Engine fault prediction method and device

Abstract

The invention provides an engine fault prediction method and device. The method comprises the following steps: acquiring historical operation data of an engine; according to the historical operation data and a Bayesian prediction model, a fault classification model of the engine is determined, and the Bayesian prediction model is a preset model of a remote monitoring platform; engine operation data collected by the sensor in the real-time operation process of the engine are obtained, wherein the engine operation data comprise first abnormal operation data; and determining a first predicted fault type of the engine according to the first abnormal operation data and a fault classification model. By means of the engine fault prediction method and device, the operation condition of the engine can be mastered in real time, the fault of the engine can be predicted, and the problem that in the prior art, due to the fact that an off-line regular test method is adopted for monitoring, a large neutral position period exists, and fault prediction of the engine is not accurate is solved.

Description

Technical Field

[0001] The present application relates to the field of engine fault diagnosis, and in particular, to an engine fault prediction method, an engine fault prediction device, a computer-readable storage medium, and an electronic device. Background Art

[0002] The engine is the core component of the power system, and its cost and major repair cost are high. Engine fault diagnosis is to collect the on-site equipment operation data from the inspection system and the online system, and establish an equipment diagnosis model based on the on-site equipment data, and use technical means such as mechanism-based diagnosis and regression algorithms to analyze the degree of engine degradation under condition monitoring.

[0003] In the prior art, the method of offline periodic testing is usually adopted to monitor the engine, but this method has a large idle period, so it is impossible to grasp the timing of major repairs in time, which can easily cause serious mechanical accidents such as bearing damage and gear knocking, resulting in large losses. If you want to shorten the major repair cycle, you need to increase the cost. The engine fault cannot be diagnosed in time, which may cause major mechanical accidents, safety accidents and property losses. Summary of the invention

[0004] The main purpose of the present application is to provide an engine fault prediction method, an engine fault prediction device, a computer-readable storage medium and an electronic device, so as to at least solve the problem of inaccurate engine fault prediction in the prior art.

[0005] In order to achieve the above-mentioned purpose, according to one aspect of the present application, an engine fault prediction method is provided, wherein a sensor is deployed on the engine, and the sensor establishes a communication connection with a remote monitoring platform through the industrial Internet. The engine fault prediction method includes the following steps: S1, obtaining the historical operation data of the engine; S2, determining the fault classification model of the engine based on the historical operation data and the Bayesian prediction model, wherein the Bayesian prediction model is a preset model of the remote monitoring platform; S3, obtaining the engine operation data collected by the sensor during the real-time operation of the engine, wherein the engine operation data includes first abnormal operation data; S4, determining the first predicted fault type of the engine based on the first abnormal operation data and the fault classification model.

[0006] Optionally, the historical operation data includes a historical fault type sample set of the engine in a fault state and a fault classification database, the historical fault type sample set includes at least one historical fault type, the fault classification database includes a historical solution sample set corresponding to the historical fault type sample set, the historical solution sample set includes at least one historical solution, and the historical solution is used to solve the fault problem corresponding to the historical fault type, and the engine fault prediction method also includes: judging whether the first predicted fault type belongs to one of the historical fault types in the historical fault type sample set, and obtaining a second judgment result; when the second judgment result indicates yes, determining a first target solution based on the historical fault types in the historical fault type sample set that are the same as the first predicted fault type and the fault classification database, and the first target solution is used to solve the fault problem corresponding to the first predicted fault type; when the second judgment result indicates no, calling the second target solution based on the first predicted fault type, and configuring the first predicted fault type and the second target solution in the fault classification database to update the fault classification database.

[0007] Optionally, a monitoring processor is deployed on the engine, and the monitoring processor is communicatively connected to the sensor. The engine fault prediction method also includes: determining whether the remote monitoring platform and the monitoring processor establish a communication connection through the industrial Internet at a first moment, and obtaining a first judgment result; when the first judgment result indicates yes, executing steps S3 and S4; when the first judgment result indicates no, determining whether the remote monitoring platform and the monitoring processor establish a communication connection through the industrial Internet at a second moment, and obtaining a second judgment result, and the second moment occurs after the first moment; when the second judgment result indicates yes, obtaining the second abnormal operation data of the engine collected by the monitoring processor during the offline period of the monitoring processor, the second predicted fault type of the engine determined by the monitoring processor based on the second abnormal operation data and the micro-fault classification model, and the third predicted fault type of the engine, and when the second predicted fault type and the third predicted fault type are different, updating the micro-fault classification model to a fault classification model, the offline period is a time period after the first moment and before the second moment, the micro-fault classification model is a preset model of the monitoring processor, and the third predicted fault type is determined by the second abnormal operation data and the fault classification model.

[0008] Optionally, the historical operation data includes a first historical signal sample set of the engine in a healthy state, healthy operation samples corresponding to the first historical signal sample set, a second historical signal sample set of the engine in a faulty state, and a historical fault type sample set corresponding to the second historical signal sample set. According to the historical operation data and the Bayesian prediction model, a fault classification model of the engine is determined, including: determining a feature mapping according to a kernel function and a first preset formula, the feature mapping including a first feature mapping between the historical operation data and the healthy operation samples and a second feature mapping between the historical operation data and the historical fault type sample set; constructing a Bayesian prediction model according to the feature mapping and a sparse Bayesian learning algorithm; training the Bayesian prediction model according to the Bayesian prediction model, the first historical signal sample set, the healthy operation samples, the second historical signal sample set and the historical fault type sample set to obtain a fault classification model.

[0009] Optionally, the first preset formula is:

[0010] Among them, K(x i , x j ) is the kernel function, x i is the i-th training sample, x j is the jth training sample, i and j are both positive integers, the training sample is any sample in the first historical signal sample set and the second historical signal sample set, x represents the position matrix, and φ(x) is the feature map;

[0011] The sparse Bayesian learning algorithm is: Among them, w i is the weight, δ is the noise that follows the N(0,σ) distribution, w represents the weight vector, t i It is label data, which is any sample in the healthy operation sample and historical fault type sample set.

[0012] Optionally, the healthy operation samples and the multiple samples in the historical fault type sample set have the same probability distribution; the probability distribution has the following formula: Where t is t i The matrix of .

[0013] Optionally, the historical operation data is vibration data, and the sensor collects the vibration data by an equal-angle resampling method.

[0014] According to another aspect of the present application, an engine fault prediction device is provided, wherein a sensor is deployed on the engine, and the sensor establishes a communication connection with a remote monitoring platform through the industrial Internet. The engine fault prediction device includes the following steps: a first acquisition module, used to acquire historical operating data of the engine; a construction module, used to determine a fault classification model of the engine based on the historical operating data and a Bayesian prediction model, wherein the Bayesian prediction model is a preset model of the remote monitoring platform; a second acquisition module, used to acquire engine operating data collected by the sensor during real-time operation of the engine, wherein the engine operating data includes first abnormal operating data; and a prediction module, used to determine a first predicted fault type of the engine based on the first abnormal operating data and the fault classification model.

