Electromechanical fault pre-diagnosis alarm device

Abstract

The invention discloses an electrical mechanical fault pre-diagnosis alarm device, and relates to the technical field of electrical mechanical faults, the electrical mechanical fault pre-diagnosis alarm device comprises a shell, the interior of the shell comprises a multi-mode sensing module, and an edge feature extraction module is electrically connected with the multi-mode sensing module through a high-speed data interface; the multi-modal sensing module collects equipment data in a multi-dimensional mode and comprehensively reflects the operation state, the edge feature extraction module effectively reduces noise and extracts rich features through multiple technologies, the mixed diagnosis model module performs deep analysis through an advanced algorithm and improves the fault diagnosis accuracy, and the self-adaptive early warning module dynamically adjusts a strategy, performs early warning in time and reduces loss. The cloud digital twin module achieves equipment health assessment and model optimization, the communication module guarantees smooth data interaction of all the modules, the monitoring end facilitates real-time checking and control of a user, real-time monitoring, fault pre-diagnosis and early warning of electrical mechanical equipment are achieved, the operation reliability and safety of the equipment are effectively improved, and the operation and maintenance cost is reduced.

Description

Technical Field

[0001] This invention relates to the field of electrical machinery fault technology, specifically to an electrical machinery fault pre-diagnosis alarm device. Background Technology

[0002] In the grand scheme of modern industrial production, electromechanical equipment can be considered the core "engine" and intelligent "brain." It integrates electric drive and mechanical operation, efficiently converting electrical energy into mechanical energy and precisely controlling every aspect of the production process. From the roaring production lines in large factories to the delicately operating instruments in precision workshops, electromechanical equipment is ubiquitous. Its stable and reliable operation is the key foundation for ensuring on-time product delivery and quality standards. Once a malfunction occurs, it can not only lead to production stoppages and order delays, but also trigger a chain reaction, damaging other related equipment and causing huge economic losses and reputational damage to the enterprise.

[0003] Many enterprises use fixed-cycle maintenance for their electrical and mechanical systems. This involves comprehensive inspections and maintenance at pre-set intervals, regardless of the equipment's actual operating condition. While this approach can prevent some faults to a certain extent, it also presents several problems. Firstly, fixed-cycle maintenance can easily lead to over-maintenance. Disassembling and replacing parts on equipment in good working order before actual maintenance needs are met not only wastes significant manpower, resources, and money, but may also introduce new potential faults due to unnecessary disassembly and assembly. Secondly, fixed-cycle maintenance is difficult to adapt to changes in the actual operating conditions of the equipment. Therefore, this invention proposes an electrical and mechanical fault pre-diagnosis alarm device. Summary of the Invention

[0004] The purpose of this invention is to provide an electrical machinery fault pre-diagnosis alarm device to address the problem mentioned in the background art where many enterprises maintain electrical machinery systems using a fixed-cycle maintenance approach. This involves performing comprehensive inspections and maintenance on equipment at pre-set time intervals, regardless of the actual operating condition. While this method can prevent some faults to a certain extent, it also presents several problems. Firstly, fixed-cycle maintenance easily leads to over-maintenance. For equipment in good operating condition, disassembly and replacement of parts before actual maintenance needs are met not only waste significant manpower, material resources, and financial resources, but may also introduce new potential faults due to unnecessary disassembly and assembly. Secondly, fixed-cycle maintenance struggles to adapt to changes in the actual operating conditions of the equipment.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] An electromechanical fault pre-diagnosis alarm device includes a housing, characterized in that: the housing contains a multimodal sensing module, which is used to collect multi-dimensional operating data of the electromechanical equipment in real time;

[0007] An edge feature extraction module is electrically connected to a multimodal sensing module via a high-speed data interface. It is used to perform noise reduction and feature extraction on the data collected by the multimodal sensing module.

[0008] The hybrid diagnostic model module is used for in-depth analysis and fault diagnosis of the input feature data and is connected to the edge feature extraction module.

[0009] The adaptive early warning module is connected to the hybrid diagnostic model module through an internal data channel. The adaptive early warning module dynamically adjusts the early warning strategy based on the diagnostic results of the hybrid diagnostic model module and preset rules, and triggers early warning in a timely manner.

