Remote monitoring method and system for operation state of electromechanical equipment

Abstract

The invention discloses a remote monitoring method and system for the operation state of electromechanical equipment, and the method comprises the steps: building a multi-dimensional parameter monitoring system, obtaining the vibration frequency, temperature and other operation parameters of the electromechanical equipment in real time through a sunflower remote control end, extracting semantic features through an optimized ERNIE model, dividing feature subsets according to the operation stage of the equipment, and carrying out the remote monitoring of the operation state of the electromechanical equipment. And an optimized ERNIE model is combined with a fault feature library to mine potential fault features, a fault recognition result is generated, and a maintenance instruction containing fault positions, types and priorities is generated through a sunflower remote control end according to the fault recognition result. The system comprises a monitoring network construction unit, a data acquisition and transmission unit, a feature extraction unit, a stage division unit, a fault identification unit and an instruction generation and sending unit which are sequentially connected to cooperatively work. Accurate monitoring and intelligent maintenance of the operation state of the electromechanical equipment are realized, and the operation and maintenance efficiency and reliability of the equipment are effectively improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of electromechanical equipment operation monitoring, and in particular to a remote monitoring method and system for the operation state of electromechanical equipment. BACKGROUND

[0002] In the current acceleration of industrial intelligence, electromechanical equipment is widely used in manufacturing, energy, transportation and many other fields, and has become the core infrastructure of production and operation. With the continuous expansion of equipment scale and the increasing complexity of structure, effective monitoring of equipment operation state is crucial to ensure production continuity, reduce maintenance costs and improve safety. Traditional manual inspection and simple sensor monitoring have been unable to meet the real-time and accurate monitoring needs of massive equipment, and remote monitoring solutions based on digital and intelligent technologies have become an inevitable trend in the industry. In this context, how to efficiently collect and analyze electromechanical equipment operation parameters, timely detect potential faults and achieve preventive maintenance has become a key problem to be solved.

[0003] Existing remote monitoring technology for the operation state of electromechanical equipment has obvious defects. Firstly, the data processing and analysis capability is insufficient. Traditional methods often treat equipment operation parameters in isolation, lack deep mining of semantic associations between multi-dimensional parameters such as vibration frequency, temperature and current, and are difficult to accurately infer the equipment operation state from parameter changes. When processing time series data, it is difficult to effectively analyze the dependence of parameters in the time and space dimensions, and when facing complex fault scenarios, it is difficult to capture early abnormal signals in time, resulting in delayed fault diagnosis.

[0004] Secondly, the self-adaptation and intelligent decision-making capability of the remote monitoring system is weak. Existing systems mostly use fixed monitoring modes and diagnostic rules, which are difficult to adapt to the differences in characteristics of different types of electromechanical equipment, as well as dynamic factors such as load changes and changes in working environment during equipment operation. When the equipment exhibits new fault modes or the operating conditions deviate from the norm, the system cannot dynamically adjust the monitoring strategy and fault diagnosis model based on real-time data, and it is also difficult to reasonably develop maintenance plans in combination with maintenance history, spare parts inventory and other information, resulting in low equipment maintenance efficiency and unreasonable resource allocation. SUMMARY

[0005] In order to overcome the shortcomings and deficiencies of the prior art, the present application provides a remote monitoring method and system for the operation state of electromechanical equipment.

[0006] The technical solution adopted by the present application is a remote monitoring method for the operation state of electromechanical equipment, comprising the following steps:

[0007] Step S1: Construct a multi-dimensional parameter monitoring system based on the physical structure of the electromechanical equipment. Classify the electromechanical equipment according to component type, function module, and operation process. Establish corresponding data collection nodes in the multi-dimensional parameter monitoring system. Build connection links between nodes based on equipment operation logic relationship and parameter correlation. Form an initial electromechanical equipment parameter monitoring network.

[0008] Step S2: Establish a data transmission channel with the initial electromechanical equipment parameter monitoring network through the Sunflower remote control terminal. Real-time acquisition of vibration frequency, temperature change, current fluctuation, and speed deviation operation parameters of electromechanical equipment according to preset sampling frequency and sampling accuracy. Form a multi-dimensional time series parameter set.

[0009] Step S3: Use the optimized ERNIE model to extract semantic features from the multi-dimensional time series parameter set. Design a multi-layer semantic encoding structure and combine parameter context association mechanism to deeply analyze the semantic dependency relationship of electromechanical equipment operation parameters in time and space dimensions. Obtain a preliminary feature set containing operation state feature vectors.

[0010] Step S4: Divide the preliminary feature set according to equipment operation stages to form multiple stage feature sub-sets. Each stage feature sub-set corresponds to a different operation stage of the electromechanical equipment.

[0011] Step S5: Use the optimized ERNIE model to identify fault patterns in the stage feature sub-set. Construct a device fault feature library and combine historical fault data and real-time operation parameters of electromechanical equipment to mine and analyze potential fault features in the stage feature sub-set. Obtain fault identification results.

[0012] Step S6: According to the fault identification results, generate equipment maintenance instructions through the Sunflower remote control terminal. The maintenance instructions include fault location, fault type, and maintenance priority information. Send the maintenance instructions to the corresponding maintenance terminal.

[0013] Further, in the optimized ERNIE model, the parameter semantic correlation degree calculation model formula is designed as follows:

[0014]

[0015] Where R s represents the parameter semantic correlation degree; p i and p j represent different electromechanical equipment operation parameters; C(p i , p j ) represents the parameter p i and p jThe context correlation value between the two is calculated by the time series correlation of the parameters and the physical meaning correlation. ij is a weight coefficient, which is determined by the importance and stability of the parameters. The higher the importance and the lower the stability of the parameters, the greater the corresponding weight coefficient.