[0015] According to another aspect of the present application, a computer-readable storage medium is provided, the computer-readable storage medium including a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute any one of the above-mentioned engine fault prediction methods.

[0016] According to another aspect of the present application, an electronic device is provided, comprising: one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and the one or more programs include means for executing any one of the above-mentioned engine fault prediction methods.

[0017] By applying the technical solution of the present application, the remote monitoring platform can train the Bayesian prediction model according to the fault characteristics in the historical operation data when acquiring the historical operation data, so as to determine the fault classification model of the engine. On this basis, since the sensors deployed on the engine establish a communication connection with the remote monitoring platform through the industrial Internet, the remote monitoring platform can obtain the engine operation data collected by the sensor during the real-time operation of the engine, and when the engine operation data includes the first abnormal operation data, the trained fault classification model can be used to predict the fault type and complete the fault identification diagnosis. It can be seen that the present application can control the operation status of the engine in real time through the analysis of the remote monitoring platform, and can predict the engine failure, which solves the problem that the engine is monitored by the offline regular test method in the prior art, and there is a large gap period, resulting in inaccurate prediction of engine failure. In addition, through the present application, maintenance personnel can make accurate diagnosis of the failure at the first time and can intervene in the maintenance based on the first predicted fault type of the engine, extend the engine overhaul cycle, enhance the pre-maintenance plan and timely control the overhaul time to avoid the occurrence of serious mechanical accidents, thereby reducing the maintenance cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The drawings constituting part of the present application are used to provide a further understanding of the present application. The exemplary embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0019] Figure 1 A hardware structure block diagram of a mobile terminal for executing an engine failure prediction method provided in an embodiment of the present application is shown;

[0020] Figure 2 A schematic flow chart of an engine fault prediction method provided according to an embodiment of the present application is shown;

[0021] Figure 3 A schematic flow chart of a specific engine fault prediction method provided according to an embodiment of the present application is shown;

[0022] Figure 4 A structural block diagram of an engine fault prediction device provided according to an embodiment of the present application is shown.

[0023] The above drawings include the following reference numerals:

[0024] 102, processor; 104, memory; 106, transmission device; 108, input and output devices. DETAILED DESCRIPTION

[0025] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0026] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0027] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present application described here. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0028] As introduced in the background technology, in the prior art, engines are monitored by offline periodic testing methods, which has a large idle period, resulting in untimely control of the timing of major repairs, which can easily cause serious mechanical accidents and increase maintenance costs. In order to solve the problem of inaccurate engine fault prediction in the prior art, the embodiments of the present application provide an engine fault prediction method, an engine fault prediction device, a computer-readable storage medium and an electronic device.

[0029] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0030] The method embodiments provided in the embodiments of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Taking running on a mobile terminal as an example, Figure 1 FIG. 1 is a hardware structure block diagram of a mobile terminal of an engine fault prediction method according to an embodiment of the present invention. Figure 1 As shown, the mobile terminal may include one or more ( Figure 1 Only one is shown in the figure) a processor 102 (the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA) and a memory 104 for storing data, wherein the mobile terminal may also include a transmission device 106 and an input / output device 108 for communication functions. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the mobile terminal. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations shown.

[0031] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as the computer program corresponding to the engine fault prediction method in the embodiment of the present invention. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, the above method is implemented. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories can be connected to the mobile terminal via a network. Examples of the above network include but are not limited to the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof. The transmission device 106 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of the mobile terminal. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, referred to as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 may be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0032] In this embodiment, an engine fault prediction method running on a mobile terminal, a computer terminal or a similar computing device is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0033] Figure 2 is a flow chart of an engine fault prediction method according to an embodiment of the present application. A sensor is deployed on the engine, and the sensor establishes a communication connection with the remote monitoring platform through the industrial Internet, such as Figure 2 As shown, the method comprises the following steps:

[0034] Step S1, obtaining historical operating data of the engine;

[0035] Specifically, on the remote monitoring platform, the historical operation data of the engine is first obtained. These historical operation data may include but are not limited to vibration signal samples of the engine in a healthy state (normal operation state), engine parameters (such as speed, intake volume, injection pulse width, ignition advance angle, throttle opening, oxygen output voltage, engine load, water temperature and mixed gas), and vibration signal samples of the engine in a fault state, engine parameters (such as speed, intake volume, injection pulse width, ignition advance angle, throttle opening, oxygen output voltage, engine load, water temperature and mixed gas).

[0036] Optionally, the acquisition of these historical operating data is achieved through multiple sensors installed on the engine, and the remote monitoring platform obtains the operating status data of the engine collected by the sensors at historical times. It can be understood that the remote monitoring platform can establish a comprehensive database by acquiring and analyzing the historical operating data of the engine, providing rich training samples and reference basis for subsequent fault classification models.

[0037] Specifically, the historical operation data may be vibration data, and the sensor may collect the vibration data by an equal-angle resampling method, thereby ensuring the accuracy and reliability of the data.

[0038] Step S2, determining the engine fault classification model based on historical operation data and the Bayesian prediction model, where the Bayesian prediction model is a preset model of the remote monitoring platform;

[0039] Specifically, the Bayesian prediction model is a prediction method based on Bayesian statistical theory. It uses existing prior knowledge and data to update the probability model through the Bayesian theorem and derive predictions for future events or unknown quantities. After obtaining the historical operating data of the engine, the Bayesian prediction model can be trained and classified by inputting the above historical operating data into the Bayesian prediction model preset on the remote monitoring platform, thereby determining the fault classification model of the engine.

[0040] Optionally, before using the historical operation data to train the Bayesian prediction model, the historical operation data is preprocessed, and then the Bayesian prediction model is trained based on the preprocessed historical operation data. The preprocessing may include but is not limited to extracting vibration features and engine parameters from signal samples of the historical operation data, cleaning the vibration features and engine parameters, removing abnormal data, and then extracting fault features therefrom.

[0041] Step S3, acquiring engine operation data collected by the sensor during real-time operation of the engine, the engine operation data including first abnormal operation data;

[0042] Specifically, during the operation of the engine, multiple sensors installed on the engine can continuously collect the operation data of the engine, thereby obtaining the engine operation data during the real-time operation of the engine. These engine operation data may include but are not limited to the real-time vibration signal of the engine and engine parameters. In particular, when the engine is in an abnormal operation state, the sensor will collect the first abnormal operation data, which reflects the current fault characteristics of the engine.

[0043] It is understandable that the engine can collect the first abnormal operation data of the engine under abnormal operation by collecting the engine operation data in real time, so as to timely find the engine failure signs. This helps to achieve early warning of engine failure, improve the real-time and accuracy of fault diagnosis, and thus reduce the downtime and maintenance costs caused by failure.

[0044] Step S4: determining a first predicted fault type of the engine according to the first abnormal operation data and the fault classification model.

