherein the distribution proportions of the service life prediction values of the training set, the validation set, and the test set are the same; The overall precision of the test set is obtained, and if the overall precision does not reach an expected threshold, the layered parameters in the LSTM path and the GRU path are adjusted to optimize the fusion model; The optimized fusion model is deployed on an edge server, and parameters of a feature fusion layer are adjusted according to real-time data.

[0016] The application further provides a compressor service life prediction system for implementing the compressor service life prediction method. An index quantization module is configured to acquire working condition parameters of the compressor, perform time sequence unification and preprocessing on the working condition parameters, and generate quantized indexes, wherein the quantized indexes include long-time sequence sensitive indexes and short-time sequence fluctuation indexes. A feature extraction module is configured to extract physical features and model automatic features of the long-time sequence sensitive indexes and the short-time sequence fluctuation indexes, and acquire a time sequence sample set. An evaluation deployment module is configured to train and test a pre-constructed fusion model using the time sequence sample set, acquire an evaluation result of the fusion model, and perform deployment based on the evaluation result.

[0017] The application further provides a computer readable storage medium having a computer program stored thereon, and the program is executed by a processor to implement the compressor service life prediction method.

[0018] The application further provides an electronic starting controller, which comprises a memory, a processor, and a computer program stored in the memory and running on the processor, and the processor implements the compressor service life prediction method when executing the computer program.

[0019] In summary, according to the compressor service life prediction method, vibration types, running types and other parameters of the compressor are screened, quantized indexes are generated after preprocessing, physical features and model automatic features in the quantized indexes are extracted, a time sequence sample set is generated, long-period aging features of physical types in the time sequence sample set are captured using LSTM paths and GRU paths arranged in parallel in a fusion model, and high-frequency fluctuation features are efficiently processed, so that the fusion model simultaneously learns slow aging trends and fast working condition influences, and robustness is improved. The long-period aging features and the high-frequency fluctuation features in the compressor are processed in parallel by using the fusion model, compared with existing service life prediction methods, the overall precision is higher, and comprehensive factors are considered more comprehensively, and the double paths can also adjust hierarchical parameters to adjust the overall precision to retain stable features and adapt to dynamic working conditions of the compressor.

[0020] Additional aspects and advantages of the application will be given, partially in the following description, partially will become obvious from the following description, or will be learned by practicing the application. BRIEF DESCRIPTION OF DRAWINGS

[0021] Figure 1The flow chart of the compressor service life prediction method in the embodiment one of the present application is shown in Fig. 1. Figure 2 The structural schematic diagram of the compressor service life prediction system in the embodiment two of the present application is shown in Fig. 2. Figure 3 The structural block diagram of the electronic control device in the embodiment four of the present application is shown in Fig. 4. DETAILED DESCRIPTION

[0022] In order to make the objects, features and advantages of the present application more clear and easy to understand, the specific embodiments of the present application are described in detail below with reference to the drawings. The embodiments of the present application are shown in the drawings. However, the present application can be realized in many different forms and is not limited to the embodiments described herein. On the contrary, the purpose of providing these embodiments is to make the disclosure of the present application more thorough and comprehensive.

[0023] It should be noted that when an element is referred to as being "fixedly attached" to another element, it can be directly on the other element or intervening elements can also be present. When an element is referred to as being "connected" or "coupled" to another element, it can be directly connected to the other element or intervening elements can also be present. The terms "vertical", "horizontal", "left", "right", "upper", "lower", and the like as used herein are used for illustration only and do not indicate or imply that the device or element must be in a specific orientation, constructed and operated in a particular orientation, and therefore should not be construed as limiting the present application.

[0024] In the present application, unless otherwise explicitly specified and limited, the terms "mounting", "connecting", "connecting", "fixing" and the like should be understood broadly, for example, it can be fixedly connected, or it can be detachably connected, or it can be integrally connected; it can be mechanically connected, or it can be electrically connected; it can be directly connected, or it can be indirectly connected through an intermediate medium, or it can be the communication inside two elements. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances. The term "and / or" used herein includes any and all combinations of one or more related listed items.