[0016] Further, in the optimized ERNIE model, a fault feature matching degree evaluation model formula is designed as:

[0017]

[0018] wherein, M f represents the fault feature matching degree; f k represents the standard fault feature in the fault feature library; f t represents the to-be-evaluated feature in the current stage feature subset; S(f k , f t ) represents the feature similarity, which is calculated by the cosine similarity and the Euclidean distance of the feature vectors; α k is a feature weight coefficient, which is determined by the severity and occurrence probability of the fault feature. The higher the severity and the greater the occurrence probability of the fault feature, the greater the corresponding weight coefficient.

[0019] Further, in the step S4, the division of the equipment running stage is dynamically adjusted according to the load change curve and the start-stop state transition point parameters of the electromechanical equipment, and the adjustment formula is:

[0020] T s = T0+ β × ΔL + γ × ΔS

[0021] wherein, T s is the adjusted stage division threshold; T0 is the initial stage division threshold; ΔL is the load change, which is calculated by the difference between the current load and the reference load; ΔS is the start-stop state change value, which is calculated according to the number of transitions and the transition frequency of the equipment start-stop state; β and γ are adjustment coefficients, which are determined by the type and working characteristics of the equipment.

[0022] Further, in the step S5, when the optimized ERNIE model is used for fault mode identification, the standard fault features in the fault feature library are dynamically updated in combination with the energy consumption fluctuation parameters and the noise decibel parameters of the electromechanical equipment, and the update formula is:

[0023]

[0024] wherein, is the updated standard fault feature; is the standard fault feature before updating; E t is the current energy consumption value; Eavg is a historical average energy consumption value; E max and E min are historical maximum and minimum energy consumption values respectively; N t is a current noise decibel value; N avg is a historical average noise decibel value; N max and N min are historical maximum and minimum noise decibel values respectively; δ is an update coefficient determined by the stability and variability parameters of the fault feature; Δf is a feature update step length.

[0025] Further, in the step S6, when generating the equipment maintenance instruction, the repair history parameter and the spare parts inventory parameter of the electromechanical equipment are introduced to construct a maintenance priority evaluation model formula as:

[0026]

[0027] wherein, P m is a maintenance priority; R h is the number of repair history of the equipment; R max is a preset maximum repair number; S c is a spare parts inventory quantity; S max is a preset maximum spare parts inventory quantity; ω1 and ω2 are weight coefficients determined by the importance and fault influence range parameters of the equipment.

[0028] Further, in the step S3, when performing semantic feature extraction by using the optimized ERNIE model, a differentiated attention allocation mechanism is designed for different types of electromechanical equipment operation parameters, and the parameters are dynamically adjusted according to the fluctuation amplitude and mutation frequency parameters of the data, and the adjustment formula is:

[0029]

[0030] wherein, A p is an adjusted attention weight; A0 is an initial attention weight; σ p is a fluctuation standard deviation of the parameter; σ max is a preset maximum fluctuation standard deviation; F p is a mutation frequency of the parameter; F max is a preset maximum mutation frequency; λ is an adjustment coefficient determined by the importance and sensitivity parameters of the parameter; ΔA is an attention adjustment step length.

[0031] Further, in the step S2, when obtaining the electromechanical equipment operation parameters through the sunflower remote control terminal, the sampling frequency is dynamically optimized in combination with the geographical position parameter and the network delay parameter of the equipment, and the optimization formula is:

[0032]

[0033] wherein, F s is the optimized sampling frequency; F0 is the initial sampling frequency; D g is the geographical distance of the device; D max is the preset maximum geographical distance; L n is the network delay time; L max is the preset maximum network delay time; η and θ are optimization coefficients determined by the type of the device and the data importance parameter.

[0034] Further, in the step S5, when the optimized ERNIE model is used to identify the fault mode of the phase feature subset, the vibration spectrum parameter and the temperature gradient parameter of the electromechanical device are combined to evaluate the credibility of the fault identification result, and the evaluation formula is:

[0035]

[0036] wherein, C r is the credibility of the fault identification result; S v is the feature matching degree of the vibration spectrum; S max is the preset maximum vibration spectrum feature matching degree; G t is the abnormality degree of the temperature gradient; G max is the preset maximum temperature gradient abnormality degree; ρ1 and ρ2 are weight coefficients determined by the fault type and the device working environment parameter.

[0037] A remote monitoring system for the running state of electromechanical equipment, the system comprising:

[0038] An electromechanical equipment multi-dimensional parameter monitoring network construction unit for constructing a multi-dimensional parameter monitoring system based on the physical structure of the electromechanical equipment;

[0039] A remote data acquisition and transmission unit connected to the electromechanical equipment multi-dimensional parameter monitoring network construction unit, for establishing a data transmission channel with the initial electromechanical equipment parameter monitoring network through a sunflower remote control terminal;

[0040] A running parameter semantic feature extraction unit connected to the remote data acquisition and transmission unit, for extracting semantic features of the multi-dimensional time series parameter set using an optimized ERNIE model;

[0041] A running phase feature division unit connected to the running parameter semantic feature extraction unit, for dividing the preliminary feature set according to the running phase of the device;

[0042] A fault mode identification unit connected to the running phase feature division unit, for identifying the fault mode of the phase feature subset using an optimized ERNIE model;

[0043] The maintenance instruction generation and sending unit is connected with the fault mode identification unit, and is used for generating a device maintenance instruction through the sunflower remote control terminal according to the fault identification result and sending the maintenance instruction to a corresponding maintenance terminal.