[0045] Specifically, after receiving the first abnormal operation data collected by the sensor, the remote monitoring platform analyzes the first abnormal operation data using the established fault classification model.

[0046] Optionally, before the fault classification model uses the first abnormal operation data for prediction, the first abnormal operation data is preprocessed, and then the Bayesian prediction model is trained based on the preprocessed first abnormal operation data. Wherein, the preprocessing may include but is not limited to extracting vibration features and engine parameters from the signal samples of the first abnormal operation data and performing data cleaning on the vibration features and engine parameters, removing abnormal data, and then extracting fault features therefrom. Then, the extracted fault features are input into the fault classification model for fault identification; the similarity between the first abnormal operation data and the class to which each fault signal in the fault classification model belongs is calculated; and the first predicted fault type of the engine is determined based on the calculation result of the similarity. By combining the first abnormal operation data and the fault classification model, the fault type of the engine can be determined quickly and accurately. This not only improves the efficiency of fault diagnosis, but also can provide maintenance personnel with accurate fault information, thereby shortening maintenance time, reducing maintenance costs, and improving the operating efficiency and reliability of the engine.

[0047] Through this embodiment, the remote monitoring platform can train the Bayesian prediction model according to the fault characteristics in the historical operation data when obtaining the historical operation data, so as to determine the fault classification model of the engine. On this basis, since the sensors deployed on the engine establish a communication connection with the remote monitoring platform through the industrial Internet, the remote monitoring platform can obtain the engine operation data collected by the sensor during the real-time operation of the engine, and when the engine operation data includes the first abnormal operation data, the trained fault classification model can be used to predict the fault type and complete the fault identification diagnosis. It can be seen that the application can control the operation status of the engine in real time through the analysis of the remote monitoring platform, and the engine failure can be predicted, which solves the problem that the engine is monitored by the offline regular test method in the prior art, and there is a large gap period, resulting in the problem that the timing of major repairs is not timely. In addition, accurate diagnosis and intervention maintenance can be carried out at the first time of the fault, extending the engine overhaul cycle, and enhancing the pre-maintenance plan to avoid the occurrence of serious mechanical accidents, thereby reducing the maintenance cost.

[0048] As a possible implementation, the historical operation data includes a historical fault type sample set of the engine in a fault state and a fault classification database, the historical fault type sample set includes at least one historical fault type, the fault classification database includes a historical solution sample set corresponding to the historical fault type sample set (in other words, the fault classification database includes at least one historical solution for solving a fault problem corresponding to at least one historical fault type in the historical fault type sample set), the historical solution sample set includes at least one historical solution, and the historical solution is used to solve the fault problem corresponding to the historical fault type. The engine fault prediction method further includes the following steps:

[0049] Step S5, determining whether the first predicted fault type belongs to one of the historical fault types in the historical fault type sample set, and obtaining a second determination result;

[0050] Specifically, the first predicted fault type determined in step S4 is matched with a historical fault type in the historical fault type sample set. If the first predicted fault type is consistent with a historical fault type in the historical fault type sample set, the second judgment result is "yes"; otherwise, the second judgment result is "no".

[0051] Step S6, when the second judgment result indicates yes, a first target solution is determined based on the historical fault type in the historical fault type sample set that is the same as the first predicted fault type and the fault classification database, and the first target solution is used to solve the fault problem corresponding to the first predicted fault type;

[0052] Specifically, when the second judgment result is "yes", the same historical fault type as the first predicted fault type can be matched from the historical fault type sample set, and the historical solution corresponding to the historical fault type can be extracted from the fault classification database according to the matched historical fault type. This process is usually completed through indexing and database query to ensure the rapid extraction of solutions. Then, a historical solution for solving the fault problem corresponding to the historical fault type matching the first predicted fault type is selected from the historical solution sample set of the fault classification database, and the historical solution can be determined as the first target solution.

[0053] Among them, in the step of selecting a historical solution for solving the fault problem corresponding to the historical fault type matching the first predicted fault type from the historical solution sample set of the fault classification database, the selection process may involve strategies such as priority sorting or historical success rate evaluation or a combination of the two to ensure that the selected solution is optimal. The selected first target solution is applied to the current fault situation to solve the fault problem corresponding to the first predicted fault type. The application of the solution may involve automated control or manual intervention or a combination of the two.

[0054] Step S7, when the second judgment result indicates no, calling the second target solution according to the first predicted fault type, and configuring the first predicted fault type and the second target solution in the fault classification database to update the fault classification database.

[0055] Specifically, when the second judgment result is "no", the second target solution is called. The second target solution may be a default emergency treatment solution, or a temporary solution generated according to a preset algorithm. The first predicted fault type and the corresponding second target solution are added to the fault classification database to form new historical fault type samples and historical solution samples to update the fault classification database, ensure the diversity of fault types and solutions in the fault classification database, and always keep it up to date, so as to provide a more accurate reference for future fault prediction.

[0056] It can be seen that by collecting and processing new fault types, the accuracy of future fault prediction can be improved and the occurrence rate of unknown faults can be reduced. By continuously updating the fault classification database, the system can continuously accumulate new fault types and solutions, and enhance the system's self-learning ability and adaptability.

[0057] As a possible implementation, a monitoring processor is deployed on the engine, and the monitoring processor is communicatively connected with the sensor. The engine fault prediction method further includes the following steps:

[0058] Step S8, determining whether the remote monitoring platform and the monitoring processor establish a communication connection through the industrial Internet at the first moment, and obtaining a first determination result;

[0059] Specifically, at the first moment, the monitoring processor performs a communication connection detection with the remote monitoring platform through the industrial Internet. This process usually involves a data packet transmission test to confirm the communication status between the two. If the communication connection is detected to be normal, the first judgment result is "yes"; otherwise, the first judgment result is "no".

[0060] Step S9, when the first judgment result indicates yes, execute steps S3 and S4;

[0061] Specifically, when the communication connection is normal, the monitoring processor will transmit the collected engine operation data to the remote monitoring platform in real time. The remote monitoring platform will further process and analyze the received data, including data cleaning, feature extraction and fault prediction, so as to analyze and obtain the first abnormal operation data in the transmitter operation data, and can determine the first predicted fault type of the engine based on the first abnormal operation data and the fault classification model.

[0062] In addition, when the first judgment result indicates yes, the remote monitoring platform can also feed back the processing result to the monitoring processor to provide real-time fault diagnosis and early warning. Through real-time data transmission and processing, it can respond quickly at the early stage of the fault and reduce the impact of the fault on the engine operation.

[0063] Step S10, when the first judgment result indicates no, judging whether the remote monitoring platform and the monitoring processor establish a communication connection through the industrial Internet at a second moment, and obtaining a second judgment result, the second moment occurring after the first moment;

[0064] Specifically, in the case where the communication connection fails at the first moment, the monitoring processor attempts to establish a communication connection with the remote monitoring platform again at the second moment. The second moment is usually set to a time point after the first moment. If the communication connection is detected to be normal, the second judgment result is "yes"; otherwise, the second judgment result is "no". By repeatedly trying to establish a communication connection, the system's fault tolerance for temporary communication failures can be improved, and fault prediction interruptions caused by communication interruptions can be reduced.

