Equipment fault early warning method and system

Abstract

The invention provides an equipment fault early warning method and system, and belongs to the technical field of fault prediction, and the method comprises the steps: obtaining the real-time working condition parameters of equipment; the real-time working condition parameters comprise real-time operation parameters and environment parameters; extracting a first feature vector based on the real-time working condition parameter; respectively calculating the similarity between the first feature vector and a plurality of second feature vectors in a model library; determining the corresponding normal state model with the maximum similarity as a target model; and inputting the real-time working condition parameters of the equipment into the target model to obtain the deviation between the real-time working condition parameters and the normal parameter range, and carrying out fault early warning on the equipment based on the deviation. According to the equipment fault early warning method and system provided by the invention, the accuracy of equipment fault early warning can be improved.

Description

Technical Field

[0001] This application belongs to the field of fault prediction technology, and more specifically, relates to a method and system for early warning of equipment faults. Background Technology

[0002] Engineering equipment (including excavators, bulldozers, demolition robots, etc.) is essential in various engineering construction and emergency rescue fields. Its technical performance and operating status directly affect project quality, work efficiency, and personnel safety. Accurate early warning of equipment malfunctions can promptly detect potential faults or abnormal conditions, preventing casualties or downtime losses caused by sudden failures.

[0003] Traditional early warning methods for engineering equipment are mostly based on fixed thresholds for fault judgment. When the operating conditions of the equipment change, it can lead to false alarms or missed alarms, making it impossible to provide accurate early warnings. Summary of the Invention

[0004] The purpose of this application is to provide a method and system for early warning of equipment failure, so as to improve the accuracy of early warning of equipment failure.

[0005] A first aspect of this application provides an equipment fault early warning method, comprising: Acquire real-time operating parameters of the equipment; the real-time operating parameters include real-time operating parameters and environmental parameters; Extract the first feature vector based on the real-time operating parameters; The similarity between the first feature vector and multiple second feature vectors in the model library is calculated respectively; wherein, each second feature vector in the model library is used to characterize a working condition classification of the equipment, and each working condition classification has a corresponding normal state model, which is used to characterize the normal parameter range of the equipment under the corresponding working condition classification. The normal state model with the highest similarity is identified as the target model; The real-time operating parameters of the equipment are input into the target model to obtain the deviation between the real-time operating parameters and the normal parameter range, so as to provide fault warning for the equipment based on the deviation.

[0006] A second aspect of this application provides an equipment fault early warning system, comprising: The data acquisition module is used to acquire the real-time operating parameters of the equipment; the real-time operating parameters include real-time operating parameters and environmental parameters. The feature extraction module is used to extract a first feature vector based on the real-time operating parameters; The feature matching module is used to calculate the similarity between the first feature vector and multiple second feature vectors in the model library respectively; wherein, each second feature vector in the model library is used to characterize a working condition classification of the equipment, and each working condition classification has a corresponding normal state model, which is used to characterize the normal parameter range of the equipment under the corresponding working condition classification. The model selection module is used to identify the normal state model with the highest similarity as the target model; The fault early warning module is used to input the real-time operating parameters of the equipment into the target model, obtain the deviation between the real-time operating parameters and the normal parameter range, and make fault early warning of the equipment based on the deviation.

[0007] A third aspect of this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement the steps of the equipment fault early warning method described above.

[0008] A fourth aspect of this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the above-described equipment fault early warning method.

[0009] The beneficial effects of the equipment fault early warning method and system provided in this application are as follows: This application pre-constructs a model library containing second feature vectors for multiple typical operating condition classifications, as well as normal state models for multiple typical operating condition classifications. Based on this, a first feature vector is extracted based on the real-time operating parameters of the equipment. The normal state model corresponding to the current operating condition classification of the equipment is determined by calculating the similarity between the first and second feature vectors, i.e., the target model. The target model can characterize the normal parameter range under the current operating condition classification. Therefore, fault warning based on the target model under the current operating condition classification can improve the accuracy of fault warning. Attached Figure Description

[0010] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1 A flowchart illustrating an equipment fault early warning method provided in an embodiment of this application; Figure 2This is a structural block diagram of an equipment fault early warning system provided in an embodiment of this application; Figure 3 This is a schematic block diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0012] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0013] To make the objectives, technical solutions, and advantages of this application clearer, the following description will be provided in conjunction with the accompanying drawings and specific embodiments.