[0044] Beneficial effects: the present application proposes a remote monitoring method and system for the running state of electromechanical equipment. In data processing, the electromechanical equipment is classified according to parts and functions, and a connection link between parameters is established, and the sunflower remote control terminal is used to collect multi-dimensional time sequence parameters such as vibration frequency and temperature in real time. The optimized ERNIE model is used to design a multi-layer semantic coding structure and a parameter context association mechanism. Through a parameter semantic correlation degree calculation model, the semantic dependency relationship of parameters in the time and space dimensions is deeply mined, the disadvantages of traditional isolated parameter processing are changed, and the running state of the equipment is accurately described. In intelligent decision-making and self-adaptation, the optimized ERNIE model combines the equipment fault feature library, historical fault data and real-time parameters, identifies the fault mode through a fault feature matching degree evaluation model, dynamically divides the equipment running stage according to parameters such as load change and start-stop state, and updates the fault feature library by using parameters such as energy consumption and noise, so that the system adapts to the dynamic change of the equipment. At the same time, a maintenance priority evaluation model is constructed by combining parameters such as maintenance history and spare parts inventory, a maintenance instruction containing the fault position, type and priority is generated, and the maintenance instruction is sent to the maintenance terminal through the sunflower remote control terminal, so as to avoid the problems of low maintenance efficiency and unreasonable resource allocation in the traditional fixed mode. The whole system cooperates, from data acquisition, feature extraction to fault diagnosis and maintenance instruction generation, forms a complete closed loop, and significantly improves the intelligent level and operation and maintenance efficiency of the remote monitoring of electromechanical equipment. BRIEF DESCRIPTION OF DRAWINGS

[0045] Figure 1 The method flowchart of the present application;

[0046] Figure 2 The system unit composition diagram of the present application. DETAILED DESCRIPTION

[0047] It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict, and the present application will be further described in detail below in combination with the drawings and specific embodiments.

[0048] As Figure 1 shown, a remote monitoring method for the running state of electromechanical equipment includes the following steps:

[0049] Step S1: constructing a multi-dimensional parameter monitoring system based on the physical structure of the electromechanical equipment, classifying and identifying the electromechanical equipment according to component types, function modules, and operation processes, establishing corresponding data collection nodes in the multi-dimensional parameter monitoring system, and constructing connection links between the nodes according to the equipment operation logic relationship and parameter correlation relationship to form an initial electromechanical equipment parameter monitoring network;

[0050] Specifically, the core of step S1 is to construct a multi-dimensional parameter monitoring system based on the physical structure of the electromechanical equipment. In this process, the electromechanical equipment is first classified and identified according to component types (such as transmission components, power components, and control components), function modules (power transmission modules and signal processing modules), and operation processes (start-up, operation, and shutdown stages). Subsequently, corresponding data collection nodes are established for each type of equipment, module, or process stage in the constructed multi-dimensional parameter monitoring system. The connection links between the nodes are constructed according to the equipment operation logic relationship (such as power transmission sequence and signal interaction path) and the parameter correlation relationship (such as the mutual influence of temperature change and current fluctuation), and finally an initial electromechanical equipment parameter monitoring network is formed, which comprehensively covers all key links and parameter correlations in equipment operation.

[0051] This step builds a structured data foundation framework for electromechanical equipment operation state monitoring, enabling scattered and isolated equipment parameters to be connected through a network, facilitating subsequent systematic data collection and analysis. In terms of implementation, a comprehensive physical structure and function analysis of the electromechanical equipment is first required, and professional equipment documents, design drawings, and other materials are used to determine classification standards and parameter monitoring points. Data collection devices are deployed at the corresponding nodes through sensors, monitoring instruments, and other equipment, and network communication technologies (such as industrial Ethernet and field bus) are used to construct connection links between the nodes, ensuring that data can be accurately and timely transmitted to the monitoring system, and the initial electromechanical equipment parameter monitoring network is built.

[0052] Step S2: establishing a data transmission channel with the initial electromechanical equipment parameter monitoring network through the Sunflower remote control terminal, and real-time acquiring the vibration frequency, temperature change, current fluctuation, and speed deviation operation parameters of the electromechanical equipment according to the preset sampling frequency and sampling accuracy to form a multi-dimensional time series parameter set;

[0053] Specifically, step S2 establishes a data transmission channel between the initial electromechanical equipment parameter monitoring network and the monitoring center by means of the sun remote control terminal. In order to ensure the effectiveness and accuracy of the obtained data, the operating parameters of the electromechanical equipment such as vibration frequency, temperature change, current fluctuation, and speed deviation are collected in real time according to the preset sampling frequency and sampling accuracy. The sampling frequency is set according to the operating characteristics of the equipment, and a higher sampling frequency is used for equipment that operates at high speed and has fast parameter changes. The sampling accuracy is determined according to the parameter monitoring requirements to ensure that subtle parameter changes can be captured, thereby forming a multi-dimensional time sequence parameter set that records the dynamic changes of the parameters during equipment operation.

[0054] This step realizes remote and real-time collection of electromechanical equipment operating parameters, breaks through geographical restrictions, and enables monitoring personnel to obtain equipment operating data without needing to be present on site. In implementation, first install the sun remote control terminal adapter module on the electromechanical equipment side and integrate it with the data collection device. Deploy the corresponding control terminal software in the monitoring center and establish stable connection between the two ends through network configuration. According to the type of equipment and operating requirements, accurately set the sampling frequency and sampling accuracy parameters in the control terminal software. After starting the data collection program, the sun remote control terminal periodically obtains operating parameters from the data collection node according to the set parameters and sends the data to the monitoring center through the network transmission channel, storing a multi-dimensional time sequence parameter set to provide raw data support for subsequent data analysis.

[0055] Step S3: using an optimized ERNIE model to extract semantic features from the multi-dimensional time sequence parameter set, and through the design of a multi-layer semantic coding structure, combining a parameter context association mechanism to deeply analyze the semantic dependency relationship of the electromechanical equipment operating parameters in the time dimension and the space dimension, obtaining a preliminary feature set containing operating state feature vectors;

[0056] Specifically, step S3 uses an optimized ERNIE model to extract semantic features from the multi-dimensional time sequence parameter set. The model designs a multi-layer semantic coding structure, combining a parameter context association mechanism to deeply analyze the semantic dependency relationship of the electromechanical equipment operating parameters in the time dimension (the trend of parameter change over time) and the space dimension (the relationship between different parameters). Through coding, decoding, and feature mapping operations on the parameters, the hidden equipment operating state information behind the parameters is mined, and the original multi-dimensional time sequence parameters are converted into a preliminary feature set containing operating state feature vectors. These feature vectors can more intuitively and accurately reflect the essential features of the equipment operating state.

