Engine health state recognition method, device, equipment, medium and program product

Abstract

The embodiment of the invention discloses an engine health state recognition method, device and equipment, a medium and a program product. Comprising the following steps: acquiring engine operation data in a to-be-detected time period, and dividing the engine operation data into at least two groups of batch data according to a preset batch division rule; inputting each batch of data into a pre-constructed health trend model, and determining a prediction result of each batch of data; wherein the health trend model is a feature model which is constructed according to engine key operation variables obtained under the stable operation condition and has a nonlinear degradation trend; determining abnormal batch data according to each batch of data and the prediction result corresponding to each batch of data; and the deviation amplitude of each abnormal batch of data is determined, and the health state of the engine is determined according to the deviation rate of each deviation amplitude in the time dimension. The engine is prevented from directly entering a serious fault under the condition of no early warning, and the engine maintenance efficiency and the locomotive driving safety are improved.

Description

Technical Field

[0001] This invention relates to the field of equipment health management technology, and in particular to a method, device, equipment, medium and program product for identifying engine health status. Background Technology

[0002] With the booming development of my country's transportation industry, the engine, as the core power unit of locomotives, directly affects the safety and reliability of the entire vehicle. As vehicle operation becomes increasingly complex and the operating environment more diverse, assessing the engine's health and monitoring degradation trends in a timely manner is crucial for ensuring driving safety and improving operational reliability.

[0003] Traditional health monitoring methods typically rely on fixed threshold alarms or expert judgment. However, these methods struggle to accurately identify potential performance degradation, early sub-health conditions, and fault signs when faced with complex operating environments with multiple operating conditions and variables. Engine health status identification using machine learning methods often requires training with large amounts of labeled data samples, resulting in limited generalization ability, complex model structures, poor interpretability, and difficulty in direct application to real-world operation and maintenance scenarios. Summary of the Invention

[0004] This invention provides a method, device, equipment, medium, and program product for identifying engine health status. It constructs a health trend model for potential nonlinear degradation of the engine and identifies potential sub-health states of the engine by comparing the predicted deviations of multiple batches of data. This reduces the impact of data fluctuations caused by environmental factors on the accuracy of the identification results and reduces the impact of expert subjective judgment on the consistency and stability of the identification results. It improves the accuracy and stability of identifying sub-health states, avoids the engine from directly entering a serious fault without warning, and improves engine maintenance efficiency and vehicle driving safety.

[0005] In a first aspect, embodiments of the present invention provide a method for identifying engine health status, including:

[0006] Acquire engine operating data within the time period to be tested, and divide the engine operating data into at least two batches according to a preset batch division rule;

[0007] Each batch of data is input into a pre-built health trend model to determine the prediction results for each batch of data; the health trend model is a feature model with a non-linear degradation trend, constructed based on key engine operating variables obtained under stable operating conditions.

[0008] Abnormal batch data are determined based on the data of each batch and the corresponding prediction results of each batch.

[0009] Determine the deviation range of each abnormal batch of data, and determine the engine health status based on the deviation rate of each deviation range over time.

[0010] Secondly, embodiments of the present invention also provide an engine health status identification device, comprising:

[0011] The data acquisition module is used to acquire engine operating data within the time period to be detected, and divide the engine operating data into at least two batches of data according to a preset batch division rule;

[0012] The prediction result determination module is used to input each batch of data into a pre-built health trend model to determine the prediction result of each batch of data; wherein, the health trend model is a feature model with a non-linear degradation trend, constructed based on key engine operating variables obtained under stable operating conditions.

[0013] The abnormal data identification module is used to identify abnormal batch data based on each batch of data and the corresponding prediction results.

[0014] The health status determination module is used to determine the deviation range of each abnormal batch of data and determine the engine health status based on the deviation rate of each deviation range in the time dimension.

[0015] Thirdly, embodiments of the present invention also provide an engine health status identification device, comprising:

[0016] At least one processor; and a memory communicatively connected to the at least one processor;

[0017] The memory stores a computer program that can be executed by at least one processor, which is then executed by the at least one processor to enable the at least one processor to perform the engine health status identification method provided in any embodiment of the present invention.

[0018] Fourthly, embodiments of the present invention also provide a storage medium containing computer-executable instructions, which, when executed by a computer processor, are used to perform the engine health status identification method of any embodiment of the present invention.

[0019] Fifthly, embodiments of the present invention also provide a computer program product, including a computer program, which, when executed by a processor, is used to perform the engine health status identification method of any embodiment of the present invention.

[0020] This invention provides a method, apparatus, device, medium, and program product for identifying engine health status. It acquires engine operating data within a specified time period and divides this data into at least two batches according to a preset batching rule. Each batch is then input into a pre-constructed health trend model to determine the prediction result for each batch. The health trend model is a feature model with a non-linear degradation trend, constructed based on key engine operating variables acquired under stable operating conditions. Abnormal batches are identified based on each batch and its corresponding prediction result. The deviation magnitude of each abnormal batch is determined, and the engine health status is determined based on the deviation rate of each deviation magnitude over time. By employing this technical solution, the acquired engine operating data within the specified time period is divided into batches, and each batch is input into a pre-constructed health trend model that satisfies a non-linear degradation trend to determine whether any batches exhibit operational abnormalities. Furthermore, after identifying abnormal batches, the engine's health status is not directly determined to be abnormal; instead, the deviation magnitude changes of different abnormal batches over time are compared to determine whether the engine is in a sub-healthy state before reaching a fault state. The process utilizes a health trend model that reflects the nonlinear degradation trend of the engine and is constructed based on key engine operating characteristics. The model performs judgments based on multiple batches of data and does not rely on human experience. This reduces the impact of data fluctuations caused by environmental factors on the accuracy of the identification results and reduces the impact of subjective expert judgments on the consistency and stability of the identification results. It improves the accuracy and stability of identifying sub-health states, prevents the engine from directly entering a serious fault without warning, and improves engine maintenance efficiency and locomotive driving safety.

