Electromechanical equipment diagnosis method of multi-dimensional operation and maintenance data training model and operation and maintenance equipment

Abstract

The invention discloses an electromechanical equipment diagnosis method and operation and maintenance equipment of a multi-dimensional operation and maintenance data training model, and relates to the technical field of electromechanical equipment operation and maintenance. A multi-dimensional sensor network is deployed for staggered-time cross sampling, electromechanical current data is acquired, a first verification port and a second verification port are introduced, a diagnosis network architecture is constructed, and the diagnosis network architecture is constructed. The method comprises the following steps: performing network access on current data of the machine, performing aggregation screening based on a first verification port and data structure conversion and packaging based on a second verification port, transmitting the data back to a data center, performing operation and maintenance fault decision and trend prediction based on a built-in operation and maintenance diagnostor, and outputting a fault diagnosis list. The method and the device are used for solving the technical problems that in the prior art, an intelligent diagnosis framework is lacked, and the effectiveness and the accuracy of data sampling and diagnosis decision cannot be guaranteed, flexible sampling is carried out with the demand as the guide, a systematic diagnosis framework is constructed, and the intelligence and the accuracy of diagnosis decision of electromechanical equipment are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of electromechanical equipment operation and maintenance, and in particular to an electromechanical equipment diagnosis method and operation and maintenance equipment based on a multi-dimensional operation and maintenance data training model. Background Art

[0002] In the current field of electromechanical equipment operation and maintenance, traditional diagnostic methods rely primarily on manual inspections and empirical judgment. This approach is not only inefficient but also struggles to accurately capture early signs of equipment failure. With the development of the Internet of Things (IoT), multidimensional sensor networks are beginning to be applied to electromechanical equipment data collection. However, existing technologies often neglect optimizing data sampling methods, resulting in inconsistent data quality. Furthermore, a lack of effective data processing methods results in data congestion, which can affect diagnostic logic. Furthermore, existing diagnostic systems are mostly based on a single fault model, making them difficult to adapt to the complex and changing fault characteristics of electromechanical equipment.

[0003] Therefore, how to efficiently and accurately collect and process the operation and maintenance data of electromechanical equipment and build an intelligent diagnostic architecture has become a technical problem that needs to be solved urgently. Summary of the Invention

[0004] The present application provides a method and equipment for diagnosing electromechanical equipment based on a multi-dimensional operation and maintenance data training model, which is used to solve the technical problems in the prior art, such as the lack of an intelligent diagnostic architecture and the inability to ensure the effectiveness and accuracy of data sampling and diagnostic decisions.

[0005] In view of the above problems, the present application provides a mechatronic equipment diagnosis method and operation and maintenance equipment based on a multi-dimensional operation and maintenance data training model.

[0006] In the first aspect, the present application provides a method for diagnosing electromechanical equipment based on a multi-dimensional operation and maintenance data training model, the method comprising: deploying a multi-dimensional sensing network in the electromechanical equipment area, and obtaining electromechanical current data by adopting a staggered cross-sampling method; introducing a first verification port and a second verification port to construct a diagnostic network architecture, wherein the first verification port is arranged at the transmission starting segment of the sensing side to perform multi-dimensional data aggregation, and the second verification port is arranged at the middle segment of the transmission to perform data vector encapsulation; based on the diagnostic network architecture, the electromechanical current data is connected to the network, and is transmitted back to the data center by performing aggregation screening based on the first verification port and data structure conversion and encapsulation based on the second verification port, and based on the built-in operation and maintenance diagnostic device, operation and maintenance fault decision-making and trend prediction are performed, and a fault diagnosis sheet is output.