[0057] This step converts the original, complex device operating parameters into feature vectors with semantic information, providing more valuable data forms for subsequent fault diagnosis and state analysis. During implementation, the optimized ERNIE model needs to be configured and trained first. The historical device operating data and known device operating state labels are used as training samples to adjust the network weights and parameters of the model to adapt to the characteristics of the mechanical and electrical equipment operating parameters. The multi-dimensional time series parameter set is input into the trained model, which is processed through multiple layers of semantic encoding structure, analyzes the semantic dependence between parameters using the parameter context association mechanism, and finally outputs a preliminary feature set containing operating state feature vectors, completing the conversion from raw data to feature data.

[0058] Step S4: dividing the preliminary feature set according to the device operating phase to form multiple phase feature sub-sets, each phase feature sub-set corresponding to a different operating phase of the mechanical and electrical equipment;

[0059] Specifically, step S4 divides the preliminary feature set according to the device operating phase to form multiple phase feature sub-sets. The device operating phase division considers parameters such as the load change curve (reflecting the change of device working intensity), start-stop state transition point (key time node of device start and stop), etc. Through analysis of these parameters, different phases of device operation are determined, such as start-up phase, stable operation phase, load adjustment phase, shutdown phase, etc. Each phase feature sub-set corresponds to a specific operating phase of the device, so that the feature vectors in the same sub-set have similar operating state background and parameter change rules.

[0060] This step further classifies and organizes the preliminary feature set to facilitate fine-grained analysis of the device state for different operating phases. During implementation, the device operating time, load, start-stop state, etc. information corresponding to the feature vectors in the preliminary feature set are extracted and analyzed. According to the pre-set phase division rules (such as dividing the operating phase according to the load threshold, determining the start-stop state transition according to the start-stop time interval), the feature vectors are assigned to the corresponding phase feature sub-sets. During the division process, the division rules and parameters can be dynamically adjusted according to the actual operation of the device to ensure that the phase division accurately reflects the change of the device operating state, providing more targeted data set for subsequent fault pattern recognition.

[0061] Step S5: using the optimized ERNIE model to perform fault pattern recognition on the phase feature sub-sets, constructing a device fault feature library, combining the historical fault data and real-time operating parameters of the mechanical and electrical equipment, and mining and analyzing the potential fault features in the phase feature sub-sets to obtain the fault recognition result;

[0062] Specifically, step S5 uses the optimized ERNIE model to identify the fault mode of the phase feature subset. By constructing a device fault feature library, collecting and organizing historical fault data of the device and corresponding operating parameter features, standard fault feature samples are formed. In the identification process, combined with the historical fault data and real-time operating parameters of the mechanical and electrical equipment, the feature vectors in the phase feature subset are compared and analyzed with the standard fault features in the fault feature library, the deep learning capability of the model is used to mine the potential fault features, and it is judged whether the current operating state of the equipment exists fault and the possible fault type, so as to obtain the fault identification result.

[0063] This step realizes the fault diagnosis of the operating state of the mechanical and electrical equipment, and discovers the potential fault of the equipment in time, providing decision basis for equipment maintenance. When implemented, first, the device fault feature library needs to be continuously enriched and improved, and through analysis of a large number of historical fault cases, representative fault features are extracted and standardized. The optimized ERNIE model is integrated with the fault feature library, and the fault identification parameters and matching rules of the model are set. The phase feature subset is input into the model, and the model calculates the similarity and matching degree of the feature vector and the standard fault feature, combines the statistical law of the historical fault data, judges whether the equipment exists fault and the fault type, and outputs detailed fault identification result, including fault possible position, fault severity estimation and other information.

[0064] Step S6: According to the fault identification result, generate device maintenance instruction through sunflower remote control terminal, the maintenance instruction contains fault position, fault type, maintenance priority information, and send the maintenance instruction to the corresponding maintenance terminal.

[0065] Specifically, step S6 generates device maintenance instruction through sunflower remote control terminal according to fault identification result. The maintenance instruction contains fault position (clearly indicates the specific components or modules where the equipment fault occurs), fault type (such as mechanical fault, electrical fault, etc.), maintenance priority (determined according to fault severity and influence on equipment operation), and other key information. The generation of these information is based on the fault identification result and the pre-set maintenance strategy rules, which ensures that the maintenance instruction can accurately guide the maintenance personnel to carry out equipment repair and maintenance work.

[0066] This step converts the fault diagnosis result into actual equipment maintenance action to ensure that the equipment resumes normal operation as soon as possible. During implementation, on the system platform of the monitoring center, a maintenance instruction generation program is automatically triggered according to the fault identification result. The program integrates and formats information such as fault location, type, priority, etc. according to the preset instruction template and rules to generate standardized equipment maintenance instructions. Through the communication function of the Sunflower remote control terminal, the maintenance instructions are sent to the corresponding maintenance terminal (such as the maintenance personnel's handheld terminal, the management system of the repair station, etc.). After receiving the instructions, the maintenance personnel can quickly prepare repair tools and spare parts according to the instruction content and go to the equipment site for targeted maintenance operation, realizing the rapid response and processing of mechanical and electrical equipment faults.

[0067] Preferably, in the optimized ERNIE model, the design parameter semantic correlation degree calculation model formula is:

[0068]

[0069] wherein R s represents the parameter semantic correlation degree; p i and p j represent different mechanical and electrical equipment operation parameters; C(p i , p j ) represents the context correlation value between parameters p i and p j , which is calculated through the time series correlation and physical meaning correlation of the parameters; w ij is a weight coefficient determined by the importance and stability properties of the parameters, and the higher the importance and the lower the stability of the parameters, the greater the corresponding weight coefficient.