[0021] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a flowchart of an engine health status identification method provided in Embodiment 1 of the present invention;

[0024] Figure 2This is a flowchart of an engine health status identification method provided in Embodiment 2 of the present invention;

[0025] Figure 3 This is a schematic diagram of the structure of an engine health status identification device provided in Embodiment 3 of the present invention;

[0026] Figure 4 This is a schematic diagram of the structure of an engine health status identification device provided in Embodiment 4 of the present invention. Detailed Implementation

[0027] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0028] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0029] Example 1

[0030] Figure 1 This is a flowchart illustrating an engine health status identification method according to Embodiment 1 of the present invention. This embodiment is applicable to situations where engine operating data is analyzed to promptly identify a sub-healthy state—either in the early stages of performance degradation or before reaching a fault threshold. The method can be executed by an engine health status identification device, which can be configured within an engine health status identification equipment. Optionally, the engine health status identification device can be an electronic device, such as a laptop, desktop computer, or vehicle-mounted computer; this embodiment does not impose any limitations on this. Optionally, the engine requiring engine health status identification in this embodiment can be a diesel engine in a railway diesel locomotive, or an engine in other locomotives; this embodiment does not impose any limitations on this.

[0031] like Figure 1 As shown, this embodiment of the invention provides a method for identifying engine health status, specifically including the following steps:

[0032] S101. Obtain engine operating data within the time period to be detected, and divide the engine operating data into at least two batches of data according to the preset batch division rules.

[0033] In this embodiment, the time period to be detected can be specifically understood as a selected time period during which the engine operating status needs to be monitored and detected. During this time period, data related to engine operation needs to be collected in order to determine the engine's operating health status based on the collected data.

[0034] In this embodiment, engine operating data can be specifically understood as data related to engine operation during the engine's operation. For example, engine operating data may include speed, pressure, temperature, load, operating time, number of starts, fuel flow, intake system parameters, and exhaust system parameters, etc., and this embodiment of the invention does not impose any limitations on these.

[0035] In this embodiment, the preset batch division rule can be specifically understood as a rule that is pre-set according to the actual situation, used to divide the engine operation data acquired within a period of time into batches according to the set logic.

[0036] In this embodiment, batch data can be specifically understood as a set of multiple engine operation data obtained by dividing engine operation data according to a preset batch division rule.

[0037] Specifically, when it is necessary to detect the health status of the engine, a time period can be selected as the detection period within the time frame during which the engine is running, and engine operation data generated during the operation of the engine within the detection period can be acquired. To improve the accuracy of engine health status identification and avoid errors that may be caused by identifying data from a single time period, this embodiment of the invention uses a preset batch division rule to divide the acquired engine operation data into at least two batches of data according to certain logical rules, so that the engine health status can be more accurately identified subsequently based on the relationship between different batches of data.

[0038] S102. Input the data of each batch into the pre-built health trend model to determine the prediction results of each batch of data.

[0039] Among them, the health trend model is a feature model with a nonlinear degradation trend, constructed based on key engine operating variables obtained under stable operating conditions.

[0040] In this embodiment, the health trend model can be specifically understood as a mathematical model used to show the changing trend of typical relationships between key vectors of engine operating data when the engine is running under healthy and stable conditions.

[0041] In this embodiment, stable operation can be specifically understood as the engine's operation when it is fault-free and not experiencing performance degradation. Key engine operating variables can be specifically understood as engine operating data types selected from various available engine operating data, corresponding to the angle from which the engine health status needs to be assessed.

[0042] Specifically, each batch of data is input into a pre-built health trend model based on the required health status assessment angle, and the health trend model outputs the prediction results corresponding to the batch data.

[0043] Understandably, depending on the different health status assessment angles required, the key engine operating variables used in the construction of the health trend model will be different, and the key engine operating variables used as inputs and outputs of the health trend model will also be different. The corresponding engine operating data in the batch data can be input into the health trend model according to the requirements of the actual health trend model used, so as to obtain prediction results that are consistent with the output capability of the health trend model.

[0044] S103. Determine the abnormal batch data based on the data of each batch and the prediction results corresponding to each batch data.

[0045] In this embodiment, abnormal batch data can be specifically understood as batch data that does not conform to the healthy operation trend of the engine.

[0046] Specifically, since the prediction results obtained by the health trend model are the engine operation data that the batch data should generate when the engine is running in a healthy state, and the batch data also includes the engine operation data generated by the engine in the current operating state, the prediction results can be compared with the corresponding engine operation data in the batch data. If there is a difference between the two, it can be said that the batch data does not meet the engine operating state trend, and at this time the batch data can be identified as abnormal batch data.

[0047] S104. Determine the deviation range of each abnormal batch of data, and determine the engine health status based on the deviation rate of each deviation range in the time dimension.

[0048] In this embodiment, the deviation magnitude can be specifically understood as the degree of deviation of abnormal batch data from the healthy operating trend. The deviation rate can be specifically understood as the rate of change of the deviation magnitude of different abnormal batch data over time. The engine health status can be specifically understood as an operating state used to indicate whether the engine is operating normally, or is in the early stage of performance degradation but has not yet reached the level of failure.

[0049] Specifically, the trend curve of each abnormal batch of data is compared with the trend curve under healthy engine operation to obtain the deviation of each abnormal batch of data. Then, the abnormal batches of data are sorted according to the time dimension, and the abnormal amplitude of each abnormal batch is determined in the time dimension to obtain the deviation rate. Then, based on the deviation rate, it is determined whether the engine is in the early stage of performance decline but has not reached the fault level, and the corresponding engine health status is determined.

