Intelligent fault diagnosis and early warning system for digital mechanical equipment

Abstract

The invention provides an intelligent fault diagnosis and early warning system for digital mechanical equipment. The system comprises a data acquisition module, a feature extraction module, an optimization training module and a diagnosis and early warning module. The data acquisition module collects historical operation data such as equipment state parameters and fault characteristic parameters. The feature extraction module inputs historical data to a network, outputs equipment state features and fault features, and further obtains a fault diagnosis probability, an evaluation value, an early warning level and an evaluation value. And the optimization training module optimizes the network based on the evaluation value and the probability until a preset number of iterations is reached, and a trained diagnosis decision network is obtained. And the diagnosis and early warning module inputs the current equipment operation data into the network to realize fault diagnosis and early warning. According to the invention, accurate diagnosis and real-time early warning of mechanical equipment faults can be realized, the reliability and maintenance efficiency of equipment operation are improved, and the downtime and maintenance cost of faults are reduced.

Description

Technical Field

[0001] The present invention relates to the field of digital mechanical equipment, and more specifically, to an intelligent fault diagnosis and early warning system for digital mechanical equipment. Background Art

[0002] In modern industrial production, the stable operation of mechanical equipment is crucial to ensuring production efficiency and product quality. With the continuous development of industrial automation, traditional fault diagnosis methods mainly rely on manual experience and regular inspections. This method is not only inefficient, but also difficult to detect potential faults in a timely and accurate manner, resulting in the inability to respond quickly when equipment fails suddenly, thus affecting production progress and increasing maintenance costs. In recent years, with the rise of digital technology, some fault diagnosis systems based on sensor data have begun to appear. They can monitor the operating status of equipment in real time, but most of these systems can only provide simple fault alarms, lack in-depth analysis and prediction capabilities of faults, and cannot meet the growing demand for equipment reliability in complex industrial environments.

[0003] In the process of implementing the embodiments of the present invention, the inventors found that there are at least the following problems or defects in the prior art: first, traditional fault diagnosis methods rely on manual experience, are highly subjective, and lack objective quantitative standards; second, although existing digital fault diagnosis systems can monitor in real time, they have limited functions in fault analysis and prediction and cannot provide early warning of potential faults; finally, these systems lack adaptability and accuracy when faced with complex and changeable industrial environments and equipment conditions, and it is difficult to meet the fault diagnosis needs under different equipment and working conditions. Summary of the invention

[0004] The invention provides a digital mechanical equipment intelligent fault diagnosis and early warning system.

[0005] In a first aspect of the present invention, a digital mechanical equipment intelligent fault diagnosis and early warning system is provided, comprising:

[0006] A data acquisition module is used to collect multiple groups of historical operation data, each group of historical operation data includes equipment status parameters, fault characteristic parameters, diagnostic actions, diagnostic results and warning levels; the collection moments corresponding to the equipment status parameters and the fault characteristic parameters are adjacent and sequential moments; the diagnostic results are calculated by a fault diagnosis function, which is constructed based on the performance indicators and fault modes of key components of the equipment; the warning level is calculated by a fault warning function, which is constructed by the equipment performance degradation rate and a preset safety threshold;

[0007] A feature extraction module is used to input the device state parameters and fault feature parameters of each group of historical operation data into the feature extraction network, respectively, to obtain the device state features and fault features; input the device state features into the diagnosis decision network, to obtain the fault diagnosis probability; input the device state features and fault features into the diagnosis result evaluation network, respectively, to obtain the first diagnosis evaluation value and the second diagnosis evaluation value; input the device state features and fault features into the warning level evaluation network, respectively, to obtain the first warning evaluation value and the second warning evaluation value; input the diagnosis result, the first diagnosis evaluation value and the second diagnosis evaluation value into the diagnosis advantage evaluation function, to obtain the diagnosis advantage evaluation value; input the warning level, the first warning evaluation value and the second warning evaluation value into the warning advantage evaluation function, to obtain the warning advantage evaluation value;

[0008] An optimization training module is used to input the diagnostic advantage evaluation value, the warning advantage evaluation value and the fault diagnosis probability into the optimization objective function and optimize to obtain an optimized diagnostic decision network; input the diagnostic result and the first diagnostic evaluation value into the loss function of the diagnostic result evaluation network and optimize to obtain an optimized diagnostic result evaluation network; input the warning level and the first warning evaluation value into the loss function of the warning level evaluation network and optimize to obtain an optimized warning level evaluation network; based on the optimized diagnostic decision network, the diagnostic result evaluation network and the warning level evaluation network, repeat the above process until the preset number of iterations is exceeded to obtain a trained diagnostic decision network;

[0009] The diagnosis and warning module is used to obtain the equipment operation data at the current moment and input it into the trained diagnosis decision network to obtain the current diagnostic action for fault diagnosis and warning; execute the current diagnostic action to obtain the equipment operation data at the next moment; and combine the historical diagnosis results and warning levels to predict the future failure trend of the equipment.

[0010] Furthermore, the optimization training module is also used to execute the current diagnostic action to obtain the equipment operation data at the next moment; calculate the current diagnostic result according to the fault diagnosis function; calculate the current warning level according to the fault warning function; construct the current operation status data based on the equipment operation data at the current moment, the equipment operation data at the next moment, the current diagnostic action, the current diagnostic result and the current warning level; optimize the diagnosis decision network, the diagnosis result evaluation network and the warning level evaluation network again based on the current operation status data, and use the re-optimized diagnosis decision network to process the equipment operation data at the next moment.

[0011] Furthermore, the optimization training module is also used to construct the loss function of the diagnosis result evaluation network and the warning level evaluation network using the mean square error, minimize the loss function through the stochastic gradient descent algorithm, update the parameters of the diagnosis result evaluation network and the parameters of the warning level evaluation network, and obtain the optimized diagnosis result evaluation network and the warning level evaluation network.

[0012] Furthermore, the expression of the fault diagnosis function is:

[0013]

[0014] in, Indicates the equipment fault diagnosis result. Indicates the performance indicators of key components of the equipment. Indicates the number of key components, Indicates The weight coefficients of the key components, Indicates Failure mode function of key components.

[0015] Furthermore, the expression of the fault warning function is:

[0016]

[0017] in, Indicates the equipment failure warning level. Indicates the rate at which device performance decreases. represents the early warning sensitivity coefficient, Indicates the preset safety threshold.

[0018] Furthermore, the expression of the diagnostic advantage evaluation function is:

[0019]

[0020] in, Indicated in The diagnostic advantage assessment value at the moment, Indicates the stage of strengths assessment, represents the discount factor, Indicates the first stage, Indicated in The fault diagnosis results at the moment, Represents the advantage value parameter, which is used to control the average degree of the advantage value. Indicated in The diagnostic evaluation value corresponding to the device status parameter at the moment.

[0021] Furthermore, the expression of the early warning advantage evaluation function is:

[0022]

[0023] in, Indicated in The early warning advantage assessment value at each moment, Indicates the stage of strengths assessment, represents the discount factor, Indicates the first stage, Indicated in Fault warning level at all times, Represents the advantage value parameter, which is used to control the average degree of the advantage value. Indicated in The warning assessment value corresponding to the equipment status parameter at the moment.