[0065] Step S11, when the second judgment result indicates yes, obtain the second abnormal operation data of the engine collected by the monitoring processor during the offline period of the monitoring processor, the second predicted fault type of the engine determined by the monitoring processor based on the second abnormal operation data and the micro-fault classification model, and the third predicted fault type of the engine, and when the second predicted fault type and the third predicted fault type are different, update the micro-fault classification model to a fault classification model, the offline period is a time period after the first moment and before the second moment, the micro-fault classification model is a preset model of the monitoring processor, and the third predicted fault type is determined by the second abnormal operation data and the fault classification model.

[0066] Specifically, during the period when the remote monitoring platform and the monitoring processor do not establish a communication connection, the monitoring processor can still obtain the real-time operation data of the engine from the sensor. After the communication is restored, the remote monitoring platform will use the second moment as the starting point for communication restoration, and obtain the second abnormal operation data collected by the monitoring processor during the offline period between the first moment and the second moment. It can be understood that the real-time operation data obtained by the monitoring processor during the above-mentioned offline period includes the above-mentioned second abnormal operation data. Then the monitoring processor determines the second predicted fault type of the engine based on the second abnormal operation data and the locally preset micro-fault classification model. At the same time, the remote monitoring platform determines the third predicted fault type of the engine based on the second abnormal operation data and the fault classification model. If the second predicted fault type is different from the third predicted fault type, it means that there may be a deviation in the micro-fault classification model. At this time, the micro-fault classification model is updated to the fault classification model to improve the accuracy and reliability of the model. The updated micro-fault classification model will be synchronized with the fault classification model of the remote monitoring platform to ensure the consistency of the two models. By comparing local and remote fault prediction results, local model deviations can be discovered and corrected in a timely manner to improve model accuracy. By updating and synchronizing the model, data consistency between the monitoring processor and the remote monitoring platform can be ensured, thereby improving the overall performance of the system.

[0067] Exemplarily, when the above-mentioned fault classification model is established using a sparse probability algorithm based on a Bayesian prediction model, the micro-fault classification model can also be established using a sparse probability algorithm based on a Bayesian prediction model, so that in the step of updating the micro-fault classification model to a fault classification model, the updating of the micro-fault classification model can be achieved by adjusting the model parameters in the micro-fault classification model corresponding to the model parameters of the fault classification model.

[0068] As a possible implementation, the historical operation data includes a first historical signal sample set of the engine in a healthy state, healthy operation samples corresponding to the first historical signal sample set, a second historical signal sample set of the engine in a faulty state, and a historical fault type sample set corresponding to the second historical signal sample set. According to the historical operation data and the Bayesian prediction model, determining the fault classification model of the engine includes the following steps:

[0069] Step S21, determining a feature mapping according to the kernel function and the first preset formula, the feature mapping including a first feature mapping between the historical operation data and the healthy operation (normal operation) sample and a second feature mapping between the historical operation data and the historical fault type sample set;

[0070] Specifically, a suitable kernel function, such as a Gaussian kernel function or a polynomial kernel function, is selected to map the original signal data into a high-dimensional feature space. The selection of the kernel function has an important influence on the effect of feature mapping.

[0071] Specifically, the first preset formula is:

[0072]

[0073] Among them, K(x i , x j ) is the kernel function, x i is the i-th training sample, x j is the jth training sample, i and j are both positive integers, the training sample is any sample in the first historical signal sample set and the second historical signal sample set, x represents the position matrix, and φ(x) is the feature map.

[0074] The input of the kernel function is a feature vector of sample data (sample data corresponding to any one sample in the first historical signal sample set and the second historical signal sample set), and the output is a similarity measure between the two samples.

[0075] Based on the above steps, by mapping any two historical signal samples in the historical operation data (any two first historical signal samples in the first historical signal sample set, any two second historical signal samples in the second historical signal sample set, or any one first historical signal sample in the first historical signal sample set and any one second historical signal sample in the second historical signal sample set) to a high-dimensional feature space, a first feature map corresponding to a healthy operation sample can be generated, or a second feature map corresponding to one of the historical fault type samples in the historical fault type sample set can be generated. This process usually involves matrix operations and feature dimension expansion.

[0076] Step S22, constructing a Bayesian prediction model based on feature mapping and sparse Bayesian learning algorithm;

[0077] Specifically, a sparse Bayesian learning algorithm is used to model the data after feature mapping. The sparse Bayesian learning algorithm is:

[0078]

[0079] Among them, w i is the weight, δ is the noise that follows the N(0,σ) distribution, w represents the weight vector, t i It is label data, which is any sample in the healthy operation sample and historical fault type sample set.

[0080] Specifically, sparse Bayesian learning can automatically select important features, reduce feature dimensions, and improve the sparsity and computational efficiency of the model. Using the first feature map and the second feature map, a Bayesian prediction model is constructed. The Bayesian prediction model has strong adaptability and can automatically update model parameters during the accumulation of new data to continuously improve the performance of the model. This process involves the initialization of model parameters, the selection of prior probabilities, and the design of likelihood functions, etc., to ensure that the model can effectively capture the characteristic relationship between the healthy state and the fault state.

[0081] Step S23, training the Bayesian prediction model according to the Bayesian prediction model, the first historical signal sample set, the healthy operation sample, the second historical signal sample set and the historical fault type sample set to obtain a fault classification model.

[0082] Specifically, multiple samples in the healthy operation sample and the historical fault type sample set have the same probability distribution; the probability distribution has the following formula:

[0083]

[0084] Where t is t i The matrix of .

[0085] Specifically, the first historical signal sample set, healthy operation sample, second historical signal sample set and historical fault type sample set are integrated into a training data set. This process involves data cleaning, normalization and label assignment to ensure data quality. The Bayesian prediction model is trained using the integrated training data set. The training process involves steps such as maximizing the posterior probability, optimizing model parameters and cross-validation to ensure the model's fitting effect and generalization ability on the training data. The trained Bayesian prediction model is evaluated, including indicators such as accuracy, recall rate and F1 score, to ensure that the classification performance of the model meets the expected requirements. Based on the training and evaluation results, the final fault classification model is generated. The model can input new engine signal data and output the corresponding fault type prediction results to achieve real-time diagnosis and early warning of engine faults.

[0086] Furthermore, we use Gaussian prior probability distribution to constrain the parameters to prevent w and σ 2 Transition fitting, the formula is as follows:

[0087] Among them, α is an N+1-dimensional hyperparameter vector.

[0088] It can be seen that the engine fault classification model can be established by using a sparse probability algorithm based on Bayesian reasoning. It can not only perform fault classification judgment on normal data and abnormal data collected in real time, but also train and update the model, making the data model calculation results more accurate and obtaining better fault classification and diagnosis results.