[0070] Specifically, in the optimized ERNIE model, the parameter semantic correlation degree calculation mechanism comprehensively considers factors such as the time series correlation and physical meaning correlation of the mechanical and electrical equipment operation parameters, and realizes accurate semantic feature extraction by assigning dynamic weights to different parameters. The time series correlation reflects the change trend and dependency relationship of the parameters in the time dimension, and the physical meaning correlation reflects the internal physical connection between the parameters. Parameters with high importance and low stability obtain greater weights, so that the model pays more attention to key parameters sensitive to equipment state changes. During implementation, the features of various operation parameters of the mechanical and electrical equipment are extracted first, and their time series characteristics and physical meaning correlations are analyzed to determine the importance and stability indicators of the parameters. According to these indicators, the weight coefficients are dynamically adjusted and integrated into the semantic encoding process of the optimized ERNIE model to realize accurate calculation of the semantic correlation degree of the mechanical and electrical equipment operation parameters and improve the model's understanding ability of the equipment operation state.

[0071] Preferably, in the optimized ERNIE model, the design fault feature matching degree evaluation model formula is:

[0072]

[0073] wherein, M f represents the fault feature matching degree; f k represents the standard fault feature in the fault feature library; f t represents the to-be-evaluated feature in the current stage feature subset; S(f k , f t ) represents the feature similarity, which is calculated by the cosine similarity and the Euclidean distance of the feature vector; a k is the feature weight coefficient, which is determined by the severity and the occurrence probability parameters of the fault feature, and the higher the severity and the greater the occurrence probability, the greater the corresponding weight coefficient.

[0074] Specifically, the fault feature matching degree evaluation mechanism in the optimized ERNIE model, combined with the standard fault features in the mechanical and electrical equipment fault feature library, realizes the accurate identification of potential faults by calculating the feature similarity and giving different feature weight coefficients. The feature similarity calculation considers multiple indicators such as the cosine similarity and the Euclidean distance of the feature vector, and comprehensively measures the similarity between features. Fault features with high severity and high occurrence probability obtain higher weights, so that the model pays more attention to fault types that have greater impact on equipment safety and operation during evaluation. When implemented, first build a standard library containing various fault features, and label and classify each standard fault feature in detail. In the fault identification process, compare the to-be-evaluated feature with the standard fault feature, calculate the similarity, and combine the weight coefficient to obtain the matching degree evaluation result, providing a quantitative basis for fault diagnosis.

[0075] Preferably, in the step S4, the division of the equipment operation stage is dynamically adjusted according to the load change curve of the mechanical and electrical equipment and the start-stop state transition point parameters, and the adjustment formula is:

[0076] T s = T0+ β × ΔL + γ × ΔS

[0077] wherein, T s is the adjusted stage division threshold; T0 is the initial stage division threshold; ΔL is the load change, which is calculated by the difference between the current load and the reference load; ΔS is the start-stop state change value, which is calculated according to the number of transitions and the transition frequency of the equipment start-stop state; β and γ are adjustment coefficients, which are determined by the type and working characteristics parameters of the equipment.

[0078] Specifically, in the device operation phase division process, the time window length adaptive adjustment mechanism dynamically optimizes the time window according to the historical flow characteristics of the electromechanical equipment. This mechanism considers the threat history frequency and network flow fluctuation. When the historical frequency is high and the flow fluctuation is large, the time window is shortened to capture the rapidly changing threat characteristics more timely; otherwise, the time window is increased to ensure sufficient threat information coverage. Through this dynamic adjustment, the time window is matched with the rhythm of the change of the device running state, and the rationality and effectiveness of the feature sequence division are improved. When implemented, the running parameters of the electromechanical equipment are continuously monitored, and the characteristic parameters such as the peak value, valley value and change standard deviation of the historical flow are counted. According to these parameters, the length of the time window is dynamically calculated and adjusted, ensuring the integrity of the data while improving the response speed of the system to the change of the device running state.

[0079] Preferably, in step S5, when the optimized ERNIE model is used for fault mode identification, the standard fault features in the fault feature library are dynamically updated in combination with the energy consumption fluctuation parameters and noise decibel parameters of the electromechanical equipment, and the updating formula is:

[0080]

[0081] wherein, is the updated standard fault feature; is the standard fault feature before updating; E t is the current energy consumption value; E avg is the historical average energy consumption value; E max and E min are the historical maximum and minimum energy consumption values, respectively; N t is the current noise decibel value; N avg is the historical average noise decibel value; N max and N min are the historical maximum and minimum noise decibel values, respectively; δ is the updating coefficient, which is determined by the stability and variability parameters of the fault feature; and Δf is the feature updating step.

[0082] Specifically, the optimized ERNIE model dynamically updates the fault feature library in the fault mode identification by combining the energy fluctuation parameters and noise decibel parameters of the electromechanical equipment. The energy fluctuation parameters reflect the energy consumption change in the equipment operation process, and the noise decibel parameters reflect the acoustic characteristics of the equipment in operation. Both of them can reflect the running state of the equipment from different angles. By incorporating these parameters into the updating process of the fault feature library, the standard fault features can timely reflect the latest running characteristics and fault modes of the equipment. In implementation, the energy consumption and noise data of the electromechanical equipment are collected in real time, and the correlation between them and the equipment running state is analyzed. When abnormal fluctuations in energy consumption or noise are found, the corresponding fault features are evaluated and updated to ensure the timeliness and accuracy of the fault feature library and improve the model's ability to identify new faults.

[0083] Preferably, in the step S6, when generating the equipment maintenance instruction, the maintenance history parameters and spare parts inventory parameters of the electromechanical equipment are introduced to construct a maintenance priority evaluation model formula as follows:

[0084]

[0085] wherein, P m is the maintenance priority; R h is the number of maintenance history of the equipment; R max is the preset maximum number of maintenance; S c is the number of spare parts inventory; S max is the preset maximum number of spare parts inventory; and ω1 and ω2 are weight coefficients determined by the importance and fault influence range parameters of the equipment.