[0024] Furthermore, the expression of the optimization objective function is:

[0025]

[0026] in, represents the optimization target value, represents the parameters of the diagnosis decision network, represents the warning weight coefficient, Indicates the decision-making strategy in diagnosis The state distribution under Indicates in status Take action The probability of Indicates in status Take action The diagnostic advantage assessment value of Indicates in status Take action The early warning advantage assessment value.

[0027] In a second aspect of the present invention, a digital mechanical equipment intelligent fault diagnosis and early warning method is provided, comprising:

[0028] Collect multiple groups of historical operation data, each group of historical operation data includes equipment status parameters, fault characteristic parameters, diagnostic actions, diagnostic results and warning levels; the collection moments corresponding to the equipment status parameters and the fault characteristic parameters are adjacent and sequential moments; the diagnostic results are calculated by a fault diagnosis function, which is constructed based on the performance indicators and fault modes of key components of the equipment; the warning level is calculated by a fault warning function, which is constructed by the equipment performance degradation rate and a preset safety threshold;

[0029] The equipment state parameters and fault feature parameters of each set of historical operation data are respectively input into the feature extraction network to obtain equipment state features and fault features; the equipment state features are input into the diagnosis decision network to obtain the fault diagnosis probability; the equipment state features and fault features are respectively input into the diagnosis result evaluation network to obtain the first diagnosis evaluation value and the second diagnosis evaluation value; the equipment state features and fault features are respectively input into the warning level evaluation network to obtain the first warning evaluation value and the second warning evaluation value; the diagnosis result, the first diagnosis evaluation value and the second diagnosis evaluation value are input into the diagnosis advantage evaluation function to obtain the diagnosis advantage evaluation value; the warning level, the first warning evaluation value and the second warning evaluation value are input into the warning advantage evaluation function to obtain the warning advantage evaluation value;

[0030] Input the diagnostic advantage evaluation value, the warning advantage evaluation value and the fault diagnosis probability into the optimization objective function and optimize to obtain an optimized diagnostic decision network; input the diagnostic result and the first diagnostic evaluation value into the loss function of the diagnostic result evaluation network and optimize to obtain an optimized diagnostic result evaluation network; input the warning level and the first warning evaluation value into the loss function of the warning level evaluation network and optimize to obtain an optimized warning level evaluation network;

[0031] Based on the optimized diagnosis decision network, diagnosis result evaluation network and warning level evaluation network, the above process is repeated until the preset number of iterations is exceeded to obtain a trained diagnosis decision network;

[0032] The equipment operation data at the current moment is obtained and input into the trained diagnosis decision network to obtain the current diagnostic action for fault diagnosis and early warning.

[0033] Furthermore, the following steps are also included: executing the current diagnostic action to obtain the equipment operation data at the next moment; calculating the current diagnostic result according to the fault diagnosis function; calculating the current warning level according to the fault warning function; constructing the current operation status data based on the equipment operation data at the current moment, the equipment operation data at the next moment, the current diagnostic action, the current diagnostic result and the current warning level; optimizing the diagnostic decision network, the diagnostic result evaluation network and the warning level evaluation network again based on the current operation status data, and using the re-optimized diagnostic decision network to process the equipment operation data at the next moment.

[0034] According to the above-mentioned embodiments of the present invention, at least the following beneficial effects are achieved: the digital mechanical equipment intelligent fault diagnosis and early warning system can realize real-time and accurate monitoring of the equipment operation status. The data acquisition module collects comprehensive historical operation data, including equipment status parameters and fault characteristic parameters, etc., to provide a rich data basis for fault diagnosis. The feature extraction module can convert these data into equipment status characteristics and fault characteristics, and further input them into the diagnosis decision network and the evaluation network to obtain the fault diagnosis probability and evaluation value, the early warning level and evaluation value, so as to realize the accurate identification of faults and accurate evaluation of the early warning level. The optimization training module continuously optimizes the network based on the evaluation value and probability until the preset number of iterations is reached, and the trained diagnosis decision network is obtained, which ensures the reliability of the diagnosis result and the timeliness of the early warning.

[0035] In addition, the system can also effectively improve the maintenance efficiency of equipment and reduce maintenance costs. The diagnostic warning module inputs the current equipment operation data into the trained diagnostic decision network, quickly obtains the current diagnostic action, and performs fault diagnosis and warning in a timely manner, so that maintenance personnel can take measures in advance to avoid production interruptions caused by sudden equipment failures. At the same time, the system's intelligent diagnosis and warning functions reduce dependence on manual experience, improve the objectivity and accuracy of diagnosis, help achieve refined management and preventive maintenance of equipment, extend the service life of equipment, and ensure the continuity and stability of production. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] The above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood by reading the detailed description below with reference to the accompanying drawings. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, in which:

[0037] Figure 1 A schematic diagram of the structure of a digital mechanical equipment intelligent fault diagnosis and early warning system provided by an embodiment of the present invention;

[0038] Figure 2 A schematic diagram of a process flow of a digital mechanical equipment intelligent fault diagnosis and early warning method provided by an embodiment of the present invention;

[0039] Figure 3 The schematic diagram schematically shows the structure of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0040] The principles and spirit of the present invention will be described below with reference to several exemplary embodiments. It should be understood that these embodiments are provided only to enable those skilled in the art to better understand and implement the present invention, and are not intended to limit the scope of the present invention in any way. On the contrary, these embodiments are provided to make the present invention more thorough and complete, and to fully convey the scope of the present invention to those skilled in the art.

[0041] Those skilled in the art know that the embodiments of the present invention can be implemented as a system, device, apparatus, method or computer program product. Therefore, the present invention can be specifically implemented in the following forms, namely: complete hardware, complete software (including firmware, resident software, microcode, etc.), or a combination of hardware and software.

[0042] It should be noted that any number of elements in the drawings is for illustration rather than limitation, and any naming is only for distinction and does not have any limiting meaning.

[0043] Reference below Figure 1 , Figure 1 This is a schematic diagram of the structure of a digital mechanical equipment intelligent fault diagnosis and early warning system provided by an embodiment of the present invention. Figure 1 As shown, a digital mechanical equipment intelligent fault diagnosis and early warning system 100 includes:

[0044] The data collection module 101 is used to collect multiple groups of historical operation data, each group of historical operation data includes equipment status parameters, fault characteristic parameters, diagnostic actions, diagnostic results and warning levels; the collection moments corresponding to the equipment status parameters and the fault characteristic parameters are adjacent and sequential moments; the diagnostic results are calculated by a fault diagnosis function, which is constructed based on the performance indicators and fault modes of key components of the equipment; the warning level is calculated by a fault warning function, which is constructed by the equipment performance degradation rate and a preset safety threshold;

[0045] The feature extraction module 102 is used to input the device state parameters and fault feature parameters of each group of historical operation data into the feature extraction network to obtain the device state feature and the fault feature; input the device state feature into the diagnosis decision network to obtain the fault diagnosis probability; input the device state feature and the fault feature into the diagnosis result evaluation network to obtain the first diagnosis evaluation value and the second diagnosis evaluation value; input the device state feature and the fault feature into the warning level evaluation network to obtain the first warning evaluation value and the second warning evaluation value; input the diagnosis result, the first diagnosis evaluation value and the second diagnosis evaluation value into the diagnosis advantage evaluation function to obtain the diagnosis advantage evaluation value; input the warning level, the first warning evaluation value and the second warning evaluation value into the warning advantage evaluation function to obtain the warning advantage evaluation value;