[0025] Embodiment one Please refer to Figure 1 , which shows a flow chart of a compressor service life prediction method in the embodiment one of the present application. The compressor service life prediction method comprises the following steps S01-S03, wherein: S01, obtaining the working condition parameters of the compressor, performing time sequence unification and preprocessing on the working condition parameters, and generating quantitative indexes, wherein the quantitative indexes include long time sequence sensitive indexes and short time sequence fluctuation indexes.

[0026] Different types of sensors are used to acquire vibration, lubrication, operational, and environmental parameters during compressor operation. Vibration and lubrication parameters are long-term sensitive indicators, while operational and environmental parameters are short-term fluctuation indicators. Vibration parameters include physical characteristics such as bearing housing acceleration and cylinder wall vibration velocity; lubrication parameters include oil contamination level and oil film thickness; operational parameters include exhaust pressure, intake temperature, and instantaneous power; and environmental parameters include ambient temperature and load rate. It is important to note that the timing of all sensors must be standardized during parameter acquisition, and different parameters should be sampled at different frequencies. Vibration parameters can be sampled at 10kHz or 5kHz. For example, bearing housing acceleration can be sampled at 10kHz, taking 1024 time-domain statistical values ​​every 10 minutes, such as peak value, kurtosis, and root mean square (RMS). Cylinder wall vibration velocity can be sampled at 5kHz. After sampling in this manner, the data can be stored in the format of component name-timestamp-quantized value.

[0027] For data cleaning after storage, box plots can be used to identify abrupt changes in pressure or temperature values, and the moving average of the preceding several acquisition cycles can be used to replace them. For missing values ​​of vibration parameters, spline differences can be used to replace them, and for missing values ​​of power, the average of the acquisition cycles symmetrical before and after the missing value can be used to replace them.

[0028] In addition, the time granularity of all cleaned data is unified to ensure that the standardized data within a unit of time includes all long-term time-sensitive indicators and short-term time-fluctuation indicators. For long-term time-series indicators, the average value is taken according to the unit of time, and for short-term time-fluctuation indicators, the sampling is directly performed in units of time, which can be one minute or two minutes.

[0029] S02. Extract the physical features and automatic model features of the long-term time-series sensitive indicators and the short-term time-series fluctuation indicators to obtain a time-series sample set.

[0030] The physical characteristics need to be calculated by classifying them into long-term sensitive indicators and short-term fluctuation indicators. Each physical characteristic corresponds to the specific operating state or deterioration pattern of the compressor. This step is based on the standardized data per unit time in step S01 for extraction.

[0031] According to the parameter of the vibration type, a time sequence vibration curve is drawn, a first window is used for sliding sampling of the time sequence vibration curve, linear regression fitting is performed on the sampling interval, the time sequence trend feature of the time sequence vibration curve is obtained, and the vibration kurtosis and waveform factor of the curve segment in each sliding sampling unit in the first window are calculated. The vibration kurtosis reflects the concentration degree of the signal peak value, and the kurtosis will be significantly increased at the initial stage of the fault. The waveform factor reflects the impact characteristics of the signal. By sliding sampling the time sequence vibration curve of the middle bearing seat or the cylinder wall of the compressor, the vibration intensity and wear condition of the parts can be quantified.

[0032] As for the lubrication parameter, the time sequence data thereof is also subjected to sliding sampling, and the week-to-week growth rate of the lubrication parameter and the oil film thickness are calculated. The week-to-week growth rate represents the contamination degree of the oil. The sliding standard deviation is calculated by using the oil film thickness, and the sliding standard deviation represents the stability of the oil film. By observing the growth trend of the sliding standard deviation, the stability of the oil film is determined, and then whether the lubrication is likely to fail is predicted.

[0033] The time sequence curves of the operation parameters and the environmental parameters are sampled by using a second window and a third window respectively, and the coefficient of variation of the operation parameters and the power load rate correlation and the corrected power of the environmental parameters are calculated according to the segment data in the sliding sampling units of the second window and the third window. The lengths of the sliding sampling units of the first window, the second window and the third window are equal, the length of the sliding sampling unit in this step is the unit time in step S01, and the first window, the second window and the third window can be set to different lengths of time. The physical characteristics of the long time sequence sensitive indicators and the short time sequence fluctuation indicators obtained by the sliding sampling units are summarized.