[0010] The cloud-based digital twin module and the hybrid diagnostic model module are bidirectionally connected via a communication module, and are used to build a digital twin model that is synchronized with the actual electrical machinery equipment in real time in the cloud.

[0011] The communication module uses a wireless method to realize data interaction between the multimodal sensing module, edge feature extraction module, hybrid diagnostic model module, adaptive early warning module and cloud digital twin module. The communication module transmits the early warning information of the adaptive early warning module to the alarm and monitoring terminal through the signal output interface.

[0012] Optionally, the multimodal sensing module includes a vibration sensor, a broadband current sensor, an infrared temperature sensor, and a partial discharge sensor. The vibration sensor uses an accelerometer to sense equipment vibration, the broadband current sensor uses current transformer technology, the infrared temperature sensor is based on the thermal effect of infrared radiation, and the partial discharge sensor uses electromagnetic coupling and ultrasonic pulse detection technology to detect partial discharge phenomena inside the equipment.

[0013] Optionally, the edge feature extraction module includes a wavelet denoising unit, a time-domain feature extraction unit, a frequency-domain envelope analysis unit, and a nonlinear feature calculation unit. The wavelet denoising unit uses wavelet transform technology to perform multi-scale decomposition and reconstruction of the input data to remove noise. The time-domain feature extraction unit extracts feature parameters such as mean, variance, and peak value from a time-domain perspective. The frequency-domain envelope analysis unit converts the data to the frequency domain through Fourier transform to analyze spectral features. The nonlinear feature calculation unit calculates nonlinear features such as permutation entropy and complexity. It is responsible for denoising and feature extraction of the data collected by the multimodal sensing module. The units interact with each other through an internal bus.

[0014] Optionally, the hybrid diagnostic model module includes a D-CNN layer that uses a convolutional neural network for local feature extraction and classification, an LSTM layer that uses a long short-term memory network to capture time series features, a feature fusion layer that fuses the outputs of the first two layers through weighted averaging and other algorithms to improve feature representation capabilities, and a Bayesian network that performs fault diagnosis based on probabilistic reasoning and outputs fault probabilities for in-depth analysis and fault diagnosis of the input feature data.

[0015] Optionally, the adaptive early warning module dynamically adjusts the early warning threshold and early warning level based on the device's operating status, historical fault data, and preset early warning rules. When the fault probability output by the hybrid diagnostic model module exceeds the set threshold, the adaptive early warning module promptly triggers the corresponding level of early warning, records the early warning information, and sends the early warning information to the alarm and monitoring terminal through the communication module.

[0016] Optionally, the cloud-based digital twin module provides model maintenance suggestions and parameter update instructions to the hybrid diagnostic model module through the communication module.

[0017] Optionally, the monitoring terminal includes a display screen, a mobile phone, and a computer. The monitoring terminal allows users to view the device's operating status, diagnostic results, and early warning information in real time. The monitoring terminal also transmits user commands to the corresponding modules via a communication module.

[0018] Optionally, the housing is mounted on the electrical equipment via a mounting plate, a display screen is mounted on the front side of the housing via a connecting plate, an interface is provided on one side of the housing, and a connecting line for connecting to electrical equipment components is fixedly connected to the interface, and alarms are symmetrically provided on the top left and right sides of the housing.

[0019] The beneficial effects of this invention are:

[0020] This invention utilizes a multimodal sensing module to collect equipment data from multiple dimensions, comprehensively reflecting its operational status. An edge feature extraction module effectively reduces noise and extracts rich features through various techniques, laying the foundation for accurate diagnosis. A hybrid diagnostic model module employs advanced algorithms for deep analysis, improving fault diagnosis accuracy. An adaptive early warning module dynamically adjusts strategies, providing timely warnings and reducing losses. A cloud-based digital twin module enables equipment health assessment and model optimization. A communication module ensures smooth data interaction between modules. The monitoring terminal allows users to conveniently view and control the equipment in real time. This invention achieves real-time monitoring, fault pre-diagnosis, and early warning of electrical and mechanical equipment, effectively improving equipment reliability and safety, reducing maintenance costs, and demonstrating high practical value. Attached Figure Description

[0021] Figure 1 This is a schematic block diagram showing the connection of an electrical and mechanical fault pre-diagnosis alarm device according to the present invention;

[0022] Figure 2 This is a system flowchart of an electrical and mechanical fault pre-diagnosis alarm device according to the present invention;

[0023] Figure 3 This is a schematic diagram of the shell structure in this invention.