[0046] The optimization training module 103 is used to input the diagnostic advantage evaluation value, the warning advantage evaluation value and the fault diagnosis probability into the optimization objective function and optimize to obtain an optimized diagnostic decision network; input the diagnostic result and the first diagnostic evaluation value into the loss function of the diagnostic result evaluation network and optimize to obtain an optimized diagnostic result evaluation network; input the warning level and the first warning evaluation value into the loss function of the warning level evaluation network and optimize to obtain an optimized warning level evaluation network; based on the optimized diagnostic decision network, the diagnostic result evaluation network and the warning level evaluation network, repeat the above process until the preset number of iterations is exceeded to obtain a trained diagnostic decision network;

[0047] The diagnosis and warning module 104 is used to obtain the equipment operation data at the current moment and input it into the trained diagnosis decision network to obtain the current diagnosis action for fault diagnosis and warning.

[0048] It should be noted that the digital mechanical equipment intelligent fault diagnosis and early warning system covers multiple key modules such as data acquisition, feature extraction, optimization training, and diagnosis and early warning. The function of the data acquisition module is to collect various historical data of mechanical equipment during operation. These data include equipment status parameters, such as temperature, pressure, vibration frequency, etc., and fault characteristic parameters, such as the intensity of abnormal sound, the fluctuation amplitude of current, etc. The acquisition moments of these parameters are adjacent and sequential moments, which means that they are obtained in the continuous process of equipment operation so as to accurately reflect the changes in equipment status. The diagnosis result is calculated by the fault diagnosis function, which is based on the performance indicators of key components of the equipment, such as the degree of wear of bearings, the efficiency of motors, etc., and the possible failure modes of these components. The warning level is determined by the fault warning function, which takes into account the rate of equipment performance degradation, such as the gradual reduction of equipment output power, and the preset safety threshold, that is, when the equipment performance degrades to what extent, the warning will be triggered.

[0049] The fault diagnosis function and warning function in the data acquisition module are based on preset expert rules and linear models, which are used to generate preliminary diagnosis results and warning levels of historical data, and provide label data for subsequent network training. The diagnostic decision network in the feature extraction module extracts nonlinear features from the original data through deep learning models (such as CNN / RNN), and combines the dynamic parameter adjustment of the optimization training module to achieve adaptive recognition of complex fault modes. The difference between the two is that the data acquisition module is based on artificial experience and fixed rules, which is suitable for fast calculation of known fault modes, but lacks the ability to generalize unknown faults and nonlinear features; the diagnostic decision network learns the deep correlation of equipment status through data-driven learning, and can be dynamically optimized to adapt to complex scenarios such as equipment aging and environmental changes, thereby improving diagnostic accuracy and warning timeliness.

[0050] Specifically, equipment status parameters refer to various physical quantities that can directly or indirectly reflect the operating status of the equipment. For example, for a motor, its current, voltage, temperature, etc. during operation are all equipment status parameters. Fault characteristic parameters are special physical quantities that appear when a device fails, such as the abnormal vibration frequency of a motor when the bearing is damaged. The fault diagnosis function is a mathematical model that takes the performance indicators and failure modes of key components of the equipment as input, and outputs the diagnosis result of whether the equipment has a fault through a certain algorithm, such as logistic regression or neural network algorithm. The warning level assessment is determined based on the rate of equipment performance degradation and the preset safety threshold. If the rate of equipment performance degradation exceeds the safety threshold, the warning level will be increased accordingly to remind the operator to take measures.

[0051] Preferably, the feature extraction module can use convolutional neural network (CNN) or recurrent neural network (RNN) in deep learning to process the collected data and extract more representative equipment status features and fault features. For example, for time series data, such as vibration signals during equipment operation, RNN can better capture the temporal relationship in the data. In the optimization training module, more advanced optimization algorithms, such as the Adam optimization algorithm, can be used to replace the traditional stochastic gradient descent algorithm to speed up the convergence of the network and improve the optimization effect.

[0052] The feature extraction module is used to input the equipment state parameters and fault feature parameters in the historical operation data into the feature extraction network to generate equipment state features and fault features; the diagnostic decision network is based on the output of the feature extraction network and combined with the diagnostic advantage evaluation value in the optimization objective function. and early warning advantage assessment value The weighted optimization results are used to generate a dynamic diagnosis strategy, whose output works together with the fault diagnosis function and warning function preset in the data acquisition module in the following scenarios: The function of the data acquisition module is used to generate training labels and initial warning thresholds; The diagnosis decision network is used to integrate multi-dimensional features in real-time diagnosis to improve the ability to identify nonlinear faults.

[0053] Traditional function methods (such as fault diagnosis function Relying on manually defined critical component weights wi and failure mode functions , its limitations include at least:

[0054] 1. Weight coefficient It needs to be preset manually and cannot be adjusted dynamically according to the actual operating status of the equipment;

[0055] 2. The function form is linear superposition ,It is difficult to capture the nonlinear characteristics of multi-component coordinated failures;

[0056] 3. Unable to automatically adapt to equipment performance degradation or changes in the external environment (such as temperature and humidity).

[0057] The diagnostic decision network of this application solves the above problems in the following ways:

[0058] 1. Use convolutional layers / recurrent layers to extract spatial and temporal dependency features in time series data;

[0059] 2. Combined diagnostic advantage assessment and early warning advantage assessment , achieving a dynamic balance between diagnostic accuracy and early warning sensitivity;

[0060] 3. By optimizing the training module, new data can be integrated in real time, so that the network can continuously adapt to changes in device status.

[0061] Furthermore, the diagnostic decision network can be designed as a multi-layer perceptron (MLP) structure, which can improve its ability to recognize complex fault modes by increasing the depth and width of the network. In the diagnostic warning module, in addition to outputting the current diagnostic action, it can also combine historical diagnostic results and warning levels to predict possible future failure trends of the equipment, providing a more comprehensive reference for preventive maintenance of the equipment. During the training process, the diagnostic decision network extracts the relationship between the equipment operating parameters and the equipment fault characteristics in the operating data, and forms a set of corresponding feature vectors with the equipment operating parameters and the corresponding equipment fault characteristics in each set of operating data. The diagnostic decision network formed after training can not only analyze the real-time operating data of the equipment to obtain the corresponding diagnostic results and fault results, but also predict the equipment operating data at the next moment. The relationship between the equipment operating parameters and the equipment fault characteristics obtained during the training process can be used to predict equipment failures in the equipment operating data at each future moment;

[0062] In some embodiments, the optimization training module is also used to execute the current diagnostic action to obtain the equipment operation data at the next moment; calculate the current diagnostic result according to the fault diagnosis function; calculate the current warning level according to the fault warning function; construct the current operation status data based on the equipment operation data at the current moment, the equipment operation data at the next moment, the current diagnostic action, the current diagnostic result and the current warning level; optimize the diagnostic decision network, the diagnostic result evaluation network and the warning level evaluation network again based on the current operation status data, and use the re-optimized diagnostic decision network to process the equipment operation data at the next moment.