[0089] In order to enable those skilled in the art to more clearly understand the technical solution of the present application, the implementation process of the engine fault prediction method of the present application will be described in detail below in combination with specific embodiments.

[0090] This embodiment relates to a specific engine fault prediction method, such as Figure 3 As shown, the following steps are included:

[0091] Step S100: First, the remote monitoring platform obtains engine signal samples, which include engine fault signal samples and normal operation signal samples. The engine signal samples are preprocessed to extract vibration characteristics and engine parameters from the engine signal samples.

[0092] The preprocessing is to perform scale normalization, and the calculation formula is:

[0093]

[0094] Among them, z is the scale-normalized data, x is the original data, is the mean of the original data, and δ is the variance of the original data.

[0095] The engine parameters include speed, air intake, injection pulse width, ignition advance angle, throttle opening, oxygen output voltage, engine load, water temperature and mixed gas. The vibration characteristics include mean, standard deviation, root square amplitude, root mean square value, peak value, waveform index, peak index, pulse index, margin index, skewness and kurtosis.

[0096] Specifically, the mean The calculation formula is: Among them, x n is the time domain sequence of the signal, n = 1, 2, ..., N, N is the number of sample points; standard deviation σ x The calculation formula is: Root amplitude x r The calculation formula is: RMS x rms The calculation formula is: Peak x p The calculation formula is: p =max|x(n)|; The calculation formula of the waveform indicator W is: The calculation formula of the peak index C is: The calculation formula of impulse index I is: The calculation formula of margin index L is: The calculation formula of skewness S is: The calculation formula of kurtosis K is:

[0097] Step S110: cleaning the vibration characteristics and engine parameters, removing abnormal data, extracting fault characteristics, and establishing an engine fault classification model based on the fault characteristics;

[0098] Specifically, fault feature extraction uses a kernel principal component analysis feature algorithm to perform dimensionality reduction processing on vibration features and engine parameters to extract fault features.

[0099] Among them, the kernel principal component analysis feature algorithm reduces the dimension of vibration features and engine parameters by first combining the vibration features of the vibration signal to be identified and the engine parameters to construct the feature parameter matrix X m×n , where m is the number of samples and n is the number of features, and then the nonlinear variation function ψ(x i ) The feature parameter matrix X m×n Map it into the high-dimensional space F, and then use the principal component analysis feature algorithm to perform feature dimensionality reduction.

[0100] Specifically, the covariance matrix of the feature space F is: T is the scale parameter.

[0101] The engine fault classification model is established using a sparse probability algorithm based on Bayesian reasoning. Among them, x i is the i-th training sample, as input, t i is the label data, as output, the algorithm formula is: Among them, w i is the weight, δ is the noise that follows the N(0,σ) distribution, x represents the position matrix, w represents the weight vector, K(x i , x j ) is the kernel function, φ(x) is composed of N×(N+1) order kernel functions, and the formula is: t i Subject to independent distribution, the sample probability expression is: Where t is t iThe matrix uses Gaussian prior probability distribution to constrain the parameters to prevent w and σ 2 Transition fitting, the formula is: Among them, α is an N+1-dimensional hyperparameter vector.

[0102] Step S120: using the engine fault signal samples to train and classify the engine fault classification model to obtain an engine fault classification result.

[0103] Step S130: Establish a fault classification database according to the engine fault classification result, store the engine fault signal sample in the corresponding fault classification database according to the classification result, and obtain the fault cause, repair suggestion and solution at the same time, and save them in the fault classification database according to the corresponding fault classification;

[0104] Step S140: installing a plurality of sensors and a monitoring processor on the engine, wherein the sensors collect vibration signals of the engine in operation and send the collected vibration signals to the monitoring processor;

[0105] The sensor collects vibration signals through the equal-angle resampling method. First, the instantaneous frequency is obtained using the Hilbert-Huang transform, then the instantaneous rotation angle is calculated, and finally the equal-angle resampling signal is obtained using cubic spline interpolation.

[0106] The equal-angle resampling method is to select appropriate upper and lower cutoff frequencies from the original signal according to the main order components, perform bandpass filtering, obtain the signal z(t), perform EMD decomposition on the signal z(t), and obtain the first IMF component, perform Hilbert transform on it to obtain the analytical signal, and then take the phase to obtain the original signal rotation angle α based on equal time interval sampling. When equal-angle resampling is performed by interpolation, the obtained rotation angle signal α is subjected to cubic spline interpolation, and then resampled at a preset frequency to obtain an equal-angle vibration signal, remove the preset amount data before and after, eliminate the endpoint effect of the cubic spline interpolation, and perform frequency domain analysis and demodulation analysis on the equal-angle resampled signal.

[0107] The Hilbert-Huang transform is used to first decompose the signal into intrinsic mode functions determined by its local characteristics, and then perform Hilbert spectrum analysis on each intrinsic mode function to obtain the corresponding Hilbert spectrum. Hilbert spectrum analysis is to perform Hilbert transform on the original signal h(t) to obtain its Hilbert transform. Then construct the analytical signal, the formula is: The phase formula of y(t) is: The envelope amplitude formula of y(t) is: By taking the derivative of the phase θ, we can get the instantaneous frequency of the signal h(t). The instantaneous frequency is combined with the envelope amplitude to obtain the time-frequency-amplitude three-dimensional spectrum array.

[0108] Step S150: The monitoring processor pre-processes, calculates and judges the vibration signal, and when abnormal vibration signal data appears, sends the abnormal vibration signal data to the remote monitoring platform;

[0109] The monitoring processor has a dual-mode working mode. The first working mode is an online mode for communicating and sending data with the remote monitoring platform. The second working mode is an offline mode in which a communication connection cannot be established with the remote monitoring platform. In the offline mode, the monitoring processor stores a micro-engine fault classification model and a micro-fault classification database. When the monitoring processor disconnects the communication connection with the remote monitoring platform, the monitoring processor can evaluate and diagnose the abnormal vibration signal data by itself. When the monitoring processor establishes a communication connection with the remote monitoring platform, the remote monitoring platform will send data regularly to update the micro-engine fault classification model and the micro-fault classification database.

[0110] The monitoring processor will save the vibration signals collected by the sensor within a certain period of time, and regularly compress and package the vibration signals and send them to the remote monitoring platform. After receiving the data, the remote monitoring platform will train the engine fault classification model with the vibration signals under normal engine conditions; after the engine fault classification model completes the fault identification and diagnosis, it will store the abnormal vibration signal data in the corresponding engine fault classification database according to the engine fault classification results. At the same time, the engine fault classification model will be trained and updated according to the data updates in the engine fault classification database at preset intervals.