[0014] Please refer to Figure 1 , Figure 1 This is a flowchart illustrating an equipment fault early warning method provided in an embodiment of this application. The equipment fault early warning method provided in this embodiment can be executed by an electronic device, and the method may include: S101: Obtain real-time operating parameters of the equipment; real-time operating parameters include real-time running parameters and environmental parameters.

[0015] In this embodiment, the real-time operating parameters of the equipment can be used to characterize the current operating condition of the equipment. These mainly include real-time operating parameters and environmental parameters. Real-time operating parameters can characterize the operating status of the equipment, including engine speed, hydraulic system pressure, gearbox oil temperature or motor current, etc. Environmental parameters can characterize the external conditions of the equipment operation, such as ambient temperature, humidity, altitude, dust concentration, electromagnetic noise or vibration intensity, etc.

[0016] Specifically, the aforementioned real-time operating parameters can be collected in real time through the sensor network (such as temperature sensors, pressure sensors, vibration sensors, GPS or environmental monitoring modules, etc.) on the equipment.

[0017] S102: Extract the first feature vector based on real-time operating parameters.

[0018] In this embodiment, the collected real-time operating parameters can be preprocessed. For example, the real-time operating parameters are first cleaned to remove isolated outliers caused by sensor false alarms. Then, the cleaned real-time operating parameters are standardized, such as mapping temperature and pressure to the [0,1] interval to unify the units. Finally, the standardized real-time operating parameters are time-series aligned to ensure that the timestamps of the multi-sensor data are synchronized.

[0019] Key features that characterize the current operating condition are extracted from the preprocessed real-time operating parameters, such as the mean, peak or rate of change of motor speed, as well as load level, ambient temperature range, altitude level, etc. The above key features are integrated into a fixed-dimensional vector to obtain the first feature vector.

[0020] S103: Calculate the similarity between the first feature vector and multiple second feature vectors in the model library respectively, and determine the normal state model with the highest similarity as the target model; each second feature vector in the model library is used to characterize a working condition classification of the equipment, and each working condition classification has a corresponding normal state model. The normal state model is used to characterize the normal parameter range of the equipment under the corresponding working condition classification.

[0021] In this embodiment, a model library can be pre-built, which includes the second feature vectors corresponding to multiple typical working condition categories (such as "high temperature and high load operation", "low temperature and empty transport", "high plateau and low speed driving", etc.) and normal state models corresponding to multiple working condition categories.

[0022] Among them, the normal state model corresponding to each working condition category is a benchmark model built based on the normal historical working condition parameters of the equipment under that working condition category. Through the analysis and learning of a large number of normal historical working condition parameters (such as operating parameters and environmental parameters), it extracts the normal parameter range under that working condition category.

[0023] Based on this, algorithms such as cosine similarity and Euclidean distance can be used to calculate the similarity between the first feature vector and all second feature vectors in the model library, quantify the matching degree between the current working condition (the working condition represented by the first feature vector) and historical typical working conditions (the working conditions corresponding to the second feature vector), and classify the working condition corresponding to the second feature vector with the highest similarity as the matching result of the current working condition.

[0024] S104: Identify the normal state model with the highest similarity as the target model.

[0025] In this embodiment, the normal state model with the highest similarity refers to the normal state model corresponding to the second feature vector with the highest similarity. The normal state model with the highest similarity is determined as the target model for subsequent fault warning under the current working condition.

[0026] S105: Input the real-time operating parameters of the equipment into the target model to obtain the deviation between the real-time operating parameters and the normal parameter range, so as to provide early warning of equipment failure based on the deviation.

[0027] In this embodiment, the target model is trained based on an autoencoder model. By inputting real-time operating parameters into the target model, the single-dimensional deviation between each real-time operating parameter and its normal parameter range can be obtained. If the single-dimensional deviation between a real-time operating parameter and its normal parameter range is greater than a preset second error threshold, it indicates that the real-time operating parameter is an abnormal parameter item; otherwise, the real-time operating parameter is a normal parameter item.

[0028] Specifically, when calculating the single-dimensional deviation between a real-time operating condition parameter and its normal parameter range, if the normal parameter range corresponding to the real-time operating condition parameter is a single limit, such as the upper limit of the current parameter being 10A, the single-dimensional deviation between the real-time operating condition parameter and its normal parameter range is calculated based on the real-time operating condition parameter and the single limit.

[0029] Taking current parameters as an example, the single-dimensional deviation between the current parameter and its normal parameter range can be calculated using the following formula: ; in, This represents a single-dimensional deviation between the current parameter and its normal parameter range. This represents the real-time value of the current parameter. This indicates the upper limit of the current parameter.