[0034] After the physical characteristics are obtained, the automatic features of the model are extracted, a preset length window and a preset step sliding window are used, the time sequence axis is taken as the reference, the continuous quantitative indicators according to the preset window length and the preset step are cut, and a plurality of time sequence segment samples are obtained. The dimension attribute of any time sequence segment sample matches the window length and the step of the sliding window. According to the fault threshold and the fault shutdown time stamp of each component of the compressor, the residual life label of the time sequence segment sample is calculated and obtained:

[0035] wherein, the residual life label is, the fault shutdown time stamp is, the end time stamp of the time sequence segment sample is.

[0036] It should be noted that the remaining life label in the embodiment is focused on compressor failure shutdown, and the failure shutdown is that the parameters in the long time sequence sensitive index and the short time sequence fluctuation index exceed the set threshold.

[0037] S03, training and testing the pre-constructed fusion model using the time sequence sample set, obtaining an evaluation result of the fusion model, and deploying based on the evaluation result.

[0038] The fusion model in the embodiment is an LSTM-GRU fusion model, which includes an input layer, a feature extraction layer, and an output layer. The feature extraction layer includes an LSTM path and a GRU path connected in parallel, the LSTM path focuses on long time sequence aging feature extraction, and the GRU path focuses on short time sequence fluctuation feature extraction. The LSTM path and the GRU path are connected to the output layer through a feature fusion layer, and the output layer includes a full connection layer and a final output layer.

[0039] The LSTM path includes a first long time extraction layer and a second long time extraction layer arranged in series. The first long time extraction layer has 256 neurons, and 256 neurons are sufficient to capture the long-term correlation of slow-changing indicators such as vibration and lubricating oil (such as the cumulative trend of bearing wear), so that the output of the first long time extraction layer can be used as the input of the next layer LSTM, further deepening the long time sequence feature mining. The second long time extraction layer has 128 neurons, and the number of neurons is reduced from 256 to 128, reducing the number of model parameters and the amount of calculation, avoiding overfitting, and at the same time compressing the features in a sliding window into a global vector (128 dimensions), which is convenient for subsequent feature fusion with the GRU path. In addition, a Dropout layer can be connected after the second long time extraction layer, and the dropout rate is set to 0.3, that is, 30% of the neurons are randomly discarded, which is used to suppress the overfitting of the long time sequence path and avoid the model overfitting the long period noise of the training set, such as abnormal vibration but non-fault signal in a certain period of time.

[0040] The GRU path includes a first short-time extraction layer and a second short-time extraction layer arranged in series, the first short-time extraction layer has 128 neurons, the GRU path is simpler and more efficient than the LSTM path, and 128 neurons are sufficient to capture the short-term correlation of fast-changing indicators such as pressure and power (such as power fluctuations caused by load changes), the output of the first short-time extraction layer can adapt to the input requirements of the second short-time extraction layer; the second short-time extraction layer has 64 neurons, which matches the lightweight positioning of the GRU path and avoids occupying too many computing resources, and the second short-time extraction layer compresses the features in a sliding window into a global vector (64 dimensions). In addition, a Dropout layer can also be connected after the second short-time extraction layer, with a dropout rate of 0.2, that is, 20% of the neurons are randomly discarded. Compared with the LSTM path, the noise of short-term fluctuation features is relatively small, a lower dropout rate can retain more effective features (such as load fluctuation signals under normal working conditions), and overfitting can be avoided.

[0041] In the feature fusion layer, the LSTM path and the GRU path can be spliced using a splicing function, and a 192-dimensional fusion feature vector is output, which contains long-term aging trend features and short-term working condition fluctuation features. The splicing fusion does not lose the feature weight of any path, and is better than the additive fusion in retaining the advantages of both paths (if added, the feature specificity of each path will be weakened), so that the subsequent fully connected layer can learn the correlation between the two types of features and RUL.