[0024] The numbers on the map are:

[0025] 1. Housing; 2. Mounting plate; 3. Connecting plate; 4. Display screen; 5. Interface; 6. Connecting cable; 7. Alarm. Detailed Implementation

[0026] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0027] The preferred embodiment of the device of the present invention will be described below.

[0028] Example 1:

[0029] As attached Figure 1 To be continued Figure 3 As shown, the present invention provides an electrical machinery fault pre-diagnosis alarm device, including a housing 1, characterized in that: the housing 1 includes a multi-modal sensing module inside, which is used to collect multi-dimensional operating data of electrical machinery in real time. Each sensor in the module performs preliminary amplification, filtering and other processing on the collected raw analog signals through the built-in signal conditioning circuit, and then transmits the processed signals to the subsequent modules through a wired connection.

[0030] The edge feature extraction module is electrically connected to the multimodal sensing module through a high-speed data interface. It is used to perform noise reduction and feature extraction on the data collected by the multimodal sensing module. The units interact with each other through an internal bus. The wavelet noise reduction unit first performs noise reduction on the input data, and then passes the processed data to the time domain feature extraction unit, the frequency domain envelope analysis unit and the nonlinear feature calculation unit in sequence for feature extraction.

[0031] The hybrid diagnostic model module is used for in-depth analysis and fault diagnosis of the input feature data. It connects with the edge feature extraction module, and the layers within the hybrid diagnostic model module transfer and compute data through the connection weights and activation functions within the neural network. Simultaneously, this module achieves bidirectional data interaction with the cloud-based digital twin module via a specific communication protocol, receiving optimization instructions from the cloud and uploading local diagnostic results.

[0032] The adaptive early warning module is connected to the hybrid diagnostic model module through an internal data channel. The adaptive early warning module dynamically adjusts the early warning strategy based on the diagnostic results of the hybrid diagnostic model module and preset rules, and triggers early warning in a timely manner. The module internally stores and adjusts the early warning strategy through logic circuits and storage units. When the early warning conditions are met, it sends an early warning signal to the communication module through the output interface.

[0033] The cloud-based digital twin module and the hybrid diagnostic model module are bidirectionally connected via a communication module. The cloud-based digital twin model is used to build a digital twin model that is synchronized with the actual electrical machinery in real time. Big data analysis technology is used to deeply mine the uploaded data to achieve equipment health status assessment and remaining service life prediction. The communication module also provides optimization suggestions and update instructions to the hybrid diagnostic model module.

[0034] The communication module uses a wireless method to realize data interaction between the multimodal sensing module, edge feature extraction module, hybrid diagnostic model module, adaptive early warning module and cloud digital twin module. The communication module transmits the early warning information of the adaptive early warning module to the alarm 7 and the monitoring terminal through the signal output interface. Users can set parameters and issue control commands through the input device of the monitoring terminal.

[0035] Example 2:

[0036] As attached Figure 1-2As shown, this embodiment is basically the same as the previous embodiment, except that the multimodal sensing module includes a vibration sensor, a wideband current sensor, an infrared temperature sensor, and a partial discharge sensor. The vibration sensor uses an accelerometer to sense the vibration of the equipment. The accelerometer can sense the vibration acceleration generated by the equipment during operation and convert it into an electrical signal. Since the original analog signal may be relatively weak and mixed with noise, the signal conditioning circuit built into the sensor will perform preliminary amplification and filtering of the electrical signal to enhance the signal strength and remove some high-frequency noise. Then, the processed signal is transmitted to the subsequent module through a wired connection. The wideband current sensor uses current transformer technology. The current transformer can proportionally transform the large current in the electrical mechanical equipment into a small current, which is convenient for measurement and transmission. Similarly, the acquired current signal, after being amplified and filtered by the built-in signal conditioning circuit, is transmitted to subsequent modules via a wired connection to reflect the current operating status of the equipment. The infrared temperature sensor, based on the thermal effect of infrared radiation, utilizes the fact that all objects above absolute zero emit infrared radiation. The sensor detects the infrared radiation energy emitted by the equipment surface and converts it into an electrical signal. The built-in signal conditioning circuit amplifies and linearizes the converted electrical signal, ensuring it accurately reflects the equipment temperature. The data is then transmitted via a wired connection. The partial discharge sensor uses electromagnetic coupling and ultrasonic pulse detection technology to detect partial discharge phenomena inside the equipment. The electromagnetic coupling part detects the electromagnetic signals generated by partial discharge, while the ultrasonic pulse detection part captures the ultrasonic signals caused by partial discharge. The sensor converts these two signals into electrical signals, which are then processed by the built-in signal conditioning circuit and transmitted via a wired connection to subsequent modules for monitoring partial discharge phenomena inside the equipment.