[0030] When calculating the single-dimensional deviation between a real-time operating parameter and its normal parameter range, if the normal parameter range corresponding to the real-time operating parameter includes an upper limit and a lower limit, such as the upper limit of the temperature parameter being 70℃ and the lower limit being -25℃, the single-dimensional deviation between the real-time operating parameter and its normal parameter range is calculated based on the real-time operating parameter, the upper limit of the real-time operating parameter, and the lower limit of the real-time operating parameter.

[0031] Taking temperature as an example, the single-dimensional deviation between the temperature parameter and its normal range can be calculated using the following formula: ; in, This represents a single-dimensional deviation between a temperature parameter and its normal range. This represents the real-time value of the temperature parameter. This indicates the upper limit of the current parameter. This indicates the lower limit of the current parameter.

[0032] In this embodiment, after obtaining the single-dimensional deviation corresponding to each real-time operating condition parameter, the first reconstruction error (overall deviation) between the real-time operating condition parameter and the normal parameter range can be obtained by calculating the sum of squares or average value of each single-dimensional deviation.

[0033] Based on this, the real-time operating parameters of the equipment are input into the target model to obtain the deviation between the real-time operating parameters and the normal parameter range, so as to provide early warning of equipment failure based on the deviation, including: Input the real-time operating parameters of the equipment into the target model to obtain the first reconstruction error of the target model; If the first reconstruction error is less than or equal to the first error threshold under the corresponding working condition category, the operating status of the output equipment is normal. If the first reconstruction error is greater than the first error threshold under the corresponding working condition category, the operating status of the output equipment is a fault state.

[0034] In this embodiment, if the first reconstruction error between the real-time operating parameters and the normal parameter range is greater than the preset first error threshold, the equipment is determined to be operating normally and no fault warning is triggered. If the deviation is greater than the preset first error threshold, the equipment is determined to be abnormal and a fault warning is triggered.

[0035] Meanwhile, when an abnormality is detected in the equipment, the target model can also output specific abnormal parameters, such as "abnormal temperature rise" or "vibration frequency deviating from the normal range". Based on this, staff can focus on the systems, components or operating links (such as cooling systems, transmission components, etc.) related to the abnormal parameters, thereby greatly narrowing the scope of investigation and improving diagnostic efficiency and accuracy.

[0036] As can be seen from the above, this embodiment pre-constructs a model library containing second feature vectors under multiple typical operating condition classifications, as well as normal state models under multiple typical operating condition classifications. Based on this, a first feature vector is extracted based on the real-time operating condition parameters of the equipment. By calculating the similarity between the first and second feature vectors, the normal state model corresponding to the current operating condition classification of the equipment is determined, i.e., the target model. The target model can characterize the normal parameter range under the current operating condition classification. Therefore, fault warning based on the target model under the current operating condition classification can improve the accuracy of fault warning.

[0037] In one embodiment of this application, the method for determining the operating condition classification includes: Acquire multiple sets of normal historical operating condition parameters for the equipment; each set of normal historical operating condition parameters includes historical operating parameters and historical environmental parameters during normal equipment operation. Multiple sets of normal historical operating condition parameters were subjected to feature extraction to obtain multiple third feature vectors; Cluster analysis is performed on multiple third feature vectors to obtain multiple cluster regions. Each cluster region corresponds to a working condition classification, and the center of each cluster region corresponds to a second feature vector.

[0038] In this embodiment, the multiple operating condition classifications of the equipment can be obtained by cluster analysis of multiple sets of normal historical operating condition parameters of the equipment.

[0039] Specifically, the multiple sets of normal historical operating condition parameters of the equipment refer to the historical operating condition parameters under normal operating conditions. These parameters can be collected in advance, covering different operating scenarios (e.g., no-load / full-load, low-speed / high-speed) and environmental conditions (e.g., normal temperature / high temperature, plains / plateaus). Each set of data includes historical operating parameters (e.g., engine speed, hydraulic pressure, oil temperature) and historical environmental parameters (e.g., ambient temperature, humidity, altitude). Then, preprocessing operations such as cleaning, normalization, and time-series alignment are performed on the multiple sets of normal historical operating condition parameters. Key features that characterize the operating condition classification are then extracted from these preprocessed parameters, and these key features are integrated into a fixed-dimensional vector to obtain the third feature vector.

[0040] Based on this, clustering algorithms such as K-means or DBSCAN are used to perform clustering analysis on multiple third feature vectors. Each cluster region formed after clustering represents a type of working condition with similar characteristics and can be classified into a working condition category. The center of each cluster region is the second feature vector corresponding to that working condition category.