[0111] Step S160: the remote monitoring platform uses the engine fault classification model to classify the abnormal vibration signal data, determines the class to which the fault signal of the abnormal vibration signal data belongs, and completes the identification and diagnosis of the fault;

[0112] The remote monitoring platform uses the engine fault classification model to classify the abnormal vibration signal data. By calculating the similarity between the fault features of the abnormal vibration signal data and the class to which each fault signal belongs in the engine fault classification model, the fault signal of the abnormal vibration signal data is determined according to the calculated similarity result to complete the fault identification. The similarity between the fault features and the class to which each fault signal belongs is calculated by first extracting the features of each historical fault signal in the class to which each fault signal belongs, and generating the feature vector of each historical fault signal.

[0113] For example, the fault signals belong to the following categories: A i×n , B p×n and C q×n , i, p and q are the number of historical fault signals in the class to which the fault signal belongs, and the feature vector of the historical fault signal is n-dimensional: XA = (xa1, xa2, ..., xa n)、XB=(xb1,xb2,…,xb n ) and XC=(xc1,xc2,…,xc n ), and then the fault features extracted from the abnormal vibration signal data X are used to generate the abnormal vibration signal data feature vector X = (x1, x2, ..., x n ), the abnormal vibration signal data X and the fault signal A i×n 、Fault signal belongs to category B p×n And the fault signal belongs to class C q×n The similarity calculation is performed on each historical fault signal in the , and the similarity vector S = (d1, d2, ..., d i )、S=(d1,d2,…,d p ) and S=(d1,d2,…,d q ), take the abnormal vibration signal data X and each fault signal belongs to class A i×n , B p×n and C q×n The mean values ​​of the similarities SA, SB and SC are compared. Taking the largest value among them, the abnormal vibration signal data X belongs to the fault category to which the fault signal belongs.

[0114] The vibration signal X to be identified and the class A to which each fault signal belongs i×n , B p×n and C q×n The mean calculation formula of the similarity is as follows: Among them, i, p and q are the fault signal class A i×n , B p×n and C q×n The number of historical fault signals in the system.

[0115] The similarity calculation between the fault characteristics of the abnormal vibration signal data and the historical fault signals within the category to which the fault signal belongs is calculated using the correlation coefficient similarity, and the calculation formula is as follows: Where X is the abnormal vibration signal data, and Y is the historical fault signal.

[0116] The fault identification and diagnosis is completed by determining the category to which the fault signal of the abnormal vibration signal data belongs.

[0117] Step S170: The remote monitoring platform extracts the fault cause and solution in the fault classification database according to the fault identification result, and sends the fault identification diagnosis result, fault cause, repair suggestion and solution to the monitoring processor;

[0118] It should be noted that after the engine fault classification model classifies the abnormal vibration signal data, when the engine fault classification result cannot find the corresponding fault cause and solution in the engine fault classification database, it will be solved by professionals, who will find the cause of the fault and formulate a solution. After the processing is completed, the cause of the fault, repair suggestions and solutions found by the professionals will be added to the fault classification database. Among them, in the step of adding the cause of the fault, repair suggestions and solutions found by the professionals into the database, a preset algorithm can be used to generate a temporary solution based on the cause of the fault, repair suggestions and solutions found by the professionals, and then added to the fault classification database through the call of the remote monitoring platform. It is understandable that professionals will also look for new causes and solutions, regularly modify and delete old causes and solutions in the fault classification database, and re-update and organize the data in the fault classification database.

[0119] Step S180: The monitoring processor is used to remind and guide the engine operator, and the operator repairs the engine according to the fault identification and diagnosis results, the cause of the fault, the repair suggestions and the solution.

[0120] From the above description, it can be seen that in this example, engine fault signal samples and normal operating signal samples are obtained through the remote monitoring platform, the signal samples are preprocessed, the vibration characteristics and engine parameters in the signal samples are extracted, the vibration characteristics and engine parameters are cleaned, the abnormal data are removed, and then the fault characteristics are extracted therefrom, an engine fault classification model is established according to the fault characteristics, the engine fault classification model is trained and classified using the engine fault signal samples to obtain the engine fault classification results, a fault classification database is established according to the engine fault classification results, the engine fault signal samples are stored in the corresponding fault classification database according to the classification results, and the cause of the fault, repair suggestions and solutions are obtained at the same time, and they are saved in the fault classification database according to the corresponding fault classification, multiple sensors and a monitoring processor are installed on the engine, and the sensor The sensor collects the vibration signal of the engine when it is running, and sends the collected vibration signal to the monitoring processor. The monitoring processor preprocesses, calculates and judges the vibration signal. When abnormal vibration signal data appears, the abnormal vibration signal data is sent to the remote monitoring platform. The remote monitoring platform uses the engine fault classification model to classify the abnormal vibration signal data, judges the class to which the fault signal of the abnormal vibration signal data belongs, and completes the identification and diagnosis of the fault. The remote monitoring platform extracts the cause and solution of the fault in the fault classification database according to the fault identification result, and sends the fault identification and diagnosis result, the cause of the fault, the repair suggestion and the solution to the monitoring processor. The monitoring processor reminds and guides the engine operator, and the staff repairs the engine according to the fault identification and diagnosis result, the cause of the fault, the repair suggestion and the solution.

[0121] The embodiment of the present application also provides an engine failure prediction device. It should be noted that the engine failure prediction device of the embodiment of the present application can be used to execute the engine failure prediction method provided by the embodiment of the present application. The device is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions that have been made will not be repeated. As used below, the term "module" can implement a combination of software and / or hardware for a predetermined function. Although the device described in the following embodiments is preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0122] The following is an introduction to the engine fault prediction device provided in the embodiments of the present application.

[0123] Figure 4 Schematic diagram of an engine fault prediction device according to an embodiment of the present application. Sensors are deployed on the engine, and the sensors establish a communication connection with the remote monitoring platform through the industrial Internet, such as Figure 4 As shown, the device includes: a first acquisition module 10, a construction module 20, a second acquisition module 30 and a prediction module 40.

[0124] A first acquisition module 10 is used to acquire historical operating data of the engine;

[0125] Specifically, the acquisition of historical operation data is achieved through multiple sensors installed on the engine. The historical operation data may be vibration data, and the sensor may collect vibration data through an equal-angle resampling method, thereby ensuring the accuracy and reliability of the data.

[0126] A construction module 20 is used to determine a fault classification model of the engine based on historical operation data and a Bayesian prediction model, where the Bayesian prediction model is a preset model of the remote monitoring platform;

[0127] Specifically, after acquiring the historical operating data of the engine, the Bayesian prediction model can be trained and classified by inputting the historical operating data into the Bayesian prediction model preset in the remote monitoring platform, thereby determining a fault classification model for the engine.

[0128] Optionally, before using the historical operation data to train the Bayesian prediction model, the historical operation data is preprocessed, and then the Bayesian prediction model is trained based on the preprocessed historical operation data. The preprocessing may include but is not limited to extracting vibration features and engine parameters from signal samples of the historical operation data, cleaning the vibration features and engine parameters, removing abnormal data, and then extracting fault features therefrom.