[0007] On the second aspect, the present application provides an operation and maintenance device, which is composed of a multi-dimensional sensor network, a diagnostic network architecture and an operation and maintenance diagnostic device; wherein, the multi-dimensional sensor network performs front-end sampling and on-site early warning, and the diagnostic network architecture and the operation and maintenance diagnostic device perform data interaction and processing, and jointly constitute a diagnostic system for electromechanical equipment.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages: The embodiment of the present application provides a method for diagnosing electromechanical equipment using a multi-dimensional operation and maintenance data training model. A multi-dimensional sensor network is deployed in the electromechanical equipment area, and electromechanical current data is obtained by adopting a staggered cross-sampling method. A first verification port and a second verification port are introduced to construct a diagnostic network architecture. The first verification port is arranged at the transmission start segment of the sensor side to perform multi-dimensional data aggregation, and the second verification port is arranged at the transmission middle segment to perform data vector encapsulation. Based on the diagnostic network architecture, the electromechanical current data is connected to the network, and is transmitted back to the data center by performing aggregation screening based on the first verification port and data structure conversion and encapsulation based on the second verification port. Based on the built-in operation and maintenance diagnostic device, operation and maintenance fault decisions and trend predictions are made, and a fault diagnosis sheet is output. The method is used to solve the technical problems in the prior art of lacking an intelligent diagnostic architecture and being unable to guarantee the effectiveness and accuracy of data sampling and diagnostic decisions. Flexible sampling is carried out based on demand, and a systematic diagnostic architecture is constructed to improve the intelligence and accuracy of electromechanical equipment diagnostic decisions. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 A schematic flow chart of a method for diagnosing electromechanical equipment using a multi-dimensional operation and maintenance data training model is provided for this application; Figure 2 This application provides a schematic diagram of the configuration flow of the sampling mode in the electromechanical equipment diagnosis method of a multi-dimensional operation and maintenance data training model. DETAILED DESCRIPTION

[0010] The present application provides a method for diagnosing electromechanical equipment and operation and maintenance equipment based on a multi-dimensional operation and maintenance data training model, deploys a multi-dimensional sensor network for staggered cross-sampling, obtains electromechanical current data, introduces a first verification port and a second verification port, builds a diagnostic network architecture, connects the electromechanical current data to the network, performs aggregation screening based on the first verification port and data structure conversion and encapsulation based on the second verification port, and transmits it back to the data center. Based on the built-in operation and maintenance diagnostic device, operation and maintenance fault decisions and trend predictions are made, and a fault diagnosis sheet is output to solve the technical problems in the existing technology of lacking an intelligent diagnostic architecture and being unable to ensure the effectiveness and accuracy of data sampling and diagnostic decisions.

[0011] Example 1:

[0012] like Figure 1 As shown, the present application provides a method for diagnosing electromechanical equipment based on a multi-dimensional operation and maintenance data training model, the method comprising: S1: Deploy a multi-dimensional sensor network in the electromechanical equipment area and obtain electromechanical current data by adopting staggered cross sampling.

[0013] In the embodiments of this application, a multidimensional sensor network refers to a network system composed of multiple different types of sensors capable of sensing a variety of physical quantities. These sensors are deployed in a targeted manner based on the type of parameters required to monitor the operation of the electromechanical equipment, such as voltage, current, vibration, and temperature. Their deployment locations are carefully planned to fully cover the key parts of the electromechanical equipment in order to accurately capture various types of information during the equipment's operation.

[0014] At the same time, staggered sampling is employed to obtain more comprehensive and accurate electromechanical current data. Staggered sampling involves flexibly and specifically setting sampling times and frequencies based on the data characteristics of different sensor sources, such as the speed of data change and whether it exhibits periodicity, as well as fault characteristics identified from historical equipment fault records. For example, for sensor sources with rapidly changing data and non-periodic, non-stationary fault characteristics, the sampling frequency may be increased, and the sampling time may be staggered with that of other sensor sources.

[0015] During the implementation process, the data characteristics of the numerous multi-dimensional sensor sources in the electromechanical equipment area are first determined in detail, including the type of sensor source (such as current sensor, vibration sensor, etc.) and the specific deployment location. Next, the historical fault records accumulated over a long period of time in the electromechanical equipment area are called up to deeply explore the fault characteristics, such as identifying whether the fault source is an overheating fault in the motor or a short circuit fault in the line. Then, based on the determined data characteristics and fault characteristics, the sampling mode is scientifically and reasonably configured. Finally, the multi-dimensional sensor network is driven by this sampling mode to obtain electromechanical current data. This data collection method can effectively avoid data redundancy, while improving the validity and reliability of the data, providing a solid data foundation for the subsequent accurate diagnosis of electromechanical equipment.