[0041] As can be seen from the above, this embodiment obtains a working condition classification by performing cluster analysis on multiple sets of normal historical working condition parameters. This classification can cover all typical working condition classifications of the equipment, ensuring that the typical working condition classification can be matched with the real-time working condition parameters, thereby accurately determining the current working condition classification.

[0042] In one embodiment of this application, the method for determining the corresponding normal state model for each operating condition category includes: Obtain multiple sets of normal historical operating condition parameters under this operating condition category; each set of normal historical operating condition parameters includes historical operating parameters and historical environmental parameters during normal equipment operation; An autoencoder model is trained based on multiple sets of normal historical operating condition parameters under this operating condition classification to obtain the normal state model corresponding to this operating condition classification.

[0043] In this embodiment, for each operating condition category, multiple sets of historical data on the normal operation of the equipment under that category can be collected in advance. These data include complete historical operating parameters (such as rotational speed, pressure, and temperature) and historical environmental parameters (such as ambient temperature and humidity, and altitude). The collected data is cleaned (outliers are removed), standardized (unitized units), and time-series aligned to form a training dataset. Subsequently, an autoencoder model is trained using this dataset as input. The autoencoder model includes an encoder and a decoder. The encoder compresses the input data into hidden layer features, and the decoder reconstructs the input data. Through multiple iterations of optimization, the reconstruction error is minimized, allowing the autoencoder model to fully learn the normal distribution and fluctuation range of parameters under that operating condition category. After training, the converged autoencoder model can be used as the normal state model corresponding to that operating condition category.

[0044] As can be seen from the above, this embodiment trains the corresponding normal state model based on multiple sets of normal historical operating condition parameters under each operating condition category, which can accurately capture the unique distribution pattern and fluctuation range of the operating condition category and improve the accuracy of fault warning under the operating condition category.

[0045] In one embodiment of this application, the method for determining the first error threshold for each operating condition category includes: Input the parameters of multiple sets of normal historical operating conditions under this operating condition category into the trained autoencoder model to obtain multiple second reconstruction errors corresponding to the parameters of multiple sets of normal historical operating conditions. By fitting the probability distribution of multiple second reconstruction errors, the standard deviation and mean of the multiple second reconstruction errors are obtained; The first error threshold corresponding to the classification of this working condition is determined based on the standard deviation and mean of multiple second reconstruction errors.

[0046] In this embodiment, for each operating condition category, multiple sets of normal historical operating condition parameters under that category are input into the trained autoencoder model. This yields multiple second reconstruction errors corresponding to the multiple sets of normal historical operating condition parameters. These multiple second reconstruction errors can serve as sample data of reconstruction errors (normal errors) under normal operating conditions of the equipment. Statistical analysis is performed on these multiple second reconstruction errors to fit their probability distribution. Taking a normal distribution as an example, the mean (μ) and standard deviation (σ) of the normal distribution can be calculated. A first error threshold is then set based on the mean and standard deviation, for example, μ + 3σ (covering 99.7% of the normal error). The corresponding quantile value is determined as the first error threshold to ensure that the reconstruction error of the normal data rarely exceeds the limit.

[0047] As can be seen from the above, this embodiment determines the first error threshold based on the error distribution of the second reconstruction error corresponding to multiple sets of normal historical operating condition parameters under each operating condition category, so that the setting of the first error threshold is more in line with the actual error law of the operating condition category, and further improves the accuracy of fault warning under the operating condition category.

[0048] In one embodiment of this application, the equipment fault early warning method further includes: The cumulative runtime for acquiring equipment; The attenuation coefficient is determined based on the cumulative running time. The first error threshold is corrected based on the attenuation coefficient to obtain the corrected first error threshold. If the first reconstruction error is greater than the first error threshold under the corresponding operating condition category, the output equipment's operating status is a fault state, including: If the first reconstruction error is greater than the corrected first error threshold, the operating status of the output equipment is a fault state.

[0049] In this embodiment, considering that the equipment will experience normal performance degradation as the usage time increases, such as the fuel consumption and temperature of older equipment being slightly higher than that of newer equipment, this embodiment records the cumulative running time of the equipment in real time and corrects the first error threshold based on the cumulative running time of the equipment, so that the setting of the first error threshold is adapted to the changes throughout the entire life cycle of the equipment.

[0050] Specifically, the attenuation coefficient can be calculated using the following formula: ; in, Indicates the attenuation coefficient. Indicates the decay rate. For example, a preset constant. , This indicates the cumulative runtime.