[0063] It should be noted that the optimization training module plays a vital role in the system. It not only performs the current diagnostic action to obtain the equipment operation data at the next moment, but also calculates the current diagnostic results and warning levels based on the fault diagnosis function and fault warning function. These data are then integrated and constructed into the current operating status data, which covers the overall operating status of the equipment at a specific moment, including but not limited to the performance indicators, fault characteristics, and warning status of the equipment. Based on these operating status data, the optimization training module can further optimize the diagnostic decision network, the diagnostic result evaluation network, and the warning level evaluation network to ensure that the network can self-adjust and update according to the latest equipment operation data to improve the accuracy of diagnosis and warning. The optimized diagnostic decision network will be used to process the equipment operation data at the next moment to achieve a dynamic, self-optimizing fault diagnosis and warning process.

[0064] Specifically, the current diagnostic action refers to the actual operation of the equipment according to the suggestions output by the diagnostic decision network, such as adjusting the operating parameters of the equipment or conducting a local inspection. The equipment operation data at the next moment refers to the new data generated by the equipment in the time period immediately after the diagnostic action is executed. The diagnostic decision network obtains the equipment adjustment operating parameter data based on the current diagnostic action, i.e., the output suggestions, and uses the adjusted operating parameter data as the equipment operation data at the next moment. At the same time, the equipment failure situation at the next moment is predicted based on the equipment operation data at the next moment, so as to achieve the purpose of predicting the possible failure trend of the equipment in the future. The fault diagnosis function and the fault warning function are preset algorithms in the system, which are used to calculate the fault diagnosis results and warning levels of the equipment according to the key performance indicators and failure modes of the equipment, as well as the performance degradation rate and safety threshold. The specific parameter settings of these functions depend on the type of equipment and the operating environment. For example, for a high-precision CNC machine tool, its fault diagnosis function may pay more attention to key performance indicators such as machining accuracy and tool wear, while the warning function may set the safety threshold according to the machining error and tool wear rate of the machine tool.

[0065] Preferably, the optimization training module can use a reinforcement learning algorithm to further optimize the network. In this case, the system will dynamically adjust the strategy of the diagnostic decision network according to the operating status and diagnostic results of the equipment to maximize the long-term diagnostic accuracy and early warning timeliness. For example, the system can train the network to recognize various failure modes by simulating different failure scenarios, and adjust the network parameters according to the simulation results.

[0066] Furthermore, in order to improve the adaptability of the system, transfer learning technology can be introduced so that the network trained on one device can be quickly adapted to similar devices, reducing the time and resource consumption of retraining. When constructing the current operating status data, in addition to the basic equipment status parameters and fault characteristic parameters, environmental factors such as temperature and humidity can also be included. These factors may affect the performance of the equipment, thereby providing support for more comprehensive fault diagnosis and early warning.

[0067] In some embodiments, the optimization training module is also used to construct the loss function of the diagnosis result evaluation network and the warning level evaluation network using the mean square error, minimize the loss function through the stochastic gradient descent algorithm, update the parameters of the diagnosis result evaluation network and the parameters of the warning level evaluation network, and obtain the optimized diagnosis result evaluation network and the warning level evaluation network.

[0068] It should be noted that the optimization training module plays a key role in network optimization in the digital mechanical equipment intelligent fault diagnosis and early warning system. The module uses mean square error to construct the loss function of the diagnosis result evaluation network and the early warning level evaluation network, which is a common method to measure the difference between the predicted value and the actual value. Through the stochastic gradient descent algorithm, the module can minimize the loss function and then update the parameters of the diagnosis result evaluation network and the early warning level evaluation network. This process enables the network to automatically adjust internal parameters according to the actual diagnosis results and early warning levels to improve the accuracy of the prediction. In short, this mechanism ensures that the system can continuously optimize its own diagnosis and early warning capabilities by learning from historical data.

[0069] Specifically, the mean square error is obtained by calculating the average of the squares of the differences between the predicted value and the true value. It is an indicator that quantifies the accuracy of the prediction. When constructing the loss function, the mean square error can intuitively reflect the deviation between the network's predicted results and the actual results. The stochastic gradient descent algorithm is an optimization algorithm that iteratively adjusts the network parameters to gradually reduce the value of the loss function. In each iteration, the algorithm calculates the gradient of the loss function based on the current parameters and updates the parameters in the opposite direction of the gradient, thereby gradually approaching the optimal solution. For example, if the fault probability predicted by the diagnostic result evaluation network deviates from the actual fault probability, the stochastic gradient descent algorithm will adjust the weights and biases of the network to reduce this deviation.

[0070] Preferably, in addition to using mean square error as the loss function, other more complex loss functions can also be considered, such as the cross entropy loss function, especially when dealing with classification problems, the cross entropy loss function can more effectively measure the difference between the predicted probability distribution and the true probability distribution. In addition, when updating network parameters, more advanced optimization algorithms can be used, such as the Adam algorithm or the RMSprop algorithm, which can usually provide faster convergence speed and better optimization effect when dealing with complex network structures and large-scale data sets.

[0071] More specifically, when implementing the stochastic gradient descent algorithm, batch gradient descent or mini-batch gradient descent can be used instead of pure stochastic gradient descent. Batch gradient descent uses the entire dataset to calculate the gradient, which can provide more stable parameter updates, but the computational cost is higher; mini-batch gradient descent strikes a balance between the two. It uses a small batch of the dataset to approximate the gradient, which not only maintains the stability of parameter updates but also reduces the computational cost.

[0072] In some embodiments, the expression of the fault diagnosis function is:

[0073]

[0074] in, Indicates the equipment fault diagnosis result. Indicates the performance indicators of key components of the equipment. Indicates the number of key components, Indicates The weight coefficients of the key components, Indicates Failure mode function of key components.

[0075] It should be noted that the fault diagnosis function is a core component in the digital mechanical equipment intelligent fault diagnosis and early warning system, which is used to calculate the fault diagnosis results of the equipment. This function comprehensively considers the performance indicators and failure modes of the key components of the equipment, quantifies and integrates these factors in the form of mathematical expressions, and thus draws a conclusion on whether the equipment has a fault. Specifically, the expression of the fault diagnosis function is:

[0076]

[0077] in, Indicates the equipment fault diagnosis result. Indicates the number of key components of the equipment, Indicates The weight coefficient of each key component reflects the importance of the component in the overall fault diagnosis. Indicates The failure mode function of each key component describes the characteristics and manifestations of the component failure. The construction of this function enables the system to evaluate the overall failure status of the equipment based on the performance and failure mode of the key components.

[0078] Specifically, the performance indicators of key equipment components may include various parameters that can reflect the operating status of the equipment, such as motor speed, bearing temperature, hydraulic system pressure, etc. Failure mode functions can be defined according to different components and failure types. For example, for a bearing whose performance has degraded due to wear, its failure mode function may be related to the degree of wear and vibration frequency. Weight coefficient The setting of can be determined according to the degree of influence of the component on the overall performance of the equipment, which is usually obtained through expert experience or data analysis. For example, if the failure of a component has a significant impact on the production efficiency of the equipment, then its weight coefficient will be relatively high.