[0050] The technical solution of this embodiment acquires engine operating data within the time period to be tested and divides the engine operating data into at least two batches according to a preset batch division rule. Each batch of data is then input into a pre-constructed health trend model to determine the prediction result for each batch. The health trend model is a feature model with a non-linear degradation trend, constructed based on key engine operating variables acquired under stable operating conditions. Abnormal batch data is determined based on each batch of data and its corresponding prediction result. The deviation magnitude of each abnormal batch of data is determined, and the engine health status is determined based on the deviation rate of each deviation magnitude over time. By adopting the above technical solution, the acquired engine operating data within the time period to be tested is divided into batches, and each batch of data is input into a pre-constructed health trend model that satisfies a non-linear degradation trend to determine whether there are any operational abnormalities in each batch of data. Furthermore, after identifying abnormal batch data, the engine health status is not directly determined to be abnormal; instead, the deviation magnitude changes of different abnormal batch data over time are compared again to determine whether the engine is in a sub-healthy state that has not yet reached a fault state. The process utilizes a health trend model that reflects the nonlinear degradation trend of the engine and is constructed based on key engine operating characteristics. The model performs judgments based on multiple batches of data and does not rely on human experience. This reduces the impact of data fluctuations caused by environmental factors on the accuracy of the identification results and reduces the impact of subjective expert judgments on the consistency and stability of the identification results. It improves the accuracy and stability of identifying sub-health states, prevents the engine from directly entering a serious fault without warning, and improves engine maintenance efficiency and locomotive driving safety.

[0051] Example 2

[0052] Figure 2This is a flowchart of an engine health status identification method provided in Embodiment 2 of the present invention. The present invention further optimizes the above-mentioned optional technical solutions by acquiring stable engine operation data under stable engine operation conditions before putting the engine health status detection into operation, and selecting independent variables and response variables that can be used to construct a health trend model according to actual needs. At the same time, content different from the independent variables and response variables can also be selected as operating condition penalty terms, so that the health trend model constructed based on this can suppress the interference of samples on the fitting results under specific environments, and obtain a health trend model that can reflect the nonlinear degradation trend of the key engine operating variables corresponding to the independent variables and response variables during engine operation. Furthermore, when performing engine health status detection, a pre-built health trend model can be used to predict the input batch data in a manner consistent with the data type of the response variable, obtaining the corresponding prediction results. These prediction results are then compared with the data in the batch that are consistent with the data type of the response variable to obtain the corresponding fitting residuals. These fitting residuals can not only be used to identify abnormal batch data, but also to determine the deviation rate between two adjacent abnormal batch data in the time dimension. Only when the deviation rate continues to rise in the time dimension is the engine health status considered to be sub-healthy. This avoids the impact of data fluctuations caused by environmental factors on the accuracy of identification when determining the engine health status in a single identification, improving the accuracy and stability of sub-health status identification, preventing the engine from directly entering a serious fault without warning, and improving engine maintenance efficiency and vehicle driving safety.

[0053] like Figure 2 As shown in the figure, an engine health status identification method provided by an embodiment of the present invention specifically includes the following steps:

[0054] S201. Obtain stable engine operating data under stable operating conditions, and determine multiple key engine operating variables based on the data types contained in the stable engine operating data.

[0055] Among the key operating variables of an engine are at least one of the following: engine speed, pressure, temperature, and load.

[0056] In this embodiment, stable engine operating data can be specifically understood as engine operating data selected from the vehicle's historical operating data, indicating that the engine is operating stably. Data types can be specifically understood as types used to indicate different operating states of the engine during operation, such as speed, pressure, temperature, load, operating time, number of starts, fuel flow, intake system parameters, and exhaust system parameters; each of these belongs to a data type within the engine operating data.

[0057] Specifically, from the historical operating data generated by vehicle operation, engine operating data under stable operating conditions is selected as stable engine operating data. Furthermore, since stable engine operating data can contain various types of engine operating data, several data types that can typically reflect the engine's health status can be selected as key engine operating variables, depending on the focus of attention regarding engine health.

[0058] S202. Select at least two key operating variables from each engine as independent and response variables respectively, and select one from the remaining key operating variables of each engine as the operating condition penalty term.

[0059] In this embodiment, the independent variable can be specifically understood as the data type corresponding to the controllable operating condition in the health trend model to be constructed. The response variable can be specifically understood as the data type in the health trend model to be constructed that will be affected by the independent variable and can serve as a key health indicator reflecting the engine's operating health status. It is understood that in this embodiment of the invention, there may be one or more independent variables and one response variable. The operating condition penalty term can be specifically understood as the data type corresponding to the constraint condition set for the health trend model to be constructed based on a specific environment or operating condition.

[0060] Specifically, based on the type of engine health status to be assessed, the main data types required for assessing that type of engine health status are determined. Key engine operating variables corresponding to controllable operating conditions are identified as independent variables, and those that can serve as key health indicators reflecting the engine's operational health status are identified as response variables. Furthermore, from the remaining key engine operating variables excluding the independent and response variables, one key engine operating variable corresponding to a specific environment or operating condition is selected as an operating condition penalty term to suppress the impact of samples under that specific environment or operating condition on the construction of the health trend model.

[0061] S203. Construct an initial health trend model based on independent and response variables, and construct an objective function using the working condition penalty term as a stability constraint.

[0062] In this embodiment, the initial health trend model can be specifically understood as a health trend model that has not yet been fitted and constrained by training samples.

[0063] In some examples, the health trend model can be constructed using a polynomial regression function. In this embodiment of the invention, rotational speed N is used as the independent variable and pressure P is used as the response variable for function fitting. To enhance the stability of the health trend model under different temperature conditions and to suppress the interference of samples under specific environments on the fitting results, temperature T is introduced as a condition penalty term, and the deviation of temperature T is used as a stability constraint factor to construct the objective function. In this embodiment of the invention, the nonlinear degradation trend of pressure over time is considered. Taking a cubic polynomial fitting function as an example, the initial health trend model is constructed as follows:

[0064]

[0065] in, Let N be the model prediction value for the i-th batch of samples. i Let β be the rotational speed value for this batch, where β = [β0, β1, β2, β3]. T The vector of model parameters to be solved is, as it can be understood, is the vector of model parameters to be solved, since the initial health trend model is constructed using a cubic polynomial fitting function as an example. Therefore, the vector of model parameters to be solved should contain four model parameters to be solved.