[0051] Based on the obtained attenuation coefficient, multiplying the attenuation coefficient by the first error threshold yields the corrected first error threshold. Determining the equipment's operating status based on this corrected first error threshold improves the accuracy of the operating status assessment.

[0052] Similarly, the same method can be used to dynamically correct the second error threshold in order to accurately determine a single abnormal parameter item.

[0053] As can be seen from the above, this embodiment corrects the first error threshold based on the cumulative running time of the equipment, so that the setting of the first error threshold is adapted to the changes throughout the entire life cycle of the equipment.

[0054] In one embodiment of this application, the normal state model with the highest similarity is determined as the target model, including: If the maximum similarity is greater than or equal to the similarity threshold, the normal state model with the highest similarity is determined as the target model. If the maximum similarity is less than the similarity threshold, the preset first model is determined as the target model; the normal parameter range corresponding to the first model includes the normal parameter range under all working condition categories.

[0055] In this embodiment, after obtaining the similarity between the first feature vector and multiple second feature vectors in the model library, if the maximum similarity is greater than or equal to a preset similarity threshold (e.g., 0.8), it indicates that the current working condition is highly matched with a typical working condition classification, and the normal state model corresponding to the typical working condition classification is determined as the target model.

[0056] If the maximum similarity is less than the similarity threshold, it indicates that the current working condition belongs to an edge scenario or a new working condition that is not covered by the model library. In this case, the preset first model is called as the target model. The normal parameter range of this model integrates the normal range of all working condition classifications and can be compatible with various working condition classifications.

[0057] As can be seen from the above, this embodiment can ensure that the target model is highly adapted to the current working condition by filtering the target model through the similarity threshold for typical working conditions already covered in the model library (such as "normal temperature high load" and "low temperature no load"), thereby improving the early warning accuracy under the current working condition classification. For special working conditions not covered in the model library, the first model is used as a "fallback mechanism" to provide a general normal benchmark and ensure that the equipment can be effectively monitored in all scenarios.

[0058] In one embodiment of this application, the method for selecting the corresponding autoencoder model for each working condition category includes: If multiple sets of normal historical operating condition parameters under this operating condition category show that the load fluctuation value corresponding to this operating condition category is less than or equal to the fluctuation threshold, and the environmental parameters are within the preset environmental parameter range, then a fully connected autoencoder is selected as the autoencoder model, and the number of encoder layers is set to the first value. If multiple sets of normal historical operating condition parameters under this operating condition category show that the load fluctuation value corresponding to this operating condition category is greater than the fluctuation threshold and the environmental parameters are within the preset environmental parameter range, then an LSTM autoencoder is selected as the autoencoder model, and the number of encoder layers is set to the second value; wherein, the second value is greater than the first value. If multiple sets of normal historical operating condition parameters under this operating condition category show that the load fluctuation value corresponding to this operating condition category is less than or equal to the fluctuation threshold, and the environmental parameters exceed the preset environmental parameter range, the noise reduction connection autoencoder is selected as the autoencoder model, and the number of encoder layers is set to the second value. If multiple sets of normal historical operating condition parameters under this operating condition category show that the load fluctuation value corresponding to this operating condition category is greater than the fluctuation threshold and the environmental parameters exceed the preset environmental parameter range, then the CNN-LSTM autoencoder is selected as the autoencoder model, and the number of encoder layers is set to the third value; where the third value is greater than the second value.

[0059] In this embodiment, when training the normal state model corresponding to each working condition classification, the autoencoder model type and encoder layer number can be dynamically adjusted based on the load fluctuation characteristics and environmental parameter features of that working condition classification, so that the normal state model can be accurately adapted to the corresponding working condition classification.

[0060] Specifically, when multiple sets of normal historical operating condition parameters under this operating condition category show that the load fluctuation value corresponding to this operating condition category is less than or equal to the fluctuation threshold, and the environmental parameters are within the preset environmental parameter range, it indicates that the load fluctuation under this operating condition category is small and the environment is normal. A simple fully connected self encoder can be selected, and the normal mode can be captured with fewer layers (first value).

[0061] When multiple sets of normal historical operating condition parameters under a certain operating condition category show that the load fluctuation value corresponding to the operating condition category is greater than the fluctuation threshold and the environmental parameters are within the preset environmental parameter range, it indicates that the load fluctuation is large but the environment is normal under the operating condition category. It is necessary to pay attention to the temporal correlation of historical operating condition parameters. Therefore, an LSTM autoencoder (used to process time series data) is selected, and the number of layers (second value) is increased to capture complex dynamic features.