[0086] Specifically, in generating the equipment maintenance instruction, the maintenance priority evaluation mechanism introduces the maintenance history parameters and spare parts inventory parameters of the electromechanical equipment, comprehensively considers the maintenance frequency and spare parts availability of the equipment, and realizes the reasonable allocation of maintenance resources. The maintenance history parameters record the past maintenance of the equipment, reflecting the reliability and vulnerability of the equipment; the spare parts inventory parameters reflect the reserve situation of the spare parts required for maintenance, affecting the timeliness of maintenance. Through the comprehensive evaluation of these two parameters, the maintenance priority is determined, and the maintenance resources are preferentially allocated to the equipment that needs urgent maintenance and has sufficient spare parts. In implementation, the equipment maintenance history database and spare parts inventory management system are established to obtain the maintenance records and spare parts inventory information of the equipment in real time. When generating the maintenance instruction, the maintenance priority is calculated according to the preset evaluation rules combined with these two parameters to ensure that the maintenance work is efficient and orderly.

[0087] Preferably, in the step S3, when performing semantic feature extraction using the optimized ERNIE model, a differentiated attention allocation mechanism is designed for different types of electromechanical equipment operation parameters, and the parameters are dynamically adjusted according to the fluctuation amplitude and mutation frequency parameters of the data, with the adjustment formula as follows:

[0088]

[0089] wherein, A p is the adjusted attention weight; A0 is the initial attention weight; σ p is the fluctuation standard deviation of the parameter; σ max is the preset maximum fluctuation standard deviation; F p is the mutation frequency of the parameter; F max is the preset maximum mutation frequency; λ is the adjustment coefficient, determined by the importance and sensitivity parameters of the parameter; ΔA is the attention adjustment step.

[0090] Specifically, in the process of semantic feature extraction by the optimized ERNIE model, the differentiated attention allocation mechanism dynamically adjusts the attention weight according to the fluctuation amplitude and mutation frequency of different types of mechanical and electrical equipment operation parameters. Parameters with large fluctuation amplitude and high mutation frequency usually reflect the changes of the equipment operation state more, so they are allocated higher attention weight, so that the model pays more attention to the changes of these key parameters. In implementation, real-time monitoring and analysis are first performed on various types of operation parameters of the mechanical and electrical equipment to calculate their fluctuation amplitude and mutation frequency. According to these parameter values, the attention weight allocation of the model is dynamically adjusted, and more computing resources and attention are concentrated on the key parameters, thereby improving the sensitivity and recognition accuracy of the model to the changes of the equipment operation state.

[0091] Preferably, in the step S2, when the mechanical and electrical equipment operation parameters are acquired through the sunflower remote control terminal, the sampling frequency is dynamically optimized in combination with the geographical position parameter and the network delay parameter of the equipment, and the optimization formula is:

[0092]

[0093] wherein, F s is the optimized sampling frequency; F0 is the initial sampling frequency; D g is the geographical position distance of the equipment; D max is the preset maximum geographical position distance; L n is the network delay time; L max is the preset maximum network delay time; η and θ are optimization coefficients, determined by the type of the equipment and the data importance parameter.

[0094] Specifically, when obtaining the operating parameters of electromechanical equipment through the Sunflower remote control terminal, the dynamic optimization mechanism of the sampling frequency combines the device's geographic location parameters and network delay parameters to achieve intelligent adjustment of the sampling frequency. Geographic location parameters affect the distance and difficulty of data transmission, while network delay parameters reflect the real-time nature of data transmission. For devices with long distances and high network delays, the sampling frequency should be appropriately reduced to ensure the stability of data transmission; conversely, the sampling frequency should be increased to obtain more detailed equipment operation information. During implementation, the device's geographic location and network connection status are monitored in real time, and the network delay time is calculated. According to the preset optimization rules, the sampling frequency is dynamically adjusted in combination with the geographic location and network delay parameters to reduce the system's communication burden and data processing pressure while ensuring data quality.

[0095] Preferably, in step S5, when the optimized ERNIE model is used to perform fault mode recognition on the stage feature subset, the reliability of the fault recognition result is evaluated in combination with the vibration spectrum parameters and temperature gradient parameters of the electromechanical equipment. The evaluation formula is:

[0096]

[0097] Among them, C r is the reliability of the fault identification result; S v is the characteristic matching degree of the vibration spectrum; S max is the preset maximum vibration spectrum feature matching degree; G t is the abnormal degree of temperature gradient; G max is the preset maximum temperature gradient abnormality; ρ1 and ρ2 are weight coefficients, which are determined by the fault type and equipment working environment parameters.

[0098] Specifically, when using the optimized ERNIE model for fault pattern identification, the credibility assessment mechanism combines the vibration spectrum parameters and temperature gradient parameters of the electromechanical equipment to comprehensively evaluate the fault identification results. Vibration spectrum parameters reflect the operating status of the equipment's mechanical components, while temperature gradient parameters reflect the thermal distribution of various parts of the equipment. Both are closely related to equipment failures. These two parameters are used to verify and supplement the fault identification results, improving the accuracy and reliability of fault diagnosis. During implementation, the equipment's vibration spectrum and temperature gradient data are simultaneously collected during the fault identification process. This data is compared and analyzed with the fault identification results, and the credibility of the fault identification results is calculated. Based on the credibility assessment results, the fault diagnosis is further confirmed or revised to ensure the accuracy of the diagnostic results and provide a reliable basis for equipment maintenance.

[0099] like Figure 2 As shown, a remote monitoring system for the operating status of electromechanical equipment includes:

[0100] The electromechanical equipment multi-dimensional parameter monitoring network construction unit is used for constructing a multi-dimensional parameter monitoring system based on the physical structure of the electromechanical equipment.

[0101] The remote data acquisition and transmission unit is connected with the electromechanical equipment multi-dimensional parameter monitoring network construction unit, and is used for establishing a data transmission channel with the initial electromechanical equipment parameter monitoring network through the sunflower remote control terminal.