[0066] Following the example above, since temperature T is introduced as a working condition penalty term, that is, temperature T can be used as a working condition stability constraint factor introduced in the fitting process of the health trend model, a temperature baseline value T can be defined at this time. ref Therefore, the objective function containing regularization and temperature penalty terms is constructed as follows:

[0067]

[0068] Understandably, the objective function described above is calculated to minimize its value in order to achieve an accurate fit to system stress and improve model robustness. Among these, The sum of squared residuals (least squares term) between stress and model predictions is used to measure the fitting error of the health trend model; the smaller the sum, the more accurate the prediction. This is the L2 regularization term, used to constrain model parameters and prevent overfitting. λ1 and λ2 are temperature deviation penalty terms used to reduce the interference of abnormal temperatures on model fitting; λ1 and λ2 are regularization weight coefficients, which can be adjusted based on experience or cross-validation.

[0069] In this embodiment of the invention, by constructing the initial health trend model and objective function as described above, the model can effectively fit the nonlinear response relationship between pressure and speed while dynamically suppressing the adverse effects on the fitting results under atypical operating conditions (such as high temperature), thereby significantly reducing the risk of overfitting under boundary conditions. This modeling method not only accurately characterizes the healthy operating characteristics of pressure changing with speed, but also implements modeling constraints on differences in the operating environment through a temperature deviation penalty mechanism, enhancing the robustness and engineering adaptability of the model under varying operating conditions.

[0070] Understandably, this modeling approach can also be applied to the construction of health trend models under other key engine operating variables, and is suitable for trend modeling and health benchmark extraction of various typical engine operating data, possessing good scalability and industrial application value.

[0071] S204. Extract data corresponding to independent variables, response variables, and operating condition penalty terms from stable engine operation data to construct a training sample set.

[0072] Specifically, after clarifying the data types corresponding to the independent variables, response variables, and operating condition penalty terms, stable engine operation data of the corresponding data types are extracted from the stable engine operation data. A set of stable engine operation data corresponding to a set of independent variables, response variables, and operating condition penalty terms acquired at the same time is used as a training sample to construct a training sample set containing multiple training samples.

[0073] S205. Train the initial health trend model using the training sample set and objective function to obtain the pre-constructed health trend model.

[0074] Specifically, each training sample in the training sample set is input into the initial health trend model, and the initial health trend model is trained using the objective function as the loss function until the preset fitting conditions are met or all training samples in the training sample set are used up. At this point, the health trend model is considered to have been trained and a pre-constructed health trend model is obtained.

[0075] It is understood that S201-S205 may perform modeling and construction only once during the health status identification process for the same type of engine, or may perform multiple constructions as needed, or may be reconstructed each time. The embodiments of the present invention do not impose any restrictions on this.

[0076] S206. Obtain engine operating data within the time period to be detected, and divide the engine operating data into at least two batches of data according to the preset batch division rules.

[0077] The preset batch division rules include:

[0078] The engine operating data is divided into equal-length batches based on a preset time window.

[0079] The engine operating data is divided into batches of equal length with the same data volume according to the preset data length.

[0080] In some examples, since the acquired engine operation data is data within a period of time to be detected, a preset time window can be constructed with a given time length according to actual needs. Starting from the beginning of the period to be detected, the engine operation data contained therein is divided into batches, so that the preset time windows are seamlessly connected within the period to be detected. The engine operation data in each preset time window is considered as the same batch of data, that is, batch data of equal length in the time dimension.

[0081] It is understandable that the amount of data contained in each batch of equal-length batches over the aforementioned time dimension may differ.

[0082] In some examples, to ensure that the amount of data in each batch of data input into the health trend model is consistent, a preset data length can be set according to actual needs. Based on the preset data length, the engine operation data obtained within the time period to be detected is divided into multiple batches of data with the same amount of data, that is, the same length in terms of data volume.

[0083] It is understood that when dividing engine operation data within the testing period into batches, appropriate methods can be selected from the preset batch division rules according to different actual needs. In addition to the two examples mentioned above, the batch division rules can also be adaptively adjusted according to actual needs, and the embodiments of the present invention do not limit this.

[0084] S207. Input the data from each batch into the pre-built health trend model to determine the prediction results for each batch of data.

[0085] Among them, the health trend model is a feature model with a nonlinear degradation trend, constructed based on key engine operating variables obtained under stable operating conditions.

[0086] Specifically, each batch of data is input into a pre-built health trend model, which then uses the data in each batch that corresponds to the independent variable in the health trend model to perform the prediction processing of the corresponding response variable and obtain the prediction results for each batch of data.

[0087] S208. For each batch of data, the data in the batch that has the same data type as the response variable in the health trend model is identified as the actual data to be compared.

[0088] Specifically, since the prediction results output by the health trend model can be understood as the values ​​that the response variables should correspond to under healthy operating conditions, and the values ​​corresponding to these response variables have also been acquired and exist in each batch of data during actual operation, in order to compare the two to determine whether there are any situations in each batch of data that do not belong to a healthy operating state during actual operation, for each batch of data, the data that has the same data type as the response variables in the health trend model can be identified as the actual data to be compared with the prediction results.

[0089] S209. Determine the fitting residuals based on the prediction results corresponding to the actual data to be compared and the batch data.

[0090] Specifically, the sum of squared residuals between the actual data to be compared and the predicted results corresponding to the batch data is calculated, and the calculated result is determined as the fitting residual.

[0091] S210. If the fitting residuals exceed the reference tolerance range of the health trend model, the batch data will be identified as abnormal batch data.