[0016] Furthermore, using staggered cross sampling, step S1 of this application includes: For the multi-dimensional sensor sources in the electromechanical equipment area, data characteristics are determined, wherein the sensor source type and deployment location are identified; historical fault records of the electromechanical equipment area are called to mine fault characteristics, wherein the fault source is identified; a sampling mode is configured according to the data characteristics and the fault characteristics; wherein the data characteristics include linear and nonlinear types, and the fault characteristics include periodic stable fault types, periodic non-stationary fault types, non-periodic stable fault types, and non-periodic non-stationary fault types.

[0017] Specifically, the multi-dimensional sensing source is a collection of sensors of various types, which are deployed in specific areas of electromechanical equipment to comprehensively collect information during the operation of the equipment.

[0018] Among them, there are a wide variety of sensor sources. For example, current sensors, which use principles such as electromagnetic induction to monitor and collect current values ​​in real time during equipment operation, show specific patterns of change during normal equipment operation and under different fault conditions. Based on sensing needs, different types of sensors are deployed in appropriate locations on electromechanical equipment. For example, temperature sensors are deployed near the windings of a motor. Since the windings generate heat due to the passage of current during operation, temperature changes here can directly reflect the load and operating status of the motor. The data characteristics may show complex changes related to factors such as the motor's operating time and load size.

[0019] Furthermore, the data characteristics of each sensor source are determined, which are classified as linear or nonlinear in this application. Furthermore, the historical fault records are detailed records of various types of faults that occurred during the past operation of electromechanical equipment. By mining the fault characteristics, information with regularity that can characterize the essential characteristics of the fault is extracted from the historical records, and the corresponding fault source, that is, the root cause of the equipment failure, is determined.

[0020] For example, when a critical component suddenly breaks inside a device, this fault source can cause dramatic and irregular changes in the device's operating parameters, exhibiting characteristics of a non-periodic and non-stationary fault. Due to its suddenness and complexity, this type of fault poses a significant threat to the normal operation of the device. Different faults occur in different locations on the device, and the sampling requirements vary depending on the fault characteristics.

[0021] Sampling patterns are further configured based on data and fault characteristics. Targeted configuration ensures the validity and completeness of the sampled data stream. Linear data characteristics indicate that the changes in the data collected by the sensor source exhibit a relatively regular linear relationship with certain factors. For example, within a certain load range, the motor current value and load size may exhibit a linear growth relationship. Nonlinear data characteristics, on the other hand, indicate that the data changes more complexly and cannot be described by a simple linear relationship, such as sensor data affected by multiple complex factors.

[0022] Fault characteristics are divided into periodic stable faults, such as the vibration abnormality caused by periodic belt relaxation mentioned above. This type of fault is periodic in time and the fault state is relatively stable; periodic non-stationary faults are faults that occur periodically, but during the fault period, the operating parameters and other characteristics of the equipment are not stable, that is, the fault characteristics show non-stationary changes within the period; non-periodic stable faults refer to faults that do not occur according to a fixed period, but after the fault occurs, the fault state of the equipment is relatively stable; non-periodic non-stationary faults, such as sudden breakage of components, have no time pattern, and the operating state of the equipment is extremely unstable after the fault occurs.

[0023] By staggered cross-sampling, a scientific and reasonable sampling pattern is constructed to obtain high-quality data that meets the needs of accurate diagnosis of electromechanical equipment.

[0024] Further, such as Figure 2 As shown, the sampling mode is configured, and step S1 of this application includes: Combine data characteristics with fault characteristics to determine a first sampling class, wherein the first sampling class is any combination of characteristics; for the first sampling class, locate the first sensing part and set a first sampling condition, wherein the sampling condition at least includes a sampling frequency; identify the first sampling condition, cross-fuse the various sampling conditions in a sampling time series, and determine a sampling mode; and drive data acquisition of the multidimensional sensor network according to the sampling mode.

[0025] Combine data characteristics with fault characteristics, for example, linear data characteristics with periodic stationary faults, or nonlinear data characteristics with aperiodic nonstationary fault. Determine a first sampling class, which can be any of these combinations. This combination allows for the development of more precise sampling strategies for different combinations of data and fault characteristics.

[0026] For the determined first sampling class, a sensing detection portion that meets the combination method is determined as the first sensing portion. The first sensing portion refers to the set of sensing sources corresponding to the first sampling class, and the data characteristics of these sensing sources and the fault characteristics associated with them meet the characteristics of the current first sampling class.