[0062] When multiple sets of normal historical operating condition parameters under this operating condition category show that the load fluctuation value corresponding to this operating condition category is less than or equal to the fluctuation threshold, and the environmental parameters exceed the preset environmental parameter range, it indicates that the load fluctuation under this operating condition category is small, but the environment is an extreme environment with large environmental interference. Therefore, a noise reduction autoencoder with noise resistance is selected, and the number of layers (second value) is increased to enhance the robustness of feature extraction.

[0063] When multiple sets of normal historical operating condition parameters under a given operating condition category show that the load fluctuation value corresponding to that operating condition category is greater than the fluctuation threshold, and the environmental parameters exceed the preset environmental parameter range, it indicates that the load fluctuation under that operating condition category is large, and the environment is an extreme environment. The operating condition simultaneously exhibits temporal dynamics and environmental noise. Therefore, a CNN-LSTM hybrid model is selected, and the maximum number of layers (the third value) is set to handle complex feature patterns. In the CNN-LSTM hybrid model, CNN stands for Convolutional Network, used to extract local features, and LSTM stands for Medium-Long-Term Memory Network, used to handle temporal correlations.

[0064] It should be noted that in autoencoder models, the number of encoder and decoder layers is the same to ensure reconstruction accuracy and simplify training. Therefore, the number of decoder layers can be determined based on the number of encoder layers.

[0065] As can be seen from the above, this embodiment dynamically adjusts the autoencoder model type and the number of encoder layers according to the complexity of the working conditions, so as to minimize the complexity of the autoencoder model while ensuring the accuracy of feature extraction.

[0066] Corresponding to the equipment fault early warning method in the above embodiments, Figure 2 This is a structural block diagram of an equipment fault early warning system provided according to an embodiment of this application. For ease of explanation, only the parts relevant to the embodiment of this application are shown. References Figure 2 The equipment fault early warning system 20 includes: a data acquisition module 21, a feature extraction module 22, a feature matching module 23, a model selection module 24, and a fault early warning module 25.

[0067] The data acquisition module 21 is used to acquire the real-time operating parameters of the equipment; the real-time operating parameters include real-time operating parameters and environmental parameters. Feature extraction module 22 is used to extract the first feature vector based on real-time operating parameters; The feature matching module 23 is used to calculate the similarity between the first feature vector and multiple second feature vectors in the model library, and to determine the normal state model with the highest similarity as the target model. Each second feature vector in the model library is used to characterize a working condition classification of the equipment. Each working condition classification has a corresponding normal state model. The normal state model is used to characterize the normal parameter range of the equipment under the corresponding working condition classification. The model selection module 24 is used to determine the normal state model with the highest similarity as the target model; The fault warning module 25 is used to input the real-time operating parameters of the equipment into the target model, obtain the deviation between the real-time operating parameters and the normal parameter range, and make fault warnings for the equipment based on the deviation.

[0068] In one embodiment of this application, the feature matching module 23 is further configured to: Acquire multiple sets of normal historical operating condition parameters for the equipment; each set of normal historical operating condition parameters includes historical operating parameters and historical environmental parameters during normal equipment operation. Multiple sets of normal historical operating condition parameters were subjected to feature extraction to obtain multiple third feature vectors; Cluster analysis is performed on multiple third feature vectors to obtain multiple cluster regions. Each cluster region corresponds to a working condition classification, and the center of each cluster region corresponds to a second feature vector.

[0069] In one embodiment of this application, for each working condition category, the feature matching module 23 is further configured to: Obtain multiple sets of normal historical operating condition parameters under this operating condition category; each set of normal historical operating condition parameters includes historical operating parameters and historical environmental parameters during normal equipment operation; An autoencoder model is trained based on multiple sets of normal historical operating condition parameters under this operating condition classification to obtain the normal state model corresponding to this operating condition classification.

[0070] In one embodiment of this application, the target model is a model trained based on an autoencoder model; the fault warning module 25 is further used for: The real-time operating parameters of the equipment are input into the target model to obtain the deviation between the real-time operating parameters and the normal parameter range. Based on this deviation, fault warnings for the equipment are provided, including: Input the real-time operating parameters of the equipment into the target model to obtain the first reconstruction error of the target model; If the first reconstruction error is less than or equal to the first error threshold under the corresponding working condition category, the operating status of the output equipment is normal. If the first reconstruction error is greater than the first error threshold under the corresponding working condition category, the operating status of the output equipment is a fault state.