[0042] The fusion feature vector is nonlinearly mapped using a fully connected layer to obtain intermediate features, and the intermediate features are linearly mapped through a final output layer to obtain the service life prediction value:

[0043] wherein, is the intermediate feature, is the fusion feature vector, is the weight matrix of the fully connected layer, is the bias vector of the fully connected layer, is the activation function;

[0044] wherein, is the service life prediction value, is the weight matrix of the final output layer, is the bias vector of the final output layer.

[0045] The time sequence sample set is divided into a training set, a validation set and a test set according to a preset ratio, and the fusion model is trained, wherein the distribution ratio of the service life prediction values of the training set, the validation set and the test set is the same. The overall accuracy of the test set is obtained, and if the overall accuracy does not reach an expected threshold, the hierarchical parameters in the LSTM path and the GRU path are adjusted to optimize the fusion model. The optimized fusion model is deployed on an edge server, and the parameters of the feature fusion layer are adjusted according to real-time data.

[0046] In summary, according to the compressor service life prediction method provided by the application, the vibration type, the operation type and other parameters of the compressor are screened, the quantization indicators are generated after preprocessing, the physical features and model automatic features in the quantization indicators are extracted, the time sequence sample set is generated, the LSTM path and the GRU path in the fusion model are used to capture the long-period aging characteristics of the physical type in the time sequence sample set and efficiently process the high-frequency fluctuation characteristics, so that the fusion model learns the slow aging trend and the fast working condition influence at the same time, and the robustness is improved. Compared with the existing service life prediction method, the overall accuracy is higher, the factors are more comprehensive, and the double path can also adjust the hierarchical parameters to adjust the overall accuracy to retain stable features and adapt to the dynamic working condition of the compressor.

[0047] Embodiment two The application also provides a compressor service life prediction system, please refer to Figure 2 , which is a structure schematic diagram of the compressor service life prediction system in the embodiment two of the application. The compressor service life prediction system comprises: An index quantization module 11 is configured to obtain the working condition parameters of the compressor, perform time sequence unification and preprocessing on the working condition parameters, and generate quantization indicators, wherein the quantization indicators comprise long-time sequence sensitive indicators and short-time sequence fluctuation indicators. A feature extraction module 12 is configured to extract the physical features and model automatic features of the long-time sequence sensitive indicators and the short-time sequence fluctuation indicators, and obtain a time sequence sample set. An evaluation and deployment module 13 is configured to train and test a pre-constructed fusion model using the time sequence sample set, obtain an evaluation result of the fusion model, and perform deployment based on the evaluation result.

[0048] Embodiment three The application also provides a computer readable storage medium having one or more computer programs stored thereon, wherein the programs are executed by a processor to implement the above-mentioned compressor service life prediction method.

[0049] Those skilled in the art will appreciate that the logic or steps represented in flow diagrams or otherwise described herein, for example, can be thought of as a list of executable instructions for implementing logic functions, and can be embodied in any computer-readable storage medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor-containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions, or a combination of both. In the context of this specification, a "computer-readable storage medium" can be any means that can contain, store, communicate, propagate or transport the program for use by or in connection with the instruction execution system, apparatus, or device.

[0050] More specific examples (a non-exhaustive list) of the computer-readable storage medium include the following: an electrical connection having one or more wires (electrical device), a portable computer diskette (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable storage medium can even be paper or another suitable medium upon which the program is printed, as the program can be electronically captured, for example, via an optical scanner, then compiled, interpreted, or otherwise processed, and stored in a computer memory in a form that is then employable by a computer.

[0051] Embodiment Four Figure 3 A structural block diagram of an electronic governor provided for Embodiment Four is shown. The electronic governor includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the compressor service life prediction method in the above embodiments when executing the program. Figure 3 The electronic governor 30 shown is merely an example and should not bring any limitation to the function and use range of the embodiments of the present application.

[0052] As shown in Figure 3 The electronic governor 30 can be in the form of a general computing device, for example, it can be a server device. The components of the electronic governor 30 can include, but are not limited to, the at least one processor 31 described above, the at least one memory 32 described above, and a bus 33 connecting different system components, including the memory 32 and the processor 31.

[0053] The bus 33 includes a data bus, an address bus, and a control bus.