[0027] Sampling conditions are further set for the first sensing component based on the sampling requirements. These first sampling conditions, which are the rules guiding the sensor's data collection, include at least a sampling frequency. The sampling frequency refers to the number of times the sensor collects data per unit time and is crucial for accurately capturing data changes. For different sampling types, an appropriate sampling frequency must be set based on their characteristics.

[0028] For example, for the combination of nonlinear data characteristics and non-periodic and non-stationary fault types, since the data changes are complex and the faults occur suddenly and irregularly, in order to be able to capture the data changes at the moment the fault occurs in time, a higher sampling frequency needs to be set; while for the combination of linear data characteristics and periodic and stationary fault types, since the data changes are relatively regular, a lower sampling frequency can be set.

[0029] Preferably, in addition to the sampling frequency, the sampling conditions may also include parameters such as sampling duration and sampling accuracy, which together constitute complete sampling conditions.

[0030] Specifically, a sampling time series refers to a series of sampling moments arranged in chronological order. Cross-fusion involves rationally arranging different combinations of sampling conditions across time to avoid interference between sensor sources and maximize system resources. Cross-fusion can stagger sampling times, preventing too many sensor sources from simultaneously collecting data at the same time and reducing system load. By setting sampling patterns, it's possible to ensure that each sensor source in a multidimensional sensor network works together according to predetermined rules, achieving efficient and accurate data collection.

[0031] Finally, data acquisition and driving of the multi-dimensional sensor network are performed based on the determined sampling pattern. This means converting the sampling pattern into actual control signals and sending them to each sensor in the multi-dimensional sensor network. This enables the acquisition of comprehensive, accurate, and representative electromechanical equipment operating data, providing a solid data foundation for subsequent fault diagnosis and operation and maintenance decisions.

[0032] S2: Introduce the first verification port and the second verification port to build a diagnostic network architecture, wherein the first verification port is arranged at the transmission starting segment of the sensor side to perform multi-dimensional data aggregation, and the second verification port is arranged at the transmission middle segment to perform data vector encapsulation.

[0033] First, the purpose of introducing the first and second verification ports must be clarified. During the monitoring of electromechanical equipment, multidimensional sensor networks collect a large amount of operational data. To ensure the quality of the data ultimately used for diagnosis and improve the accuracy and reliability of diagnosis, specific verification and processing mechanisms are required.

[0034] In this application, the data transmission process is divided into two ports for processing to avoid the complexity of data processing in a single stage that affects the overall continuity and provide a coherent and standardized diagnostic architecture.

[0035] The first verification port is located at the sensor-side transmission start segment. This segment is the initial section where data from each sensor begins transmission. This segment aggregates and pre-processes multi-linear sensor data, integrating and summarizing diverse data collected from sensors of different types and locations.

[0036] Specifically, it includes classifying and counting the data, removing some obviously erroneous or duplicate data, making the data more orderly and refined to form a data set containing a variety of information, preparing for subsequent processing and analysis.

[0037] The second verification port is then deployed mid-transmission. This refers to the point in between when data is transmitted from the sensor to the data center. This is because after initial processing at the first verification port, the data requires further format conversion and packaging for subsequent processing and analysis.

[0038] If all processing is done at the first verification port, the continuity of transmission of new and old sampling data cannot be guaranteed. Targeted port configuration can ensure the accuracy of data processing.

[0039] The second verification port's primary function is to perform data vectorization. This involves converting pre-processed data into a vector format with a specific format. For example, the second verification port extracts and transforms features from the data aggregated by the first verification port, converting it into a series of vectors with specific dimensions and meaning. These vectors contain key information about the operating status of electromechanical equipment and are formatted in a unified format, making them easier to process by subsequent machine learning algorithms or diagnostic models.

[0040] By introducing the first verification port and the second verification port and placing them in appropriate positions respectively, and performing the corresponding multi-dimensional data aggregation and data vector encapsulation functions, a complete diagnostic network architecture is constructed, which can effectively process and optimize the data collected by the multi-dimensional sensor network of electromechanical equipment, and provide high-quality data support for subsequent fault diagnosis and operation and maintenance decisions.