[0129] A second acquisition module 30, for acquiring engine operation data collected by the sensor during real-time operation of the engine, the engine operation data including first abnormal operation data;

[0130] Specifically, during the operation of the engine, multiple sensors installed on the engine can continuously collect the operation data of the engine, thereby obtaining the engine operation data during the real-time operation of the engine. In particular, when the engine is in an abnormal operation state, the sensor will collect the first abnormal operation data, which reflects the current fault characteristics of the engine.

[0131] The prediction module 40 is used to determine a first predicted fault type of the engine according to the first abnormal operation data and the fault classification model.

[0132] Specifically, after receiving the first abnormal operation data collected by the sensor, the remote monitoring platform analyzes the first abnormal operation data using the established fault classification model.

[0133] Optionally, before the fault classification model uses the first abnormal operation data for prediction, the first abnormal operation data is preprocessed, and then the Bayesian prediction model is trained based on the preprocessed first abnormal operation data.

[0134] As a possible implementation method, the historical operation data includes a historical fault type sample set of the engine in a fault state and a fault classification database, the historical fault type sample set includes at least one historical fault type, the fault classification database includes a historical solution sample set corresponding to the historical fault type sample set, the historical solution sample set includes at least one historical solution, and the historical solution is used to solve the fault problem corresponding to the historical fault type. The device also includes the following: a first judgment module, a first determination module and a first update module.

[0135] A first judgment module is used to judge whether the first predicted fault type belongs to one of the historical fault types in the historical fault type sample set, and obtain a second judgment result;

[0136] A first determination module is used to determine a first target solution based on the historical fault type in the historical fault type sample set that is the same as the first predicted fault type and the fault classification database when the second judgment result indicates yes, and the first target solution is used to solve the fault problem corresponding to the first predicted fault type;

[0137] The first updating module is used to call the second target solution according to the first predicted fault type when the second judgment result indicates no, and configure the first predicted fault type and the second target solution in the fault classification database to update the fault classification database.

[0138] As a possible implementation, a monitoring processor is deployed on the engine, and the monitoring processor is communicatively connected with the sensor. The device also includes: a second judgment module, an execution module, a third judgment module and a second update module.

[0139] A second judgment module is used to judge whether the remote monitoring platform and the monitoring processor establish a communication connection through the industrial Internet at the first moment, and obtain a first judgment result;

[0140] an execution module, configured to execute the second acquisition module and the prediction module when the first judgment result indicates yes;

[0141] A third judgment module is used to judge whether the remote monitoring platform and the monitoring processor establish a communication connection through the industrial Internet at a second moment when the first judgment result indicates no, and obtain a second judgment result, the second moment occurring after the first moment;

[0142] The second update module is used to obtain the second abnormal operation data of the engine collected by the monitoring processor during the offline period of the monitoring processor, the second predicted fault type of the engine determined by the monitoring processor based on the second abnormal operation data and the micro-fault classification model, and the third predicted fault type of the engine when the second judgment result indicates yes, and update the micro-fault classification model to a fault classification model when the second predicted fault type and the third predicted fault type are different. The offline period is a time period after the first moment and before the second moment, the micro-fault classification model is a preset model of the monitoring processor, and the third predicted fault type is determined by the second abnormal operation data and the fault classification model.

[0143] As a possible implementation method, the historical operation data includes a first historical signal sample set of the engine in a healthy state, healthy operation samples corresponding to the first historical signal sample set, a second historical signal sample set of the engine in a faulty state, and a historical fault type sample set corresponding to the second historical signal sample set. The construction module includes: a determination submodule, a construction submodule and a training submodule.

[0144] A determination submodule is used to determine a feature map according to a kernel function and a first preset formula, wherein the feature map includes a first feature map between historical operation data and healthy operation samples and a second feature map between historical operation data and a historical fault type sample set; wherein the first preset formula is:

[0145]

[0146] Among them, K(x i , x j ) is the kernel function, x i is the i-th training sample, x j is the jth training sample, i and j are both positive integers, the training sample is any sample in the first historical signal sample set and the second historical signal sample set, x represents the position matrix, and φ(x) is the feature map.

[0147] The submodule is used to construct a Bayesian prediction model based on feature mapping and sparse Bayesian learning algorithm; wherein the sparse Bayesian learning algorithm is:

[0148]

[0149] Among them, w i is the weight, δ is the noise that follows the N(0,σ) distribution, w represents the weight vector, t i It is label data, which is any sample in the healthy operation sample and historical fault type sample set.

[0150] The training submodule is used to train the Bayesian prediction model according to the Bayesian prediction model, the first historical signal sample set, the healthy operation sample, the second historical signal sample set and the historical fault type sample set to obtain a fault classification model. Among them, the healthy operation sample and the multiple samples in the historical fault type sample set have the same probability distribution; the probability distribution has the following formula:

[0151]

[0152] Where t is t i The matrix of .

[0153] The engine fault prediction device includes a processor and a memory. The first acquisition module 10, the construction module 20, the second acquisition module 30 and the prediction module 40 are all stored in the memory as program units, and the processor executes the program units stored in the memory to implement corresponding functions. The modules are all located in the same processor; or, the modules are located in different processors in any combination.

[0154] The processor includes a kernel, which retrieves the corresponding program unit from the memory. One or more kernels can be set, and the kernel parameters can be adjusted to solve the problem that the engine is monitored by an offline periodic test method in the prior art, which has a large idle period, resulting in the delay in controlling the timing of major repairs, which can easily cause serious mechanical accidents and increase maintenance costs.

[0155] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0156] An embodiment of the present invention provides a computer-readable storage medium, which includes a stored program, wherein when the program is running, the device where the computer-readable storage medium is located is controlled to execute any engine fault prediction method.

[0157] An embodiment of the present invention provides an electronic device, characterized in that it includes: one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and the one or more programs include a method for executing any one of the engine fault prediction methods.

[0158] Obviously, those skilled in the art should understand that the above modules or steps of the present invention can be implemented by a general computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be executed in a different order than here, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present invention is not limited to any specific combination of hardware and software.

[0159] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0160] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0161] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0162] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0163] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0164] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0165] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0166] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0167] From the above description, it can be seen that the above embodiments of the present application achieve the following technical effects:

[0168] 1) This application provides an engine fault prediction method. When the remote monitoring platform obtains historical operation data, it can train a Bayesian prediction model according to the fault characteristics in the historical operation data, so as to determine the fault classification model of the engine. On this basis, since the sensors deployed on the engine establish a communication connection with the remote monitoring platform through the industrial Internet, the remote monitoring platform can obtain the engine operation data collected by the sensor during the real-time operation of the engine, and when the engine operation data includes the first abnormal operation data, the trained fault classification model can be used to predict the fault type and complete the fault identification diagnosis. It can be seen that this application can control the operation status of the engine in real time through the analysis of the remote monitoring platform, and can predict the engine failure, which solves the problem that the engine is monitored by the offline regular test method in the prior art, and there is a large gap period, resulting in inaccurate prediction of engine failure. In addition, through this application, maintenance personnel can make accurate diagnosis of the fault at the first time and can intervene in the maintenance based on the first predicted fault type of the engine, extend the engine overhaul cycle, enhance the pre-maintenance plan and timely control the overhaul time to avoid the occurrence of serious mechanical accidents, thereby reducing maintenance costs.