[0102] The running parameter semantic feature extraction unit is connected with the remote data acquisition and transmission unit, and is used for performing semantic feature extraction on the multi-dimensional time sequence parameter set by using an optimized ERNIE model.

[0103] The running phase feature division unit is connected with the running parameter semantic feature extraction unit, and is used for dividing the preliminary feature set according to the equipment running phase.

[0104] The fault mode recognition unit is connected with the running phase feature division unit, and is used for performing fault mode recognition on the phase feature sub-set by using an optimized ERNIE model.

[0105] The maintenance instruction generation and sending unit is connected with the fault mode recognition unit, and is used for generating equipment maintenance instructions through the sunflower remote control terminal according to the fault recognition result and sending the equipment maintenance instructions to a corresponding maintenance terminal.

[0106] The present application provides a kind of electromechanical equipment running state remote monitoring method and system, for the problem of traditional technology to electromechanical equipment running parameter processing extensive, correlation is not enough, the system constructs multi-dimensional parameter monitoring system, with electromechanical equipment physical structure and functional module as foundation, establishes the data acquisition node and connection link covering vibration frequency, temperature, current etc.Parameter.By sunflower remote control terminal, with dynamic optimization sampling frequency, obtain multi-dimensional time sequence parameter, ensure data real-time and integrity.Optimized ERNIE model is with multilayer semantic coding structure and context association mechanism, in combination with parameter semantic correlation degree calculation model, according to parameter importance, stability etc.Attribute distribution weight, depth analysis parameter in time and space dimension semantic dependency, change traditional isolated analysis parameter mode, realize the accurate description of equipment running state and early abnormal signal capture.

[0107] In the face of the poor self-adaptability and the lack of scientific decision-making of the traditional monitoring system, the system utilizes the optimized ERNIE model, combines the equipment fault feature library, historical fault data and real-time operation parameters, and performs fault mode recognition through a fault feature matching degree evaluation model. According to the parameters such as load change and start-stop state, the equipment operation stage is dynamically divided, and the fault feature library is updated in real time according to parameters such as energy consumption and noise, so that the system can adapt to the change of equipment working condition and new faults. In the maintenance link, the maintenance history, spare parts inventory and other parameters are introduced to construct a maintenance priority evaluation model, and the sunflower remote control terminal generates a maintenance instruction containing the fault position, type and priority, so as to realize the reasonable allocation of resources. From data acquisition, feature extraction to fault diagnosis and maintenance decision-making, the whole process cooperates, so that the system has strong self-adaptability and intelligent decision-making level, effectively improves the operation and maintenance efficiency of electromechanical equipment, reduces the fault risk and maintenance cost, and provides strong support for the intelligent upgrading of industrial production.

[0108] In the description of the present application, it should be pointed out that, unless otherwise specified and limited, the terms "arrangement", "installation", "connection", "connection", "fixing" should be understood broadly, for example, it can be fixed connection, or detachable connection, or integral connection; it can be mechanical connection, or electrical connection; it can be direct connection, or indirect connection through intermediate medium, or internal communication of two elements. For ordinary skilled in the art, the specific meaning of the above terms in the present application can be understood through specific circumstances.

[0109] Although the embodiments of the present application have been shown and described, it can be understood by those skilled in the art that various equivalent changes, modifications, replacements and variations of the embodiments can be made without departing from the principles and spirits of the present application, and the scope of the present application is defined by the appended claims and their equivalent scope.

Claims

1. A remote monitoring method for the operating status of electromechanical equipment, characterized in that: The following steps are involved: Step S1: constructing a multi-dimensional parameter monitoring system based on the physical structure of the electromechanical equipment, classifying and identifying the electromechanical equipment according to component type, functional module, and operation process, establishing corresponding data acquisition nodes in the multi-dimensional parameter monitoring system, and building connection links between the nodes based on the logical relationship between the equipment operation and the parameter association relationship to form an initial electromechanical equipment parameter monitoring network; Step S2: Establishing a data transmission channel with the initial electromechanical equipment parameter monitoring network through the Sunflower remote control terminal, obtaining the vibration frequency, temperature change, current fluctuation, and speed deviation operating parameters of the electromechanical equipment in real time according to the preset sampling frequency and sampling accuracy, and forming a multi-dimensional time series parameter set; Step S3: Using the optimized ERNIE model to extract semantic features from the multi-dimensional time series parameter set, by designing a multi-layer semantic encoding structure and combining the parameter context association mechanism to deeply analyze the semantic dependency relationship of the electromechanical equipment operating parameters in the time dimension and the spatial dimension, a preliminary feature set including the operating state feature vector is obtained; Step S4: Dividing the preliminary feature set according to the equipment operation stage to form multiple stage feature subsets, each stage feature subset corresponding to a different operation stage of the electromechanical equipment; Step S5: Using the optimized ERNIE model to perform fault pattern recognition on the stage feature subset, by building an equipment fault feature library and combining historical fault data and real-time operating parameters of electromechanical equipment, mining and analyzing potential fault features in the stage feature subset to obtain a fault recognition result; Step S6: Based on the fault identification result, a device maintenance instruction is generated through the Sunflower remote control terminal, the maintenance instruction including the fault location, fault type, and maintenance priority information, and the maintenance instruction is sent to the corresponding maintenance terminal.

2. A remote monitoring method for the operating status of electromechanical equipment according to claim 1, characterized in that: The optimized ERNIE model and the calculation model formula of the design parameter semantic relevance are as follows: Among them, R s Indicates the semantic association of parameters; p i and p j Respectively represent different electromechanical equipment operating parameters; C(p i , p j ) represents the parameter p i and p j The contextual correlation value between them is calculated by the time series correlation and physical meaning correlation of the parameters; ij is the weight coefficient, which is determined by the importance and stability of the parameter. The higher the importance and the lower the stability of the parameter, the larger the corresponding weight coefficient.