[0071] In one embodiment of this application, for each operating condition category, the fault early warning module 25 is further configured to: Input the parameters of multiple sets of normal historical operating conditions under this operating condition category into the trained autoencoder model to obtain multiple second reconstruction errors corresponding to the parameters of multiple sets of normal historical operating conditions. By fitting the probability distribution of multiple second reconstruction errors, the standard deviation and mean of the multiple second reconstruction errors are obtained; The first error threshold corresponding to the classification of this working condition is determined based on the standard deviation and mean of multiple second reconstruction errors.

[0072] In one embodiment of this application, for each operating condition category, the fault early warning module 25 is further configured to: The cumulative runtime for acquiring equipment; The attenuation coefficient is determined based on the cumulative running time. The first error threshold is corrected based on the attenuation coefficient to obtain the corrected first error threshold. If the first reconstruction error is greater than the first error threshold under the corresponding operating condition category, the output equipment's operating status is a fault state, including: If the first reconstruction error is greater than the corrected first error threshold, the operating status of the output equipment is a fault state.

[0073] In one embodiment of this application, the model selection module 24 is further configured to: If the maximum similarity is greater than or equal to the similarity threshold, the normal state model with the highest similarity is determined as the target model. If the maximum similarity is less than the similarity threshold, the preset first model is determined as the target model; the normal parameter range corresponding to the first model includes the normal parameter range under all working condition categories.

[0074] See Figure 3 , Figure 3 This is a schematic block diagram of an electronic device provided according to an embodiment of this application. Figure 3 The electronic device 300 in this embodiment may include one or more processors 301, one or more input devices 302, one or more output devices 303, and one or more memories 304. The processors 301, input devices 302, output devices 303, and memories 304 communicate with each other via a communication bus 305. The memories 304 store computer programs, including program instructions. The processors 301 execute the program instructions stored in the memories 304. Specifically, the processors 301 are configured to invoke the program instructions to perform the functions of each module / unit in the above-described device embodiments, for example... Figure 2 The functions of the data acquisition module 21, feature extraction module 22, feature matching module 23, model selection module 24, and fault early warning module 25 are shown.

[0075] It should be understood that, in the embodiments of this application, the processor 301 may be a central processing unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0076] Input device 302 may include a touchpad, a fingerprint sensor (for collecting the user's fingerprint information and fingerprint orientation information), a microphone, etc., and output device 303 may include a display (LCD, etc.), a speaker, etc.

[0077] The memory 304 may include read-only memory and random access memory, and provides instructions and data to the processor 301. A portion of the memory 304 may also include non-volatile random access memory. For example, the memory 304 may also store preset constants such as model libraries and similarity thresholds.

[0078] In specific implementations, the processor 301, input device 302, and output device 303 described in the embodiments of this application can execute the implementation method described in the equipment fault early warning method provided in the embodiments of this application, or they can execute the implementation method of the electronic device described in the embodiments of this application, which will not be repeated here.

[0079] In another embodiment of this application, a computer-readable storage medium is provided. This computer-readable storage medium stores a computer program, which includes program instructions. When executed by a processor, the program instructions implement all or part of the processes in the methods described above. Alternatively, the computer program can instruct related hardware to complete the process. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include any entity or device capable of carrying computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0080] The computer-readable storage medium can be an internal storage unit of the electronic device in any of the foregoing embodiments, such as a hard disk or memory of the electronic device. The computer-readable storage medium can also be an external storage device of the electronic device, such as a plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, etc., equipped on the electronic device. Furthermore, the computer-readable storage medium can include both internal and external storage units of the electronic device. The computer-readable storage medium is used to store computer programs and other programs and data required by the electronic device. The computer-readable storage medium can also be used to temporarily store data that has been output or will be output.

[0081] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this application.

[0082] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the electronic devices and units described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0083] In the several embodiments provided in this application, it should be understood that the disclosed electronic devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, modules may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connections shown or discussed may be indirect coupling or communication connections through some interfaces or units, or they may be electrical, mechanical, or other forms of connection.

[0084] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of the embodiments of this application, depending on actual needs.

[0085] Furthermore, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The integrated modules described above can be implemented in hardware or as software functional modules.