[0092] In this embodiment, the reference tolerance range can be specifically understood as the range of fitting errors that can be allowed between the actual data during engine operation and the health trend model, which is preset according to the actual situation.

[0093] Specifically, if the fitting residuals of a batch of data exceed the reference tolerance range pre-set for the health trend model based on actual conditions, it can be assumed that the engine may not have been operating in a healthy state when generating this batch of data. Alternatively, it can be assumed that there may have been some acquisition errors during the engine's operation and data generation process. In other words, there may be a significant difference between this batch of data and the engine operating data generated under healthy operating conditions, potentially indicating an anomaly. In this case, the batch of data can be identified as abnormal. Otherwise, it can be assumed that the batch of data is not abnormal, and the next batch of data can be evaluated.

[0094] S211. The fitting residuals of abnormal batch data are determined as the deviation magnitude.

[0095] S212. Two adjacent abnormal batches of data in the time dimension are identified as an abnormal batch data group, and the deviation rate of the abnormal batch data group is determined according to the deviation magnitude corresponding to the abnormal batch data group.

[0096] In this embodiment, the deviation rate can be specifically understood as the deviation of the engine operating conditions corresponding to two abnormal batches of data in the abnormal batch data group from the engine's healthy operating condition. Since the more severe the engine's abnormal operating condition, the greater its deviation from the engine's healthy state, the deviation magnitude corresponding to the abnormal batch data later in the time dimension should be greater than the deviation magnitude corresponding to the abnormal batch data earlier in the time dimension. Therefore, its deviation rate can be determined to be positive, which can be used to reflect whether the abnormal conditions of the engine during operation continue to worsen.

[0097] Specifically, to avoid misjudging the engine's health status due to data acquisition errors, and because the continuous occurrence of abnormalities better reflects the stability of abnormal conditions in the engine, two adjacent abnormal batches of data in the time dimension can be identified as an abnormal batch data group based on the acquisition time of each abnormal batch of data. At the same time, since each abnormal batch of data has its own deviation range, the rate of change between the two deviation ranges in an abnormal batch data group can be calculated to obtain the deviation rate of the abnormal batch data group.

[0098] S213. Determine whether each offset rate continues to increase in the time dimension. If yes, execute S214; otherwise, execute S215.

[0099] Specifically, the offset rates of each adjacent abnormal batch of data are compared sequentially to see if they show a continuous upward trend over time. If they do, it can be assumed that the speed at which the engine's operating state changes from healthy to abnormal is gradually increasing. Although there is no fault yet, a degradation trend has already emerged. In this case, S214 can be executed, and the engine's health status is considered to be in a sub-healthy state. If the offset rates do not show a continuous upward trend, it can be assumed that the speed at which the engine's operating state changes from healthy to abnormal is not gradually increasing. The occurrence of abnormal batch data may be related to acquisition errors rather than due to a degradation trend in the engine. In this case, S215 can be executed, and the engine's health status is considered to be in a healthy state.

[0100] S214. Determine the sub-healthy state as the engine's healthy state.

[0101] S215. Determine the health status as the engine health status.

[0102] Optionally, after determining the engine health status based on the deviation rate of each deviation magnitude over time, the method further includes:

[0103] Obtain the health status audit results corresponding to some or all of the abnormal batch data;

[0104] Based on the audit results of each health status and the corresponding engine health status, the engine recall rate and model recognition accuracy are determined.

[0105] Among them, the engine recall rate is the proportion of abnormal batches whose engine health status is in a sub-healthy state among the abnormal batches whose health status audit results are in a sub-healthy state.

[0106] Among the abnormal batches of data where the model identification accuracy is equivalent to the engine health status being sub-healthy, the proportion of abnormal batches of data with the health status audit result being sub-healthy is also included.

[0107] In this embodiment, the health status audit result can be specifically understood as the judgment result of whether the engine is in a sub-healthy state after the abnormal batch data is manually annotated or reviewed by experts.

[0108] In this embodiment, the engine recall rate can be specifically understood as the proportion of abnormal batches of data that are actually in a sub-healthy state and are successfully identified by the health trend model.

[0109] In this embodiment, the model recognition accuracy can be specifically understood as the proportion of abnormal batches of data that are identified as being in a sub-healthy state based on the health trend model, but are actually in a sub-healthy state rather than a healthy state. It can be used to reflect the accuracy of the health trend model's prediction of the input data.

[0110] Specifically, to improve the identification capability and stability of engine sub-health status using the model-based health trend model and offset judgment in practical engineering applications, a model evaluation mechanism can be introduced to evaluate the health trend model and quantitatively monitor its classification performance in batch-level trend change identification tasks. During the operation of the health trend model, each batch of data can be used as an evaluation unit. After identifying abnormal batches of data based on the health trend model, some or all of the abnormal batches are manually labeled or reviewed by experts to obtain the health status review results corresponding to some or all of the abnormal batches. Then, a binary classification confusion matrix can be constructed to measure the identification performance of the health trend model. Simultaneously, two key performance indicators, engine recall rate and model recognition accuracy, are defined to measure the model's capabilities. For abnormal batches of data whose health status review results indicate a sub-healthy state, the proportion of these abnormal batches whose engine health status is determined to be sub-healthy by the health trend model can be determined, and this proportion is used as the engine recall rate. Similarly, for abnormal batches of data whose engine health status is identified as sub-healthy, the proportion of these abnormal batches whose health status review results indicate a sub-healthy state can be determined, and this proportion is used as the model recognition accuracy. Based on the above two evaluation results, in the subsequent optimization process of the health trend model, priority will be given to improving the engine recall rate to reduce the risk of missing key deviation states. At the same time, the false alarm rate will be controlled through a regularization mechanism to improve the model's recognition accuracy, thereby reducing misjudgment interference while ensuring the safety of locomotive operation.