[0037] Furthermore, the edge feature extraction module includes a wavelet denoising unit, a time-domain feature extraction unit, a frequency-domain envelope analysis unit, and a nonlinear feature calculation unit;

[0038] The wavelet denoising unit employs wavelet transform technology to perform multi-scale decomposition and reconstruction of the input data to remove noise. Wavelet transform has the characteristic of multi-scale analysis, enabling the input data to be decomposed into wavelet spaces of different scales. At different scales, noise and useful signals exhibit different characteristics. By selecting an appropriate threshold to process the wavelet coefficients, wavelet coefficients corresponding to noise are removed, while wavelet coefficients corresponding to useful signals are retained or enhanced. Finally, wavelet reconstruction is performed to obtain the denoised signal. The data processed by this unit is transmitted to the time-domain feature extraction unit via an internal bus.

[0039] The time-domain feature extraction unit extracts feature parameters such as mean, variance, and peak value from a time-domain perspective to describe the signal's variation characteristics along the time axis. The mean reflects the average level of the signal, the variance reflects the signal's dispersion, and the peak value represents the signal's maximum amplitude. These feature parameters provide time-domain information for subsequent fault diagnosis. The processed data is transmitted to the frequency-domain envelope analysis unit via the internal bus.

[0040] The frequency domain envelope analysis unit converts the data to frequency domain spectral characteristics using Fourier transform. The spectrum displays the magnitude and phase information of each frequency component in the signal. By processing the spectrum using envelope analysis technology, the envelope of the spectrum is extracted, highlighting the modulation information in the signal, which helps in detecting periodic faults in the equipment. After processing, the data is transmitted to the nonlinear feature calculation unit via the internal bus.

[0041] The nonlinear feature calculation unit calculates nonlinear features such as permutation entropy and complexity, and is responsible for noise reduction and feature extraction of data collected by the multimodal sensing module. The units interact with each other via an internal bus. Permutation entropy measures the complexity and irregularity of a time series, while complexity reflects the dynamic characteristics of the system. These nonlinear features can capture the nonlinear dynamics of equipment operation, providing more comprehensive information for fault diagnosis.

[0042] Furthermore, the hybrid diagnostic model module includes a 1D-CNN layer that uses a convolutional neural network for local feature extraction and classification, an LSTM layer that uses a long short-term memory network to capture time series features, a feature fusion layer that uses algorithms such as weighted averaging to fuse the outputs of the first two layers to improve feature representation, and a Bayesian network that performs fault diagnosis based on probabilistic reasoning and outputs fault probabilities for in-depth analysis and fault diagnosis of the input feature data.

[0043] It is worth further describing that the convolutional neural network performs sliding convolution operations on the input data through convolution kernels to automatically extract local features in the data. The 1D-CNN layer can capture local patterns and feature relationships in the data, providing basic feature information for subsequent fault diagnosis.

[0044] Long Short-Term Memory (LSTM) networks are a special type of recurrent neural network that can effectively handle long-term dependencies in sequential data. In electrical machinery fault diagnosis, equipment operation data often has time-series characteristics, and LSTM layers can better understand the changing trends of equipment operating status.

[0045] The output features of different layers have different characteristics and information. The feature fusion layer integrates these features to improve the feature representation ability and enable the model to understand the data features more comprehensively.