3. A remote monitoring method for the operating status of electromechanical equipment according to claim 1, characterized in that: The optimized ERNIE model designs a fault feature matching evaluation model formula as follows: Among them, M f represents the fault feature matching degree; f k Represents the standard fault signature in the fault signature library; f t Indicates the features to be evaluated in the feature subset of the current stage; S(f k , f t ) represents the feature similarity, which is calculated by the cosine similarity and Euclidean distance of the feature vector; α k is the feature weight coefficient, which is determined by the severity and occurrence probability parameters of the fault feature. The higher the severity and the greater the probability of occurrence of the fault feature, the greater the corresponding weight coefficient.

4. A remote monitoring method for the operating status of electromechanical equipment according to claim 1, characterized in that: In step S4, the division of the equipment operation phase is dynamically adjusted based on the load change curve of the electromechanical equipment and the start-stop state transition point parameters, and the adjustment formula is: T s =T0+β×ΔL+γ×ΔS Among them, T s is the adjusted stage division threshold; T0 is the initial stage division threshold; ΔL is the load change, which is calculated by the difference between the current load and the reference load; ΔS is the start-stop state change value, which is calculated based on the number and frequency of start-stop state transitions of the equipment; β and γ are adjustment coefficients, which are determined by the type of equipment and operating characteristic parameters.

5. The remote monitoring method for the operating status of electromechanical equipment according to claim 1, characterized in that: In step S5, when the optimized ERNIE model is used for fault pattern recognition, the standard fault features in the fault feature library are dynamically updated in combination with the energy consumption fluctuation parameters and noise decibel parameters of the electromechanical equipment. The update formula is: in, It is the updated standard fault signature; is the standard fault feature before the update; E t is the current energy consumption value; E avg is the historical average energy consumption value; E max and E min are the historical maximum and minimum energy consumption values ​​respectively; N t is the current noise decibel value; N avg is the historical average noise decibel value; N max and N min are the historical maximum and minimum noise decibel values ​​respectively; δ is the update coefficient, which is determined by the stability and variability parameters of the fault feature; Δf is the feature update step size.

6. A remote monitoring method for the operating status of electromechanical equipment according to claim 1, characterized in that: In step S6, when generating the equipment maintenance instruction, the maintenance history parameters and spare parts inventory parameters of the electromechanical equipment are introduced to construct the maintenance priority evaluation model formula: Among them, P m To maintain priority; R h is the maintenance history of the equipment; R max is the preset maximum number of maintenance times; S c is the spare parts inventory quantity; S max is the preset maximum spare parts inventory quantity; ω1 and ω2 are weight coefficients, which are determined by the importance of the equipment and the fault impact range parameters.

7. A remote monitoring method for the operating status of electromechanical equipment according to claim 1, characterized in that: In step S3, when using the optimized ERNIE model to extract semantic features, a differentiated attention allocation mechanism is designed for different types of electromechanical equipment operating parameters. The parameters are dynamically adjusted according to the fluctuation amplitude and mutation frequency parameters of the data. The adjustment formula is: Among them, A p is the adjusted attention weight; A0 is the initial attention weight; σ p is the standard deviation of the parameter fluctuation; σ max is the preset maximum fluctuation standard deviation; F p is the mutation frequency of the parameter; F max is the preset maximum mutation frequency; λ is the adjustment coefficient, which is determined by the importance and sensitivity parameters of the parameter; ΔA is the attention adjustment step.

8. A remote monitoring method for the operating status of electromechanical equipment according to claim 1, characterized in that: In step S2, when obtaining the operating parameters of the electromechanical equipment through the Sunflower remote control terminal, the sampling frequency is dynamically optimized in combination with the geographical location parameters and network delay parameters of the equipment. The optimization formula is: Among them, F s is the optimized sampling frequency; F0 is the initial sampling frequency; D g D is the geographical distance of the device; max is the preset maximum geographical distance; L n is the network delay time; L max is the preset maximum network delay time; η and θ are optimization coefficients, which are determined by the device type and data importance parameters.

9. A remote monitoring method for the operating status of electromechanical equipment according to claim 1, characterized in that: In step S5, when the optimized ERNIE model is used to perform fault pattern recognition on the stage feature subset, the reliability of the fault recognition result is evaluated in combination with the vibration spectrum parameters and temperature gradient parameters of the electromechanical equipment. The evaluation formula is: Among them, C r is the reliability of the fault identification result; S v is the characteristic matching degree of the vibration spectrum; S max is the preset maximum vibration spectrum feature matching degree; G t is the abnormal degree of temperature gradient; G max is the preset maximum temperature gradient abnormality; ρ1 and ρ2 are weight coefficients, which are determined by the fault type and equipment working environment parameters.

10. A remote monitoring system for the operating status of electromechanical equipment, characterized in that: The system includes: A multi-dimensional parameter monitoring network construction unit for electromechanical equipment, used to build a multi-dimensional parameter monitoring system based on the physical structure of electromechanical equipment; A remote data acquisition and transmission unit, connected to the electromechanical equipment multi-dimensional parameter monitoring network construction unit, is used to establish a data transmission channel with the initial electromechanical equipment parameter monitoring network through the Sunflower remote control terminal; An operating parameter semantic feature extraction unit is connected to the remote data acquisition and transmission unit and is used to extract semantic features from the multi-dimensional time series parameter set using an optimized ERNIE model; an operation stage feature division unit, connected to the operation parameter semantic feature extraction unit, for dividing the preliminary feature set according to the device operation stage; a fault pattern recognition unit connected to the operation stage feature division unit, and configured to perform fault pattern recognition on the stage feature subset using an optimized ERNIE model; The maintenance instruction generation and sending unit is connected to the fault mode recognition unit and is used to generate equipment maintenance instructions through the Sunflower remote control terminal according to the fault recognition result and send them to the corresponding maintenance terminal.

Images