[0111] For example, if the health trend model determines that a batch of data belongs to an abnormal batch, and the health status review results confirm that the batch of data is indeed generated by an engine in a sub-healthy state during operation, then the health trend model's identification result can be considered correct, and this can be used as a true example to be introduced into subsequent supplementary training for the health trend model. Conversely, if the health trend model determines that a batch of data belongs to an abnormal batch, and the health status review results confirm that the batch of data is not generated by an engine in a sub-healthy state during operation, then the health trend model's identification result can be considered incorrect, and this can be used as a false positive example to be introduced into subsequent supplementary training for the health trend model. During training; if the health trend model determines that the batch data does not belong to the abnormal batch data, and the health status review results determine that the batch data is generated by the engine in a sub-healthy state during operation, then the health trend model's identification result can be considered incorrect, and it can be introduced as a false negative example into subsequent supplementary training for the health trend model; if the health trend model determines that the batch data does not belong to the abnormal batch data, and the health status review results determine that the batch data is not generated by the engine in a sub-healthy state during operation, then the health trend model's identification result can be considered correct, and it can be introduced as a true negative example into subsequent supplementary training for the health trend model.

[0112] In some examples, to more clearly describe the engine health status identification method provided in the embodiments of the present invention, a certain type of freight rail diesel locomotive is used as an example for illustration.

[0113] Before actual deployment, historical data samples with stable operation and no fault records during the historical operation period will be selected, classified and archived according to vehicle number, and a health reference model will be constructed based on the archived data. In this example, the relationship between engine speed and oil inlet pressure will be modeled to construct a health trend model. During the modeling process, historical data of diesel engines under stable operating conditions and without fault records will be selected as samples. Engine speed will be used as the input parameter, and oil inlet pressure will be used as the output target. A cubic polynomial regression modeling structure will be adopted. By minimizing the error between the predicted and actual values, the model parameters will be fitted. This modeling method can accurately reflect the nonlinear trend of oil pressure changing with engine speed. The structure is clear and suitable for deployment in resource-limited environments such as vehicle-mounted edge devices. After the model training is completed, it can be exported as a standard format parameter configuration file and integrated into the monitoring system to support rapid inference and analysis of newly collected batches of data.

[0114] In actual use, key parameters of the diesel locomotive during operation will be collected first, including but not limited to diesel engine speed, oil inlet pressure, and high-temperature coolant temperature, corresponding to "acquiring engine operating data within the time period to be detected" in S101. The sampling period for key parameters during operation can be set according to the equipment configuration, typically between several seconds and tens of seconds. The collected data is organized in chronological order and divided into batches according to a fixed quantity as the unit of analysis, corresponding to "dividing the engine operating data into at least two batches of data according to the preset batch division rules" in S101.

[0115] Then, batch-wise, the engine speed data of the current batch is input into the health trend model to obtain the corresponding oil pressure prediction result, which corresponds to "inputting the data of each batch into the pre-built health trend model to determine the prediction result of each batch of data" in S102. The oil pressure prediction result can then be compared with the actual pressure observation value of the current batch. Through residual calculation, it is assessed whether the operating state of the batch deviates significantly from the health model. If the residual value of a batch exceeds the set trend deviation threshold, the system marks it as an "abnormal batch" and automatically records relevant variables, batch number, abnormality degree, and other core information, which corresponds to "determining abnormal batch data based on the data of each batch and the corresponding prediction results of each batch of data" in S103. Considering the volatility of the engineering site, the system will not directly trigger an alarm for a single abnormal batch. Instead, it adopts a trend stability mechanism: performing trend analysis on the residual signals of multiple consecutive batches to identify whether they show a continuous expansion or non-linear acceleration trend. When the system detects multiple batches of continuous deviation and the rate of trend change is greater than the set threshold, it indicates that the vehicle state may be entering a slow degradation process, and this state is defined as a "sub-healthy state". Vehicles in this state typically do not yet exhibit obvious functional malfunctions, but without intervention, they may further evolve into actual damage. The system will immediately display "sub-health" on the local terminal and recommend paying close attention to the diesel engine system's operation, which corresponds to "determining the deviation range of each abnormal batch of data and determining the engine health status based on the deviation rate of each deviation range in the time dimension" in S104.

[0116] Understandably, after a vehicle is determined to have a stability deviation and have entered a sub-healthy state using a health trend model, a preliminary analysis of the root causes of this deviation can be conducted based on the health trend model. By combining the deviation of the current batch of data relative to the health data output by the health trend model, the slope of the trend curve, and the median fluctuation pattern, and referring to a pre-built knowledge base, a preliminary inference can be made about the faults causing the vehicle's sub-healthy state. The search method is closely related to the input parameters and output targets used in the construction of the health trend model.

[0117] For example, a continuous drop in oil pressure accompanied by fluctuations in engine speed may be related to a decline in the performance of the lubrication system; a rise in the temperature of the high-temperature coolant and a high ambient temperature may indicate a reduction in the efficiency of the cooling system; frequent high-load operation accompanied by an overall drop in oil pressure may be related to the degradation of the mechanical system. For the faults identified above, maintenance recommendations can be generated, including arranging manual inspections, adjusting operating parameters, and planned replacement of relevant components, to ensure that vehicle faults are detected promptly and addressed before they become serious, thereby improving vehicle operational stability.