[0054] The memory 32 can include volatile memory, such as RAM 321 (Random Access Memory), and / or cache memory 322, and further can include non-volatile memory, such as ROM 323 (Read Only Memory).

[0055] The memory 32 can also include a program tool 325 having a set of one or more program modules 324, such as an operating system, one or more application programs, other program modules, and program data, and can include an implementation of a network environment, each or a combination thereof.

[0056] The processor 31, through the running of the computer program stored in the memory 32, executes various function applications and data processing, such as the compressor service life prediction method of the present application as described above.

[0057] The electronic controller 30 can also communicate with one or more external devices 34 such as a keyboard or other input device, a pointing device, etc. through I / O interface 35 (Input / Output interface). Further, the electronic controller 30 can communicate with one or more networks, such as a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet, through network adapter 36. As Figure 3 illustrated, the network adapter 36 communicates with the other modules of the model generated electronic controller 30 through the bus 33. It should be appreciated that although not shown, other hardware and / or software modules could be used in conjunction with the model generated electronic controller 30 including, but not limited to, microcode, device drivers, redundant processing units, disk drive arrays, RAID (Redundant Array of Independent Disks) systems, tape drives, and data archival storage systems, etc.

[0058] It should be noted that although several units / modules or sub-units / modules of the electronic controller are mentioned in the foregoing detailed description, such a division is merely exemplary and not mandatory. Indeed, according to embodiments of the present application, features and functions of two or more units / modules described above can be embodied in one unit / module. Conversely, features and functions of one unit / module described above can be further divided into a plurality of units / modules.

[0059] In the description of the specification, the description referring to the terms "one embodiment", "some embodiments", "an example", "a specific example", or "some examples" and the like means that a specific feature, structure, material or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the present application. Illustrative representations of the above terms in the specification do not necessarily refer to the same embodiment or example. Also, the specific features, structures, materials, or characteristics described can be combined in any suitable manner in one or more embodiments or examples.

[0060] The above embodiments only express several implementation manners of the present application, which are described in a more specific and detailed manner, but cannot be understood as a limitation on the patent scope of the present application. It should be noted that, for ordinary skilled persons in the art, several modifications and improvements can be made without departing from the concept of the present application, which all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application should be subject to the appended claims.

Claims

1. A method for predicting the service life of a compressor, characterized by, The compressor service life prediction method comprises: Obtaining the working condition parameters of the compressor, time series unification and preprocessing of the working condition parameters, and generating quantitative indicators, wherein the quantitative indicators include long-time sequence sensitive indicators and short-time sequence fluctuation indicators; Extracting physical characteristics and model automatic characteristics of the long-time sequence sensitive indicators and the short-time sequence fluctuation indicators, and obtaining a time series sample set; Training and testing the pre-constructed fusion model using the time series sample set, obtaining the evaluation results of the fusion model, and deploying based on the evaluation results.

2. The compressor life prediction method according to claim 1, characterized by, In the steps of obtaining the quantitative indicators of the compressor and time series unification and preprocessing of the quantitative indicators: Different types of sensors are used to collect the working condition parameters of the compressor at different frequencies, and the working condition parameters are subjected to data cleaning, time series alignment and standardization processing to generate quantitative indicators; Based on the time series of the quantitative indicators, the quantitative indicators include long-time sequence sensitive indicators and short-time sequence fluctuation indicators, the long-time sequence sensitive indicators at least include vibration signals and lubrication parameters of compressor components, and the short-time sequence fluctuation indicators at least include operating parameters and environmental parameters of the compressor.

3. The compressor life prediction method according to claim 1, characterized by, The step of extracting the physical characteristics and model automatic characteristics of the long-time sequence sensitive indicators and the short-time sequence fluctuation indicators comprises: According to the vibration signal, a time series vibration curve is drawn, a first window is used to sample the time series vibration curve, a linear regression fitting is performed on the sampling interval, the time series trend characteristics of the time series vibration curve are obtained, and the vibration kurtosis and waveform factor of the curve segment in each sliding sampling unit in the first window are calculated, and the weekly growth rate and oil film thickness of the lubrication parameter are calculated; The second window and the third window are used to sample the time series curves of the operating parameters and the environmental parameters respectively, and the variation coefficient of the operating parameters and the power load rate correlation and correction power of the environmental parameters are calculated according to the segment data in the sliding sampling unit of the second window and the third window; Wherein, the length of the sliding sampling unit of any window is equal, and the physical characteristics of the long-time sequence sensitive indicators and the short-time sequence fluctuation indicators obtained by the sliding sampling unit are summarized.