[0079] Preferably, the weight coefficient in the fault diagnosis function Instead of being fixed, it can be automatically adjusted through machine learning algorithms. For example, algorithms such as support vector machines (SVM) or neural networks can be used to train the model to automatically learn the weight of each key component based on historical failure data and equipment operation data.

[0080] Furthermore, the failure mode function More complex models, such as fault simulation based on physical models, or combined with data-driven methods can be used to more accurately characterize the fault characteristics.

[0081] More specifically, when implementing the fault diagnosis function, we can consider introducing real-time data stream processing technology so that the system can receive and process equipment operation data in real time and update the fault diagnosis results in a timely manner. At the same time, in order to improve the accuracy of diagnosis, we can use multi-source data fusion methods to comprehensively analyze data from different sensors to reduce the errors that may be caused by a single data source.

[0082] In some embodiments, the expression of the fault warning function is:

[0083]

[0084] in, Indicates the equipment failure warning level. Indicates the rate at which device performance decreases. represents the early warning sensitivity coefficient, Indicates the preset safety threshold.

[0085] It should be noted that the fault warning function is used to evaluate the fault warning level of the equipment in the intelligent fault diagnosis and warning system of digital mechanical equipment, and its expression is:

[0086]

[0087] In this formula, Indicates the equipment failure warning level. Indicates the rate of device performance degradation, reflecting how fast the device performance deteriorates over time; Represents the warning sensitivity coefficient, which is used to adjust the sensitivity of the warning level; Represents the preset safety threshold, which is the critical value that triggers an early warning when the device performance drops to a certain level. This function quantifies the failure risk of the device through the relationship between the device performance degradation rate and the preset safety threshold, and provides early warning information for maintenance personnel so that they can take timely measures.

[0088] Specifically, the rate at which device performance decreases It can be calculated by monitoring the changes in the key performance indicators of the equipment over time, for example, by analyzing the rate at which the motor efficiency decreases with the running time. It can be set according to actual needs. If you want the system to be more sensitive to failures, you can appropriately reduce Preset safety threshold It is determined based on the equipment's safe operation standards and historical fault data. It is a critical value of a performance indicator. When the equipment performance drops to this value, the system will issue an early warning. For example, for a generator, if its output power drops to 80% of the rated power, it can be considered to have reached the preset safety threshold.

[0089] Preferably, the parameters in the fault warning function are , and Automatic adjustments can be made through data analysis and machine learning methods. For example, historical failure data and equipment operation data can be used to train a regression model to predict the rate at which equipment performance deteriorates. At the same time, the early warning sensitivity coefficient It can be dynamically adjusted according to different equipment types and operating environments to meet different early warning needs.

[0090] More specifically, when implementing the fault warning function, real-time monitoring data and prediction models can be combined to predict the downward trend of equipment performance in advance, and the preset safety threshold can be dynamically adjusted according to the prediction results. In addition, in order to improve the accuracy of early warning, multi-dimensional performance indicators can be introduced to comprehensively evaluate the failure risk of equipment, rather than relying solely on a single performance indicator.

[0091] In some embodiments, the diagnostic advantage evaluation function is expressed as:

[0092]

[0093] in, Indicated in The diagnostic advantage assessment value at the moment, Indicates the stage of strengths assessment, represents the discount factor, Indicates the first stage, Indicated in The fault diagnosis results at the moment, Represents the advantage value parameter, which is used to control the average degree of the advantage value. Indicated in The diagnostic evaluation value corresponding to the device status parameter at the moment.

[0094] It should be noted that the diagnostic advantage evaluation function is used in the digital mechanical equipment intelligent fault diagnosis and early warning system to quantitatively evaluate the advantages of diagnostic actions at different time stages. The expression of this function is:

[0095]

[0096] in, Indicates at time The diagnostic advantage assessment value of represents the number of stages of advantage evaluation, represents the discount factor used to adjust the current value of future rewards; Indicates at time The fault diagnosis results, represents the advantage value parameter, which is used to control the average degree of advantage value; and Respectively, in time and This function evaluates the effectiveness and value of diagnostic actions by comparing diagnostic results and evaluation values ​​at different time points.

[0097] Specifically, the parameters in the diagnostic advantage evaluation function are set as follows: Number of stages It can be determined based on the time span and complexity of the diagnostic process. For example, for a short diagnostic process, May be smaller; discount factor It is usually set between 0 and 1. The closer the value is to 1, the more importance is attached to the future diagnosis results. It is used to balance the diagnostic evaluation values ​​at different time points. The setting of its value depends on the requirements for diagnostic stability. For example, if you want the diagnostic results to be more stable, you can increase The value of .

[0098] Preferably, the parameters in the diagnostic advantage evaluation function can be automatically adjusted by a machine learning algorithm to adapt to different diagnostic scenarios and device characteristics. For example, a reinforcement learning algorithm can be used to dynamically adjust the discount factor based on historical diagnostic data and device operation data. and the advantage value parameter , in order to optimize diagnostic strategies.

[0099] More specifically, when implementing the diagnostic advantage evaluation function, we can consider introducing a time decay factor to make the function pay more attention to recent diagnostic results, thereby increasing the sensitivity to the current state of the equipment. In addition, in order to improve the accuracy of the evaluation, we can combine multiple diagnostic indicators and methods to comprehensively evaluate the advantages of diagnostic actions, rather than relying solely on a single fault diagnosis result.

[0100] In some embodiments, the early warning advantage evaluation function is expressed as:

[0101]

[0102] in, Indicated in The early warning advantage assessment value at each moment, Indicates the stage of strengths assessment, represents the discount factor, Indicates the first stage, Indicated in Fault warning level at all times, Represents the advantage value parameter, which is used to control the average degree of the advantage value. Indicated in The warning assessment value corresponding to the equipment status parameter at the moment.

[0103] The early warning advantage evaluation function is an important part of the intelligent fault diagnosis and early warning system of digital mechanical equipment, which is used to evaluate the effectiveness and value of early warning signals at different time stages. The expression of this function is:

[0104]

[0105] in, Indicates at time The early warning advantage assessment value, represents the number of stages of advantage evaluation, is the discount factor used to adjust the current value of future warning signals; Indicates at time Fault warning level, is the advantage value parameter, which is used to control the average degree of advantage value; and Respectively, in time and The warning evaluation value corresponding to the device status parameter. This function helps the system evaluate the effectiveness of the warning strategy by comparing the warning level and evaluation value at different times, so as to adjust the warning mechanism in time.

[0106] Specifically, the parameters in the early warning advantage evaluation function are set as follows: Number of stages It can be determined based on the monitoring cycle of the early warning system and the operating characteristics of the equipment. For example, for high-risk equipment that requires frequent monitoring, May be larger; discount factor It is usually set between 0 and 1. The closer its value is to 1, the higher the importance attached to future warning signals. Used to balance the early warning evaluation values ​​at different times. The setting of its value depends on the requirements for early warning stability. For example, if you want the early warning system to respond more smoothly to changes in equipment status, you can appropriately increase The value of .

[0107] Preferably, the parameters in the early warning advantage evaluation function can be optimized by a data-driven approach. For example, the discount factor can be automatically adjusted by a machine learning algorithm using historical early warning data and equipment failure records. and the advantage value parameter , to improve the accuracy and response speed of the early warning system.