[0118] The technical solution of this invention involves acquiring stable engine operation data under stable operating conditions before engine health status detection, and selecting independent and response variables from the data to construct a health trend model based on actual needs. Additionally, different content from the independent and response variables can be selected as a condition penalty term. This allows the health trend model constructed based on this data to suppress the interference of samples on the fitting results under specific environmental conditions, resulting in a health trend model that reflects the nonlinear degradation trend of key engine operating variables corresponding to the independent and response variables during engine operation. Furthermore, when performing engine health status detection, a pre-built health trend model can be used to predict the input batch data in a manner consistent with the data type of the response variable, obtaining the corresponding prediction results. These prediction results are then compared with the data in the batch that are consistent with the data type of the response variable to obtain the corresponding fitting residuals. These fitting residuals can not only be used to identify abnormal batch data, but also to determine the deviation rate between two adjacent abnormal batch data in the time dimension. Only when the deviation rate continues to rise in the time dimension is the engine health status considered to be sub-healthy. This avoids the impact of data fluctuations caused by environmental factors on the accuracy of identification when determining the engine health status in a single identification, improving the accuracy and stability of sub-health status identification, preventing the engine from directly entering a serious fault without warning, and improving engine maintenance efficiency and vehicle driving safety.

[0119] Example 3

[0120] Figure 3 This is a schematic diagram of the structure of an engine health status identification device provided in Embodiment 3 of the present invention, as shown below. Figure 3 As shown, the engine health status identification device includes a data acquisition module 31, a prediction result determination module 32, an abnormal data determination module 33, and a health status determination module 34.

[0121] The data acquisition module 31 is used to acquire engine operating data within the time period to be detected, and divide the engine operating data into at least two batches of data according to a preset batch division rule; the prediction result determination module 32 is used to input each batch of data into a pre-built health trend model to determine the prediction result of each batch of data; wherein, the health trend model is a feature model with a non-linear degradation trend constructed based on key engine operating variables acquired under stable operating conditions; the abnormal data determination module 33 is used to determine abnormal batch data based on each batch of data and the prediction result corresponding to each batch of data; the health status determination module 34 is used to determine the deviation magnitude of each abnormal batch of data, and determine the engine health status based on the deviation rate of each deviation magnitude in the time dimension.

[0122] The technical solution of this invention divides the acquired engine operating data within the detection time period into batches, and inputs each batch of data into a pre-constructed health trend model that satisfies a non-linear degradation trend. This determines whether any batch of data exhibits operational anomalies. After identifying abnormal batches of data, the system does not directly determine if the engine's health is abnormal. Instead, it compares the deviations of different abnormal batches of data over time to determine if the engine is in a sub-healthy state before reaching a fault. This process utilizes a health trend model that reflects the engine's non-linear degradation trend and is constructed based on key engine operating characteristics. The judgment involves multiple batches of data and does not rely on human experience, reducing the impact of data fluctuations caused by environmental factors on the accuracy of the identification results. It also reduces the impact of expert subjective judgment on the consistency and stability of the identification results, improving the accuracy and stability of sub-healthy state identification. This prevents the engine from directly entering a serious fault without warning, improving engine maintenance efficiency and vehicle operating safety.

[0123] Optional, the abnormal data determination module 33 is specifically used for:

[0124] For each batch of data, the data in the batch that has the same data type as the response variable in the health trend model is identified as the actual data to be compared;

[0125] Determine the fitting residuals based on the prediction results corresponding to the actual data to be compared and the batch data;

[0126] If the fitting residuals exceed the reference tolerance range of the health trend model, the batch data will be identified as abnormal batch data.

[0127] Optional, the health status determination module 34 is specifically used for:

[0128] The fitting residuals of abnormal batch data are determined as the deviation magnitude.

[0129] Two adjacent abnormal batches of data in the time dimension are identified as an abnormal batch data group, and the deviation rate of the abnormal batch data group is determined according to the deviation magnitude of the abnormal batch data group.

[0130] If the deviation rates continue to rise over time, the sub-healthy state is determined to be the engine healthy state; otherwise, the healthy state is determined to be the engine healthy state.

[0131] Optionally, preset batch division rules include: dividing engine operation data into equal-length batches based on a preset time window; or dividing engine operation data into equal-length batches with consistent data volume based on a preset data length.

[0132] Optionally, the engine health status recognition device also includes:

[0133] The trend model construction module is used to acquire stable engine operating data under stable operating conditions before acquiring engine operating data for the period to be detected, and to determine multiple key engine operating variables based on the data types contained in the stable engine operating data; at least two of these key engine operating variables are selected as independent and response variables, respectively, and one of the remaining key engine operating variables is selected as a condition penalty term; an initial health trend model is constructed based on the independent and response variables, and an objective function is constructed using the condition penalty term as a stability constraint; data corresponding to the independent, response, and condition penalty terms are extracted from the stable engine operating data to construct a training sample set; the initial health trend model is trained using the training sample set and the objective function to obtain a pre-constructed health trend model; the key engine operating variables include at least: engine speed, pressure, temperature, and load.

[0134] Optionally, the engine health status recognition device also includes:

[0135] The evaluation module is used to determine the engine health status based on the deviation rate of each deviation magnitude over time, and then obtain the health status review results corresponding to some or all of the abnormal batch data. Based on each health status review result and the corresponding engine health status, the module determines the engine recall rate and model recognition accuracy. The engine recall rate is the proportion of abnormal batch data with a sub-healthy engine health status among those with a health status review result. The model recognition accuracy is the proportion of abnormal batch data with a sub-healthy engine health status among those with a health status review result.

[0136] The engine health status identification device provided in this embodiment of the invention can execute the engine health status identification method provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method.

[0137] Example 4

[0138] Figure 4 This is a schematic diagram of the structure of an engine health status identification device according to Embodiment 4 of the present invention. The engine health status identification device 40 can be an electronic device, or a wearable device (such as a helmet, glasses, watch, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the invention described and / or claimed herein.

[0139] like Figure 4 As shown, the engine health status identification device 40 includes at least one processor 41 and a memory, such as a read-only memory (ROM) 42 and a random access memory (RAM) 43, communicatively connected to the at least one processor 41. The memory stores computer programs executable by at least two processors. The processor 41 can perform various appropriate actions and processes based on the computer program stored in the ROM 42 or loaded from storage unit 48 into the RAM 43. The RAM 43 can also store various programs and data required for the operation of the engine health status identification device 40. The processor 41, ROM 42, and RAM 43 are interconnected via a bus 44. An input / output (I / O) interface 45 is also connected to the bus 44.