4. The compressor life prediction method according to claim 3, characterized by, After obtaining the physical characteristics, the model automatic characteristics are extracted, a preset length window and a preset step sliding window are used, the time series axis is taken as the reference, a plurality of time series segment samples are obtained by cutting the continuous quantitative indicators according to the preset window length and the preset step, and the dimension attribute of any time series segment sample matches the window length and the step of the sliding window; According to the fault threshold and the fault shutdown time stamp of each component of the compressor, the residual life label of the time series segment sample is calculated: wherein, is a remaining life tag, is a failure downtime timestamp, is a timing segment sample end timestamp; According to the extracted physical characteristics and model automatic characteristics, the time series sample set obtained by the fusion model training is output.

5. The compressor life prediction method according to claim 1, characterized by, The constructed fusion model comprises an input layer, a feature extraction layer, and an output layer, wherein: The feature extraction layer comprises a parallelly connected LSTM path and a GRU path, the LSTM path and the GRU path are connected to the output layer through a feature fusion layer, and the output layer comprises a full connection layer and a final output layer; The LSTM path comprises a first long-time extraction layer and a second long-time extraction layer arranged in series, the first long-time extraction layer has 256 neurons, and the second long-time extraction layer has 128 neurons; The GRU path comprises a first short-time extraction layer and a second short-time extraction layer arranged in series, the first short-time extraction layer has 128 neurons, and the second short-time extraction layer has 64 neurons; The LSTM path and the GRU path are spliced using a feature fusion layer to output a fusion feature vector.

6. The compressor life prediction method according to claim 5, characterized by, The fusion feature vector is nonlinearly mapped using a fully connected layer to obtain an intermediate feature, and the intermediate feature is linearly mapped through a final output layer to obtain a service life prediction value: wherein, is an intermediate feature, is a fusion feature vector, is a weight matrix of the fully connected layer, is a bias vector of the fully connected layer, is an activation function; wherein, is a usage life prediction value, is a weight matrix of the final output layer, is a bias vector of the final output layer.

7. The compressor life prediction method according to claim 1, characterized by, The step of training and testing the pre-built fusion model using the time series sample set, obtaining an evaluation result of the fusion model, and deploying based on the evaluation result comprises: The time series sample set is divided into a training set, a validation set, and a test set according to a preset ratio, and the fusion model is trained, wherein the distribution ratio of the service life prediction values of the training set, the validation set, and the test set is the same; The overall accuracy of the test set is obtained, and if the overall accuracy does not reach an expected threshold, the layered parameters in the LSTM path and the GRU path are adjusted to optimize the fusion model; The optimized fusion model is deployed on an edge server, and the parameters of the feature fusion layer are adjusted according to real-time data.

8. A compressor life prediction system characterized by, The compressor service life prediction system is used to implement the compressor service life prediction method of any one of claims 1-7, and the system comprises: An index quantization module for obtaining working condition parameters of a compressor, performing time series unification and preprocessing on the working condition parameters, and generating quantized indexes, wherein the quantized indexes include long-time sequence sensitive indexes and short-time sequence fluctuation indexes; A feature extraction module for extracting physical features and model automatic features of the long-time sequence sensitive indexes and the short-time sequence fluctuation indexes, and obtaining a time series sample set; An evaluation and deployment module for training and testing a pre-built fusion model using the time series sample set, obtaining an evaluation result of the fusion model, and deploying based on the evaluation result.

9. A computer readable storage medium having stored thereon a computer program, characterized in that, The program is executed by the processor to implement the compressor service life prediction method of any one of claims 1-7.

10. An electronic actuator comprising a memory, a processor, and a computer program stored on the memory and running on the processor, characterized in that, The processor executes the computer program to implement the compressor service life prediction method of any one of claims 1-7.

Images