[0046] Bayesian networks construct a probabilistic relationship model between faults and features, dynamically updating the posterior probability of fault occurrence by combining real-time collected feature data. This network comprehensively considers various factors, providing a reliable probabilistic basis for fault diagnosis and enabling in-depth analysis and fault diagnosis of input feature data. Data transfer and computation occur between layers within the module through connection weights and activation functions within the neural network. Simultaneously, bidirectional data interaction is achieved with the cloud-based digital twin module via a specific communication protocol, receiving optimization instructions from the cloud and uploading local diagnostic results.

[0047] Furthermore, the adaptive early warning module dynamically adjusts the early warning threshold and level based on the equipment's operating status, historical fault data, and preset early warning rules. When the fault probability output by the hybrid diagnostic model module exceeds the set threshold, the adaptive early warning module promptly triggers the corresponding level of early warning, records the early warning information, and sends the early warning information to the alarm 7 and the monitoring terminal through the communication module.

[0048] Furthermore, the cloud-based digital twin module provides model maintenance suggestions and parameter update instructions to the hybrid diagnostic model module through the communication module.

[0049] Furthermore, the monitoring terminal includes a display screen 4, a mobile phone, and a computer. Users can view the device's operating status, diagnostic results, and early warning information in real time through the monitoring terminal. The monitoring terminal transmits user commands to the corresponding modules through the communication module.

[0050] Specifically, during the pre-diagnosis process, the vibration sensor, broadband current sensor, infrared temperature sensor and partial discharge sensor in the multimodal sensing module collect data such as vibration, current, temperature and partial discharge of the equipment using corresponding technologies. After being processed by the built-in signal conditioning circuit, the data is transmitted to the edge feature extraction module via a wired connection.

[0051] The wavelet denoising unit of the edge feature extraction module performs denoising processing on the input data, and then passes the processed data sequentially to the time domain feature extraction unit, the frequency domain envelope analysis unit, and the nonlinear feature calculation unit for feature extraction. The units exchange data through the internal bus.

[0052] Then, the 1D-CNN layer of the hybrid diagnostic model module performs local feature extraction and classification, the LSTM layer captures time-series features, the feature fusion layer fuses the outputs of the first two layers, and the Bayesian network performs fault diagnosis based on probabilistic inference and outputs the fault probability. Simultaneously, this module interacts with the cloud-based digital twin module via a communication interface, receiving optimization instructions and uploading diagnostic results.

[0053] The adaptive early warning module dynamically adjusts the early warning threshold and level based on the diagnostic results of the hybrid diagnostic model module, combined with the equipment operating status, historical fault data and preset rules. When the fault probability exceeds the threshold, the corresponding level of early warning is triggered, the early warning information is recorded and sent to the alarm 7 and the monitoring terminal through the communication module.

[0054] Meanwhile, the cloud-based digital twin module uses big data analytics to deeply mine uploaded data, assess the health status of the equipment and predict its remaining lifespan, and provide optimization suggestions and update instructions for the hybrid diagnostic model module.

[0055] Finally, users can view equipment information in real time through the monitor display screen 4, mobile phone and computer, and send parameter settings and control commands to the corresponding modules through the communication module to remotely monitor and control the device. This enables real-time monitoring, fault diagnosis and early warning of electrical and mechanical equipment, and improves the reliability and safety of equipment operation.

[0056] Example 3:

[0057] As attached Figure 3 As shown, this embodiment is basically the same as the previous embodiment, except that the housing 1 is mounted on the electrical equipment through the mounting plate 2, the front side of the housing 1 is equipped with the display screen 4 through the connecting plate 3, the side of the housing 1 is provided with the interface 5, and the interface 5 is fixedly connected with the connecting line 6 connected to the electrical equipment component, and the top of the housing 1 is symmetrically provided with alarms 7.

[0058] The embodiments of the present invention are given for the purposes of illustration and description. Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Any changes, modifications, substitutions and variations made by those skilled in the art to the above embodiments within the scope of the present invention should be included within the protection scope of the present invention.