[0140] Multiple components in the engine health status identification device 40 are connected to the I / O interface 45, including: an input unit 46, such as a button, touch screen, etc.; an output unit 47, such as various types of displays, speakers, etc.; a storage unit 48, such as a disk, optical disk, etc.; and a communication unit 49, such as a network card, modem, wireless transceiver, etc. The communication unit 49 allows the engine health status identification device 40 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0141] Processor 41 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 41 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 41 performs the various methods and processes described above, such as the engine health status identification method.

[0142] In some embodiments, the engine health status identification method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 48. In some embodiments, part or all of the computer program may be loaded and / or installed on the engine health status identification device 40 via ROM 42 and / or communication unit 49. When the computer program is loaded into RAM 43 and executed by processor 41, one or more steps of the engine health status identification method described above may be performed. Alternatively, in other embodiments, processor 41 may be configured to perform the engine health status identification method by any other suitable means (e.g., by means of firmware).

[0143] Optionally, embodiments of the present invention also provide a computer program product, including a computer program that, when executed by a processor, implements the engine health status identification method as provided in any embodiment of the present invention.

[0144] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0145] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0146] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0147] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0148] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or computing systems that include middleware components (e.g., application servers), or computing systems that include frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0149] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0150] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and no limitation is imposed herein.

[0151] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for identifying the health status of an engine, characterized in that, include: Acquire engine operating data within the time period to be tested, and divide the engine operating data into at least two batches of data according to a preset batch division rule; Each batch of data is input into a pre-constructed health trend model to determine the prediction result of each batch of data; wherein, the health trend model is a feature model with a non-linear degradation trend constructed based on key engine operating variables obtained under stable operating conditions; Abnormal batch data are determined based on each batch of data and the prediction results corresponding to each batch of data; Determine the deviation magnitude of each of the abnormal batches of data, and determine the engine health status based on the deviation rate of each deviation magnitude over time.

2. The engine health status identification method according to claim 1, characterized in that, Before acquiring the engine operating data for the time period to be detected, the method further includes: Acquire stable engine operating data under stable operating conditions, and determine multiple key engine operating variables based on the data types contained in the stable engine operating data; At least two of the key engine operating variables are selected as independent and response variables, respectively, and one of the remaining key engine operating variables is selected as the operating condition penalty term. An initial health trend model is constructed based on the independent variables and the response variables, and an objective function is constructed using the working condition penalty term as a stability constraint factor. Data corresponding to the independent variable, the response variable, and the operating condition penalty term are extracted from the stable engine operating data to construct a training sample set; The initial health trend model is trained using the training sample set and the objective function to obtain a pre-constructed health trend model; The key operating variables of the engine include at least one of the following: engine speed, pressure, temperature, and load.

3. The engine health status identification method according to claim 1, characterized in that, The step of determining abnormal batch data based on each batch of data and the prediction results corresponding to each batch of data includes: For each batch of data, the data in the batch that has the same data type as the response variable in the health trend model is determined as the actual data to be compared; The fitting residual is determined based on the prediction results corresponding to the actual data to be compared and the batch data. If the fitting residual exceeds the reference tolerance range of the health trend model, the batch data is identified as abnormal batch data.

4. The engine health status identification method according to claim 3, characterized in that, Determine the deviation magnitude of each of the abnormal batches of data, and determine the engine health status based on the deviation rate of each deviation magnitude over time, including: The fitting residual of the abnormal batch data is determined as the deviation magnitude. Two adjacent abnormal batches of data in the time dimension are identified as an abnormal batch data group, and the deviation rate of the abnormal batch data group is determined according to the deviation magnitude corresponding to the abnormal batch data group. If the deviation rates described above continue to increase over time, the sub-healthy state is determined to be the engine healthy state; otherwise, the healthy state is determined to be the engine healthy state.

5. The engine health status identification method according to claim 1, characterized in that, The preset batch division rules include: The engine operating data is divided into equal-length batches according to a preset time window; or The engine operating data is divided into batches of equal length with consistent data volume according to a preset data length.

6. The engine health status identification method according to any one of claims 1-5, characterized in that, After determining the engine health status based on the deviation rate of each deviation magnitude over time, the method further includes: Obtain the health status audit results corresponding to some or all of the abnormal batch data; Based on the health status review results and the engine health status corresponding to each health status review result, determine the engine recall rate and model recognition accuracy; The engine recall rate is the proportion of abnormal batches of data whose engine health status is in a sub-healthy state among the abnormal batches of data whose health status audit result is in a sub-healthy state. The model recognition accuracy refers to the proportion of abnormal batches of data whose engine health status is sub-healthy, and whose health status audit result is sub-healthy.

7. An engine health status identification device, characterized in that, include: The data acquisition module is used to acquire engine operating data within the time period to be detected, and to divide the engine operating data into at least two batches of data according to a preset batch division rule. The prediction result determination module is used to input the batch data into the pre-constructed health trend model and determine the prediction result of each batch data; wherein, the health trend model is a feature model with a nonlinear degradation trend constructed based on the key operating variables of the engine obtained under stable operating conditions. An abnormal data determination module is used to determine abnormal batch data based on each batch of data and the prediction results corresponding to each batch of data. The health status determination module is used to determine the deviation range of each batch of abnormal data, and to determine the engine health status based on the deviation rate of each deviation range in the time dimension.

8. An engine health status identification device, characterized in that, include: At least one processor; as well as A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the engine health status identification method of any one of claims 1-6.

9. A storage medium containing computer-executable instructions, characterized in that, The computer-executable instructions, when executed by a computer processor, are used to perform the engine health status identification method as described in any one of claims 1-6.

10. A computer program product, characterized in that, It includes a computer program that, when executed by a processor, implements the engine health status identification method as described in any one of claims 1-6.

Images