[0169] 2) The present application provides an engine fault prediction device, which includes: a first acquisition module, a construction module, a second acquisition module and a prediction module. The first acquisition module is used to acquire the historical operation data of the engine; the construction module is used to determine the fault classification model of the engine according to the historical operation data and the Bayesian prediction model, and the Bayesian prediction model is a preset model of the remote monitoring platform; the second acquisition module is used to acquire the engine operation data collected by the sensor during the real-time operation of the engine, and the engine operation data includes the first abnormal operation data; the prediction module is used to determine the first predicted fault type of the engine according to the first abnormal operation data and the fault classification model. Through the analysis of the remote monitoring platform, the device can control the operation status of the engine in real time, and can predict the engine failure, which solves the problem that the engine is monitored by the offline regular test method in the prior art, and there is a large gap period, resulting in inaccurate prediction of engine failure. In addition, through this application, maintenance personnel can make accurate diagnosis at the first time of the fault and can intervene in the maintenance based on the first predicted fault type of the engine, extend the engine overhaul cycle, enhance the pre-maintenance plan and timely control the overhaul time to avoid the occurrence of serious mechanical accidents, thereby reducing maintenance costs.

[0170] The above are only preferred embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. An engine failure prediction method, characterized in that: The engine is equipped with a sensor, and the sensor is connected to the remote monitoring platform through the industrial Internet. The engine fault prediction method includes the following steps: S1, obtaining historical operating data of the engine; S2. Determine a fault classification model of the engine according to the historical operation data and a Bayesian prediction model, wherein the Bayesian prediction model is a preset model of the remote monitoring platform; S3, acquiring engine operation data collected by the sensor during the real-time operation of the engine, wherein the engine operation data includes first abnormal operation data; S4. Determine a first predicted fault type of the engine according to the first abnormal operation data and the fault classification model.

2. The engine failure prediction method according to claim 1, characterized in that: The historical operation data includes a historical fault type sample set of the engine in a fault state and a fault classification database, the historical fault type sample set includes at least one historical fault type, the fault classification database includes a historical solution sample set corresponding to the historical fault type sample set, the historical solution sample set includes at least one historical solution, and the historical solution is used to solve the fault problem corresponding to the historical fault type. The engine fault prediction method also includes: Determine whether the first predicted fault type belongs to one of the historical fault types in the historical fault type sample set, and obtain a second determination result; When the second judgment result indicates yes, determining a first target solution according to the historical fault type in the historical fault type sample set that is the same as the first predicted fault type and the fault classification database, wherein the first target solution is used to solve the fault problem corresponding to the first predicted fault type; In the case where the second judgment result indicates no, a second target solution is called according to the first predicted fault type, and the first predicted fault type and the second target solution are configured in the fault classification database to update the fault classification database.

3. The engine failure prediction method according to claim 1, characterized in that: The engine is provided with a monitoring processor, the monitoring processor is in communication connection with the sensor, and the engine fault prediction method further comprises: Determine whether the remote monitoring platform and the monitoring processor establish a communication connection through the industrial Internet at a first moment, and obtain a first determination result; If the first judgment result indicates yes, execute step S3 and step S4; When the first judgment result indicates no, determining whether the remote monitoring platform and the monitoring processor establish a communication connection through the industrial Internet at a second moment, and obtaining a second judgment result, the second moment occurring after the first moment; When the second judgment result indicates yes, the second abnormal operation data of the engine collected by the monitoring processor during the offline period of the monitoring processor, the second predicted fault type of the engine determined by the monitoring processor based on the second abnormal operation data and the micro-fault classification model, and the third predicted fault type of the engine are obtained, and when the second predicted fault type and the third predicted fault type are different, the micro-fault classification model is updated to the fault classification model, the offline period is the time period after the first moment and before the second moment, the micro-fault classification model is a preset model of the monitoring processor, and the third predicted fault type is determined by the second abnormal operation data and the fault classification model.

4. The engine failure prediction method according to any one of claims 1 to 3, characterized in that: The historical operation data includes a first historical signal sample set of the engine in a healthy state, healthy operation samples corresponding to the first historical signal sample set, a second historical signal sample set of the engine in a faulty state, and a historical fault type sample set corresponding to the second historical signal sample set. According to the historical operation data and the Bayesian prediction model, a fault classification model of the engine is determined, including: Determine a feature map according to a kernel function and a first preset formula, wherein the feature map includes a first feature map between the historical operation data and the healthy operation sample and a second feature map between the historical operation data and the historical fault type sample set; Constructing the Bayesian prediction model according to the feature mapping and the sparse Bayesian learning algorithm; The Bayesian prediction model is trained according to the Bayesian prediction model, the first historical signal sample set, the healthy operation samples, the second historical signal sample set and the historical fault type sample set to obtain the fault classification model.

5. The engine failure prediction method according to claim 4, characterized in that: The first preset formula is: Among them, K(x i , x j ) is the kernel function, x i is the i-th training sample, x j is the jth training sample, i and j are both positive integers, the training sample is any one of the first historical signal sample set and the second historical signal sample set, x represents a position matrix, and φ(x) is the feature map; The sparse Bayesian learning algorithm is: Among them, w i is the weight, δ is the noise that follows the N(0,σ) distribution, w represents the weight vector, t i is label data, and the label data is any one of the healthy operation samples and the historical fault type sample set.

6. The engine failure prediction method according to claim 5, characterized in that: The healthy operation sample and the multiple samples in the historical fault type sample set have the same probability distribution; The probability distribution has the following formula: Wherein, t is the i The matrix of .

7. The engine failure prediction method according to any one of claims 1 to 3, characterized in that: The historical operation data is vibration data, and the sensor collects the vibration data by an equal-angle resampling method.

8. An engine failure prediction device, characterized in that: The engine is equipped with a sensor, and the sensor is connected to the remote monitoring platform through the industrial Internet. The engine fault prediction device includes the following steps: A first acquisition module, used for acquiring historical operation data of the engine; A construction module, used to determine a fault classification model of the engine according to the historical operation data and a Bayesian prediction model, wherein the Bayesian prediction model is a preset model of the remote monitoring platform; A second acquisition module, configured to acquire engine operation data collected by the sensor during the real-time operation of the engine, wherein the engine operation data includes first abnormal operation data; A prediction module is used to determine a first predicted fault type of the engine according to the first abnormal operation data and the fault classification model.

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 engine fault prediction method according to any one of claims 1 to 7.

10. An electronic device, characterized in that: include: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and the one or more programs include a method for executing the engine fault prediction method described in any one of claims 1 to 7.

Images