[0108] More specifically, when implementing the early warning advantage evaluation function, it is possible to consider introducing a dynamic adjustment mechanism to update parameter settings in real time according to the actual operating conditions of the equipment and changes in the external environment. In addition, in order to improve the reliability of early warning, a variety of early warning indicators and methods can be combined, such as the trend of equipment performance degradation, the frequency of abnormal events, etc., to comprehensively evaluate the advantages of early warning signals, rather than relying solely on a single warning level.

[0109] In some embodiments, the expression of the optimization objective function is:

[0110]

[0111] in, represents the optimization target value, represents the parameters of the diagnosis decision network, represents the warning weight coefficient, Indicates the decision-making strategy in diagnosis The state distribution under Indicates in status Take action The probability of Indicates in status Take action The diagnostic advantage assessment value of Indicates in status Take action The early warning advantage assessment value.

[0112] It should be noted that the optimization objective function is used to guide the optimization process of the network in the digital mechanical equipment intelligent fault diagnosis and early warning system to ensure that the diagnosis decision network can achieve the best balance between diagnosis accuracy and early warning effectiveness. The expression of this function is:

[0113]

[0114] in, represents the optimization target value, represents the parameters of the diagnosis decision network, Represents the warning weight coefficient, which is used to adjust the importance of warning in the optimization process; Indicates the decision-making strategy in diagnosis The state distribution under is the discount factor used to adjust the current value of future rewards; Indicates at time The diagnostic advantage assessment value of taking action, Represents the early warning advantage evaluation value. This function optimizes the parameters of the diagnosis decision network by comprehensively considering the diagnosis advantage and early warning advantage to improve the overall performance of the system.

[0115] Specifically, the parameters in the optimization objective function are set as follows: The parameters of the diagnosis decision network Including the weights and biases of the network, these parameters are continuously updated through the training process to improve the accuracy of diagnosis; warning weight coefficient It can be set according to the importance and urgency of the warning. For example, in high-risk equipment, the importance of the warning may be higher, so The value of can be set larger; the discount factor It is usually set between 0 and 1. The closer the value is to 1, the more attention is paid to future diagnosis and warning results. For example, if you want the system to pay more attention to the long-term health of the equipment, you can increase The value of .

[0116] Preferably, the parameters in the optimization objective function can be optimized by an automated tuning method. For example, techniques such as Bayesian optimization or genetic algorithms can be used to automatically search for the optimal and value to maximize the optimization objective function .

[0117] More specifically, when implementing the optimization objective function, we can consider introducing a multi-objective optimization strategy, taking into account multiple objectives such as diagnostic accuracy and early warning timeliness, and finding the optimal parameter combination by constructing the Pareto frontier. In addition, in order to improve the efficiency of the optimization process, we can use distributed computing or parallel computing technology to accelerate the parameter search and network training process.

[0118] The above-mentioned embodiments of the present invention have the following beneficial effects: the digital mechanical equipment intelligent fault diagnosis and early warning system collects multiple groups of historical operation data such as equipment state parameters and fault characteristic parameters through the data acquisition module, and the acquisition time of these parameters is adjacent and sequential time, which can ensure the timeliness and relevance of the collected data, and provide accurate data support for subsequent fault diagnosis and early warning. The feature extraction module inputs the collected data into the feature extraction network to obtain equipment state characteristics and fault characteristics, and then inputs them into the diagnosis decision network, the diagnosis result evaluation network and the early warning level evaluation network respectively to obtain the corresponding evaluation value and probability. This process can realize the deep mining and analysis of equipment status and fault characteristics, and provide a scientific basis for fault diagnosis and early warning. The optimization training module inputs the evaluation value and probability into the optimization objective function for optimization until the preset number of iterations is exceeded to obtain the trained diagnosis decision network, which can continuously improve the accuracy and reliability of the diagnosis decision network, so that it can better adapt to the fault diagnosis needs under different equipment and working conditions. The diagnosis and early warning module obtains the equipment operation data at the current moment and inputs it into the trained diagnosis decision network to obtain the current diagnostic action for fault diagnosis and early warning, thereby realizing real-time monitoring and rapid response to equipment failures and improving the stability and safety of equipment operation.

[0119] The optimization training module in the system is also used to execute the current diagnostic action, obtain the equipment operation data at the next moment, and optimize the diagnostic decision network, diagnostic result evaluation network and warning level evaluation network based on this, which can realize dynamic optimization and continuous improvement of the network, so that it can self-adjust and update according to the actual operation of the equipment, and further improve the accuracy of diagnosis and warning. In addition, the module also uses the mean square error to construct the loss function, and minimizes the loss function through the stochastic gradient descent algorithm, updates the network parameters, and obtains the optimized evaluation network, which can effectively reduce the error of the diagnosis and warning results and improve the overall performance of the system. The setting of the fault diagnosis function and the fault warning function can accurately calculate the fault diagnosis results and warning levels based on the performance indicators and failure modes of the key components of the equipment, as well as the equipment performance degradation rate and the preset safety threshold, and provide clear guidance for the maintenance and management of the equipment. The introduction of the diagnostic advantage evaluation function and the warning advantage evaluation function can quantitatively evaluate the advantages of diagnosis and warning, further optimize the diagnostic decision and warning strategy, and improve the decision quality and warning effect of the system. The construction of the optimization objective function can comprehensively consider the diagnostic advantage evaluation value, the early warning advantage evaluation value and the fault diagnosis probability, and realize the comprehensive optimization of the diagnosis decision network, so as to maximize the diagnostic advantage and early warning advantage while satisfying the early warning weight coefficient, and improve the overall diagnosis and early warning capabilities of the system.

[0120] like Figure 2 As shown, in some embodiments, a digital mechanical equipment intelligent fault diagnosis and early warning method 200 includes:

[0121] Step 201, collecting multiple groups of historical operation data, each group of historical operation data includes equipment state parameters, fault characteristic parameters, diagnostic actions, diagnostic results and warning levels; the collection moments corresponding to the equipment state parameters and the fault characteristic parameters are adjacent and sequential moments; the diagnostic results are calculated by a fault diagnosis function, which is constructed based on performance indicators and fault modes of key components of the equipment; the warning level is calculated by a fault warning function, which is constructed by the equipment performance degradation rate and a preset safety threshold;

[0122] Step 202, input the device state parameters and fault feature parameters of each group of historical operation data into the feature extraction network respectively, to obtain the device state features and fault features; input the device state features into the diagnosis decision network to obtain the fault diagnosis probability; input the device state features and fault features into the diagnosis result evaluation network respectively, to obtain the first diagnosis evaluation value and the second diagnosis evaluation value; input the device state features and fault features into the warning level evaluation network respectively, to obtain the first warning evaluation value and the second warning evaluation value; input the diagnosis result, the first diagnosis evaluation value and the second diagnosis evaluation value into the diagnosis advantage evaluation function, to obtain the diagnosis advantage evaluation value; input the warning level, the first warning evaluation value and the second warning evaluation value into the warning advantage evaluation function, to obtain the warning advantage evaluation value;

[0123] Step 203, input the diagnostic advantage evaluation value, the warning advantage evaluation value and the fault diagnosis probability into the optimization objective function and optimize to obtain an optimized diagnostic decision network; input the diagnostic result and the first diagnostic evaluation value into the loss function of the diagnostic result evaluation network and optimize to obtain an optimized diagnostic result evaluation network; input the warning level and the first warning evaluation value into the loss function of the warning level evaluation network and optimize to obtain an optimized warning level evaluation network;

[0124] Step 204, based on the optimized diagnosis decision network, diagnosis result evaluation network and warning level evaluation network, repeat the above process until the preset number of iterations is exceeded to obtain a trained diagnosis decision network;

[0125] Step 205, obtain the equipment operation data at the current moment, and input it into the trained diagnosis decision network to obtain the current diagnosis action for fault diagnosis and early warning.