[0086] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. An equipment failure early warning method characterized by, The method comprises: acquiring real-time working condition parameters of equipment; the real-time working condition parameters comprise real-time operation parameters and environmental parameters; extracting a first feature vector based on the real-time working condition parameters; calculating the similarity between the first feature vector and a plurality of second feature vectors in a model library respectively; each second feature vector in the model library is used to represent a working condition classification of the equipment, each working condition classification has a corresponding normal state model, and the normal state model is used to represent a normal parameter range of the equipment under the corresponding working condition classification; determining a normal state model with the largest corresponding similarity as a target model; inputting the real-time working condition parameters of the equipment into the target model to obtain a deviation between the real-time working condition parameters and the normal parameter range, and performing equipment fault early warning based on the deviation.

2. The method of claim 1, wherein, The determination method of the working condition classification comprises: acquiring a plurality of sets of normal historical working condition parameters of the equipment; each set of normal historical working condition parameters comprises historical operation parameters and historical environmental parameters when the equipment is normally operated; performing feature extraction on the plurality of sets of normal historical working condition parameters respectively to obtain a plurality of third feature vectors; performing clustering analysis on the plurality of third feature vectors to obtain a plurality of clustering regions, each clustering region corresponding to a working condition classification, and the center of each clustering region corresponding to a second feature vector.

3. The method of claim 1, wherein, For each working condition classification, the determination method of the corresponding normal state model comprises: acquiring a plurality of sets of normal historical working condition parameters under the working condition classification; each set of normal historical working condition parameters comprises historical operation parameters and historical environmental parameters when the equipment is normally operated; training an autoencoder model based on the plurality of sets of normal historical working condition parameters under the working condition classification to obtain the normal state model corresponding to the working condition classification.

4. The method of claim 1, wherein, The target model is a model trained based on an autoencoder model; inputting the real-time working condition parameters of the equipment into the target model to obtain a deviation between the real-time working condition parameters and the normal parameter range, and performing equipment fault early warning based on the deviation, comprising: inputting the real-time working condition parameters of the equipment into the target model to obtain a first reconstruction error of the target model; if the first reconstruction error is less than or equal to a first error threshold value under the corresponding working condition classification, outputting that the running state of the equipment is a normal state; if the first reconstruction error is greater than the first error threshold value under the corresponding working condition classification, outputting that the running state of the equipment is a fault state.

5. The method of claim 4, wherein, For each working condition classification, the determination method of the first error threshold value comprises: inputting the plurality of sets of normal historical working condition parameters under the working condition classification into the trained autoencoder model respectively to obtain a plurality of second reconstruction errors corresponding to the plurality of sets of normal historical working condition parameters respectively; fitting the probability distribution of the plurality of second reconstruction errors to obtain the standard deviation and mean value of the plurality of second reconstruction errors; determining the first error threshold value corresponding to the working condition classification based on the standard deviation and mean value of the plurality of second reconstruction errors.

6. The method of claim 5, wherein, The method further comprises: acquiring the cumulative running time of the equipment; determining a decay coefficient based on the cumulative running time; correcting the first error threshold value based on the decay coefficient to obtain a corrected first error threshold value; The first reconstruction error is greater than a first error threshold corresponding to the working condition classification, and the running state of the equipment is output as a fault state, including: The first reconstruction error is greater than a first error threshold corresponding to the working condition classification, and the running state of the equipment is output as a fault state.

7. The method of claim 1, wherein, The normal state model corresponding to the maximum similarity is determined as the target model, including: If the maximum similarity is greater than or equal to a similarity threshold, the normal state model corresponding to the maximum similarity is determined as the target model. If the maximum similarity is less than the similarity threshold, a first preset model is determined as the target model; the normal parameter range corresponding to the first model contains the normal parameter ranges of all working condition classifications.

8. An equipment failure early warning system, characterized by, Including: The data acquisition module is configured to acquire real-time working condition parameters of the equipment. The real-time working condition parameters include real-time running parameters and environmental parameters. The feature extraction module is configured to extract a first feature vector based on the real-time working condition parameters. The feature matching module is configured to calculate similarities between the first feature vector and a plurality of second feature vectors in a model library, respectively; each second feature vector in the model library is used to represent one working condition classification of the equipment, each working condition classification has a corresponding normal state model, and the normal state model is used to represent a normal parameter range of the equipment under the corresponding working condition classification. The model selection module is configured to determine the normal state model corresponding to the maximum similarity as a target model. The fault warning module is configured to input the real-time working condition parameters of the equipment into the target model to obtain a deviation between the real-time working condition parameters and the normal parameter range, and perform fault warning of the equipment based on the deviation.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, The processor executes the computer program to implement the steps of the method of any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, the computer program comprising instructions that, when executed by a computer, cause the computer to perform the method of any one of claims 1 to 9. The computer program is executed by the processor to implement the steps of the method of any one of claims 1 to 7.

Images