Claims

1. An electrical and mechanical fault pre-diagnosis alarm device, comprising a housing (1), characterized in that: The interior of the housing (1) includes a multimodal sensing module, which is used to collect multi-dimensional operating data of the electrical mechanical equipment in real time; An edge feature extraction module is electrically connected to a multimodal sensing module via a high-speed data interface. It is used to perform noise reduction and feature extraction on the data collected by the multimodal sensing module. The hybrid diagnostic model module is used for in-depth analysis and fault diagnosis of the input feature data and is connected to the edge feature extraction module. The adaptive early warning module is connected to the hybrid diagnostic model module through an internal data channel. The adaptive early warning module dynamically adjusts the early warning strategy based on the diagnostic results of the hybrid diagnostic model module and preset rules, and triggers early warning in a timely manner. The cloud-based digital twin module and the hybrid diagnostic model module are bidirectionally connected via a communication module, and are used to build a digital twin model that is synchronized with the actual electrical machinery equipment in real time in the cloud. The communication module uses a wireless method to realize data interaction between the multimodal sensing module, edge feature extraction module, hybrid diagnostic model module, adaptive early warning module and cloud digital twin module. The communication module transmits the early warning information of the adaptive early warning module to the alarm (7) and the monitoring terminal through the signal output interface.

2. The electrical machinery fault pre-diagnosis alarm device according to claim 1, characterized in that: The multimodal sensing module includes a vibration sensor, a broadband current sensor, an infrared temperature sensor, and a partial discharge sensor. The vibration sensor uses an accelerometer to sense equipment vibration, the broadband current sensor uses current transformer technology, the infrared temperature sensor is based on the thermal effect of infrared radiation, and the partial discharge sensor uses electromagnetic coupling and ultrasonic pulse detection technology to detect partial discharge phenomena inside the equipment.

3. The electrical machinery fault pre-diagnosis alarm device according to claim 1, characterized in that: The edge feature extraction module includes a wavelet denoising unit, a time-domain feature extraction unit, a frequency-domain envelope analysis unit, and a nonlinear feature calculation unit. The wavelet denoising unit uses wavelet transform technology to decompose and reconstruct the input data at multiple scales to remove noise. The time-domain feature extraction unit extracts feature parameters such as mean, variance, and peak value from the time domain perspective to describe the signal's variation characteristics on the time axis. The frequency-domain envelope analysis unit converts the data to the frequency domain using Fourier transform to analyze the spectral features. The nonlinear feature calculation unit calculates nonlinear features such as permutation entropy and complexity, and is responsible for denoising and feature extraction of the data collected by the multimodal sensing module. The units interact with each other through an internal bus.

4. The electrical and mechanical fault pre-diagnosis alarm device according to claim 1, characterized in that: The hybrid diagnostic model module includes a 1D-CNN layer that uses a convolutional neural network for local feature extraction and classification, an LSTM layer that uses a long short-term memory network to capture time series features, a feature fusion layer that uses weighted averaging and other algorithms to fuse the outputs of the first two layers to improve feature representation, and a Bayesian network that performs fault diagnosis based on probabilistic reasoning and outputs fault probabilities for in-depth analysis and fault diagnosis of the input feature data.

5. The electrical machinery fault pre-diagnosis alarm device according to claim 1, characterized in that: The adaptive early warning module dynamically adjusts the early warning threshold and early warning level according to the operating status of the equipment, historical fault data and preset early warning rules. When the fault probability output by the hybrid diagnostic model module exceeds the set threshold, the adaptive early warning module promptly triggers the corresponding level of early warning and records the early warning information. The early warning information is then sent to the alarm (7) and the monitoring terminal through the communication module.

6. The electrical machinery fault pre-diagnosis alarm device according to claim 1, characterized in that: The cloud-based digital twin module provides model maintenance suggestions and parameter update instructions to the hybrid diagnostic model module through the communication module.

7. The electrical machinery fault pre-diagnosis alarm device according to claim 1, characterized in that: The monitoring terminal includes a display screen (4), a mobile phone and a computer. The monitoring terminal allows users to view the operating status, diagnostic results and early warning information of the equipment in real time. The monitoring terminal transmits user instructions to the corresponding modules through the communication module.

8. The electrical and mechanical fault pre-diagnosis alarm device according to claim 1, characterized in that: The housing (1) is mounted on the electrical equipment via the mounting plate (2). The front side of the housing (1) is fitted with a display screen (4) via a connecting plate (3). An interface (5) is provided on one side of the housing (1), and a connecting line (6) for connecting to electrical equipment components is fixedly connected to the interface (5). Alarms (7) are symmetrically provided on the top left and right sides of the housing (1).

Images