[0126] It is understandable that the steps and references in the digital mechanical equipment intelligent fault diagnosis and early warning method 200 are Figure 1 The modules described above correspond to the modules in the digital mechanical equipment intelligent fault diagnosis and early warning system. Therefore, the modules, features and beneficial effects described above for the digital mechanical equipment intelligent fault diagnosis and early warning system are also applicable to the digital mechanical equipment intelligent fault diagnosis and early warning method 200 and the operations contained therein, and will not be repeated here.

[0127] In some embodiments, the following steps are also included: executing the current diagnostic action to obtain the equipment operation data at the next moment; calculating the current diagnostic result according to the fault diagnosis function; calculating the current warning level according to the fault warning function; constructing the current operation status data based on the equipment operation data at the current moment, the equipment operation data at the next moment, the current diagnostic action, the current diagnostic result and the current warning level; re-optimizing the diagnostic decision network, the diagnostic result evaluation network and the warning level evaluation network based on the current operation status data, and using the re-optimized diagnostic decision network to process the equipment operation data at the next moment.

[0128] It should be noted that the digital mechanical equipment intelligent fault diagnosis and early warning method covers the complete process from data collection to fault diagnosis and early warning. First, multiple sets of historical operation data are collected, including equipment state parameters, fault characteristic parameters, diagnostic actions, diagnostic results and early warning levels. Equipment state parameters refer to physical quantities that can directly reflect the operating state of the equipment, such as temperature, pressure, vibration, etc.; fault characteristic parameters are special physical quantities that appear when the equipment fails, such as the intensity of abnormal sound, the fluctuation amplitude of current, etc. The diagnosis result and early warning level are calculated by the fault diagnosis function and the fault early warning function, respectively. These two functions are constructed based on the performance indicators and failure modes of the key components of the equipment, as well as the equipment performance degradation rate and the preset safety threshold. Through the feature extraction network, the equipment state parameters and fault characteristic parameters are converted into equipment state features and fault features, and then the fault diagnosis probability and evaluation value, early warning level and evaluation value are obtained. These evaluation values ​​and probabilities are then input into the optimization objective function for optimization until the preset number of iterations is exceeded to obtain the trained diagnosis decision network. Finally, the equipment operation data at the current moment is obtained and input into the trained diagnosis decision network to obtain the current diagnosis action for fault diagnosis and early warning.

[0129] Specifically, the collection of equipment status parameters and fault characteristic parameters can be achieved through various sensors installed on the equipment, such as temperature sensors, pressure sensors, vibration sensors, etc. The specific forms of fault diagnosis functions and fault warning functions can be designed according to the type of equipment and the operating environment. For example, for a motor, its fault diagnosis function may be constructed based on performance indicators such as the motor's current, voltage, and temperature, as well as possible fault modes such as overheating and short circuit. The feature extraction network can use convolutional neural networks (CNNs) or recurrent neural networks (RNNs) in deep learning. The specific parameter settings include the number of network layers, the number of neurons in each layer, the activation function, etc. The preset number of iterations in the optimization objective function can be determined based on actual needs and computing resources. For example, if the computing resources are sufficient, the number of iterations can be appropriately increased to improve the training effect of the network.

[0130] Preferably, in order to improve the accuracy of diagnosis and the timeliness of warning, an attention mechanism can be introduced into the feature extraction network so that the network can pay more attention to the features related to fault diagnosis and warning. In the optimization objective function, a regularization term, such as L1 or L2 regularization, can be introduced to prevent the network from overfitting.

[0131] More specifically, when implementing diagnostic actions, the output of the diagnostic decision network can be verified twice in combination with the advice of the expert system to improve the reliability of the diagnosis. In addition, in order to cope with changes in the equipment operating environment, the parameters in the fault diagnosis function and the fault warning function can be updated regularly to ensure that the system can adapt to new operating conditions.

[0132] Reference below Figure 3 , which shows a schematic diagram of the structure of an electronic device 300 suitable for implementing some embodiments of the present invention. The electronic devices in some embodiments of the present invention may include but are not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 3 The terminal device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0133] like Figure 3 As shown, the electronic device 300 may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 301, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 302 or a program loaded from a storage device 308 to a random access memory (RAM) 303. In the RAM 303, various programs and data required for the operation of the electronic device 300 are also stored. The processing device 301, the ROM 302, and the RAM 303 are connected to each other via a bus 304. An input / output (I / O) interface 305 is also connected to the bus 304.

[0134] Typically, the following devices may be connected to the I / O interface 305: an input device 306 including, for example, a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 307 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 308 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 309. The communication device 309 may allow the electronic device 300 to communicate with other devices wirelessly or by wire to exchange data. Figure 3The electronic device 300 is shown with various devices, but it should be understood that it is not required to implement or possess all the devices shown. More or fewer devices may be implemented or possessed instead. Figure 3 Each block shown in the figure may represent one device, or may represent multiple devices as required.

[0135] Furthermore, the storage medium of the embodiment of the present application stores program instructions that can implement all the above methods, wherein the program instructions can be stored in the above storage medium in the form of a software product, including several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) or a processor to execute all or part of the steps of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, or terminal devices such as a computer, a server, a mobile phone, and a tablet.

[0136] The above descriptions are only some preferred embodiments of the present invention and an explanation of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the embodiments of the present invention is not limited to the technical solutions formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above inventive concept. For example, the above features are replaced with (but not limited to) technical features with similar functions disclosed in the embodiments of the present invention.

Claims

1. A digital mechanical equipment intelligent fault diagnosis and early warning system, characterized in that: include: A data acquisition module is used to collect multiple groups of historical operation data, each group of historical operation data includes equipment status parameters, fault characteristic parameters, diagnostic actions, diagnostic results and warning levels; the acquisition moments corresponding to the equipment status parameters and the fault characteristic parameters are adjacent moments; the diagnostic results are calculated by a fault diagnosis function, and the fault diagnosis function is constructed based on the performance indicators and fault modes of key components of the equipment; The warning level is calculated by a fault warning function, and the fault warning function is constructed by the equipment performance degradation rate and a preset safety threshold; A feature extraction module is used to input the device state parameters and fault feature parameters of each set of historical operation data into the feature extraction network to obtain device state features and fault features; Input the equipment state characteristics into the diagnosis decision network to obtain the fault diagnosis probability; input the equipment state characteristics and the fault characteristics into the diagnosis result evaluation network respectively to obtain the first diagnosis evaluation value and the second diagnosis evaluation value; input the equipment state characteristics and the fault characteristics into the warning level evaluation network respectively to obtain the first warning evaluation value and the second warning evaluation value; input the diagnosis result, the first diagnosis evaluation value and the second diagnosis evaluation value into the diagnosis advantage evaluation function to obtain the diagnosis advantage evaluation value; input the warning level, the first warning evaluation value and the second warning evaluation value into the warning advantage evaluation function to obtain the warning advantage evaluation value; An optimization training module is used to input the diagnostic advantage evaluation value, the warning advantage evaluation value and the fault diagnosis probability into the optimization objective function and optimize to obtain an optimized diagnostic decision network; input the diagnostic result and the first diagnostic evaluation value into the loss function of the diagnostic result evaluation network and optimize to obtain an optimized diagnostic result evaluation network; input the warning level and the first warning evaluation value into the loss function of the warning level evaluation network and optimize to obtain an optimized warning level evaluation network; based on the optimized diagnostic decision network, the diagnostic result evaluation network and the warning level evaluation network, repeat the above process until the preset number of iterations is exceeded to obtain a trained diagnostic decision network; The diagnosis and warning module is used to obtain the equipment operation data at the current moment and input it into the trained diagnosis decision network to obtain the current diagnostic action for fault diagnosis and warning; execute the current diagnostic action to obtain the equipment operation data at the next moment; and combine the historical diagnosis results and warning levels to predict the future failure trend of the equipment.

2. According to claim 1, a digital mechanical equipment intelligent fault diagnosis and early warning system is characterized in that: The optimization training module is also used to execute the current diagnostic action to obtain the equipment operation data at the next moment; calculate the current diagnostic result according to the fault diagnosis function; calculate the current warning level according to the fault warning function; construct the current operation status data based on the equipment operation data at the current moment, the equipment operation data at the next moment, the current diagnostic action, the current diagnostic result and the current warning level; optimize the diagnosis decision network, the diagnosis result evaluation network and the warning level evaluation network again based on the current operation status data, and use the re-optimized diagnosis decision network to process the equipment operation data at the next moment.

3. According to claim 1, a digital mechanical equipment intelligent fault diagnosis and early warning system is characterized in that: The optimization training module is also used to construct the loss function of the diagnosis result evaluation network and the warning level evaluation network using the mean square error, minimize the loss function through the stochastic gradient descent algorithm, update the parameters of the diagnosis result evaluation network and the parameters of the warning level evaluation network, and obtain the optimized diagnosis result evaluation network and the warning level evaluation network.

4. According to claim 1, a digital mechanical equipment intelligent fault diagnosis and early warning system is characterized in that: The expression of the fault diagnosis function is: in, Indicates the equipment fault diagnosis result. Indicates the performance indicators of key components of the equipment. Indicates the number of key components, Indicates The weight coefficients of the key components, Indicates Failure mode functions of key components.

5. According to claim 1, a digital mechanical equipment intelligent fault diagnosis and early warning system is characterized in that: The expression of the fault warning function is: in, Indicates the equipment failure warning level. Indicates the rate at which device performance decreases. represents the early warning sensitivity coefficient, Indicates the preset safety threshold.

6. According to claim 1, a digital mechanical equipment intelligent fault diagnosis and early warning system is characterized in that: The expression of the diagnostic advantage evaluation function is: in, Indicated in The diagnostic advantage assessment value at the moment, Indicates the stage of strengths assessment, represents the discount factor, Indicates the first stage, Indicated in The fault diagnosis results at the moment, Represents the advantage value parameter, which is used to control the average degree of the advantage value. Indicated in The diagnostic evaluation value corresponding to the device status parameter at the moment.

7. According to claim 1, a digital mechanical equipment intelligent fault diagnosis and early warning system is characterized in that: The expression of the early warning advantage evaluation function is: in, Indicated in The early warning advantage assessment value at each moment, Indicates the stage of strengths assessment, represents the discount factor, Indicates the first stage, Indicated in Fault warning level at all times, Represents the advantage value parameter, which is used to control the average degree of the advantage value. Indicated in The warning assessment value corresponding to the equipment status parameter at the moment.

8. According to claim 1, a digital mechanical equipment intelligent fault diagnosis and early warning system is characterized in that: The expression of the optimization objective function is: in, represents the optimization target value, represents the parameters of the diagnosis decision network, represents the warning weight coefficient, Indicates the decision-making strategy in diagnosis The state distribution under Indicates in status Take action The probability of Indicates in status Take action The diagnostic advantage assessment value of Indicates in status Take action The early warning advantage assessment value.

9. A digital mechanical equipment intelligent fault diagnosis and early warning method, characterized in that: The following steps are involved: Collecting multiple groups of historical operation data, each group of historical operation data includes equipment status parameters, fault characteristic parameters, diagnostic actions, diagnostic results and warning levels; the collection moments corresponding to the equipment status parameters and the fault characteristic parameters are adjacent and sequential moments; the diagnostic results are calculated by a fault diagnosis function, and the fault diagnosis function is constructed based on the performance indicators and fault modes of key components of the equipment; The warning level is calculated by a fault warning function, and the fault warning function is constructed by the equipment performance degradation rate and a preset safety threshold; The equipment state parameters and fault feature parameters of each set of historical operation data are respectively input into the feature extraction network to obtain equipment state features and fault features; Input the equipment state characteristics into the diagnosis decision network to obtain the fault diagnosis probability; input the equipment state characteristics and the fault characteristics into the diagnosis result evaluation network respectively to obtain the first diagnosis evaluation value and the second diagnosis evaluation value; input the equipment state characteristics and the fault characteristics into the warning level evaluation network respectively to obtain the first warning evaluation value and the second warning evaluation value; input the diagnosis result, the first diagnosis evaluation value and the second diagnosis evaluation value into the diagnosis advantage evaluation function to obtain the diagnosis advantage evaluation value; input the warning level, the first warning evaluation value and the second warning evaluation value into the warning advantage evaluation function to obtain the warning advantage evaluation value; Input the diagnostic advantage evaluation value, the warning advantage evaluation value and the fault diagnosis probability into the optimization objective function and optimize to obtain an optimized diagnostic decision network; input the diagnostic result and the first diagnostic evaluation value into the loss function of the diagnostic result evaluation network and optimize to obtain an optimized diagnostic result evaluation network; input the warning level and the first warning evaluation value into the loss function of the warning level evaluation network and optimize to obtain an optimized warning level evaluation network; Based on the optimized diagnosis decision network, diagnosis result evaluation network and warning level evaluation network, the above process is repeated until the preset number of iterations is exceeded to obtain a trained diagnosis decision network; The equipment operation data at the current moment is obtained and input into the trained diagnosis decision network to obtain the current diagnostic action for fault diagnosis and early warning.

10. A digital mechanical equipment intelligent fault diagnosis and early warning method according to claim 9, characterized in that: The following steps are also included: The current diagnosis result is obtained by calculating the fault diagnosis function; the current warning level is obtained by calculating the fault warning function; The current operation status data is constructed based on the equipment operation data at the current moment, the equipment operation data at the next moment, the current diagnostic action, the current diagnostic result and the current warning level; the diagnostic decision network, the diagnostic result evaluation network and the warning level evaluation network are optimized again based on the current operation status data, and the equipment operation data at the next moment is processed using the re-optimized diagnostic decision network.

Images