[0041] Furthermore, step S2 of the present application includes: connecting the multi-dimensional sensor network and the data center of the electromechanical diagnostic system via a transmission path, deploying a first verification port and a second verification port on the transmission path, and constructing a diagnostic network architecture; Among them, constructing a diagnostic network architecture includes: for the multi-dimensional sensor network, building a transmission path with the sensor deployment location as the starting location and the data center of the electromechanical diagnostic system as the end location; performing multi-threaded aggregation of sensor sampling return, and deploying a first verification port in the transmission path; deploying a second verification port in the transmission path, wherein the second verification port is placed after the first verification port.

[0042] First, establish the transmission path. A multidimensional sensor network is a network system composed of numerous sensors with different functions distributed across various locations on electromechanical equipment. These sensors are responsible for collecting various data during equipment operation in real time. Using the sensor deployment location as the starting point means planning data transmission from the actual location of each sensor installation. The electromechanical diagnostic system's data center is the core hub of the entire diagnostic system. It is responsible for receiving, storing, analyzing, and processing large amounts of data from the multidimensional sensor network to make accurate fault diagnosis and operation and maintenance decisions. Therefore, it is used as the end point of the transmission path.

[0043] When building a transmission path, it is necessary to consider a variety of factors, such as data transmission stability, transmission speed, anti-interference ability, etc.

[0044] Next, a first verification port is deployed, using multi-threaded aggregation of sensor sampling returns. Each sensor in the multidimensional sensor network corresponds to a data thread. Multi-threaded aggregation aggregates and summarizes multidimensional data. Specifically, after the multi-threaded data is aggregated and summarized using the first verification port, a single thread is determined for subsequent transmission. The aggregated data is then initially verified and processed. For example, data deduplication and filtering are performed to remove duplicate or erroneous data and improve data quality.

[0045] Furthermore, a second verification port is deployed in the transmission path. After the initial processing by the first verification port, the data is further feature-vectorized and encapsulated to ensure the convenience of subsequent diagnosis. Data vector encapsulation converts the initially processed data into a vector form in a specific format. It contains key information about the operating status of the electromechanical equipment and has a unified format, which is convenient for subsequent machine learning algorithms or diagnostic models to process. In the entire transmission path, the second verification port plays a connecting role. It further optimizes the quality and format of the data, providing strong support for accurate diagnosis in the data center.

[0046] Through the above steps, we have successfully built a complete diagnostic network architecture. On the basis of ensuring data flow efficiency, it can effectively and efficiently transmit the data collected by the multi-dimensional sensor network to the data center of the electromechanical diagnostic system, providing a solid foundation for fault diagnosis and operation and maintenance decision-making of electromechanical equipment.

[0047] Exemplarily, the configuration of the first verification port and the second verification port can be guided by the specific function of the port, and trained until convergence using a sample-driven training method, and further deployed on the transmission path to ensure its automated operation and processing.

[0048] S3: Based on the diagnostic network architecture, the electromechanical current data is connected to the network, and is transmitted back to the data center by performing aggregation screening based on the first verification port and data structure conversion and encapsulation based on the second verification port. Based on the built-in operation and maintenance diagnostic device, operation and maintenance fault decisions and trend predictions are made, and a fault diagnosis sheet is output.

[0049] First, the electromechanical current data is connected to the network based on the diagnostic network architecture. The diagnostic network architecture is an integrated whole consisting of a multi-dimensional sensor network, transmission paths, and the first and second verification ports deployed within it. It provides a stable and efficient framework for the transmission and processing of electromechanical current data. Electromechanical current data, as key data reflecting the operating status of electromechanical equipment, is connected to the transmission channels established by the diagnostic network architecture, enabling it to be transmitted according to established paths and rules.

[0050] Subsequently, aggregate screening based on the first verification port is performed. Aggregate screening involves aggregating and integrating dispersed electromechanical current data from different sensor sources and screening and checking it to ensure data quality and validity. Data is also screened based on pre-set rules, such as normal current and voltage ranges. Data outside the normal range is further analyzed to determine whether it is truly abnormal data or erroneous data caused by sensor failure or interference. If the current data transmitted by a sensor is found to deviate significantly from the normal operating range and is short-lived, it may be due to a momentary anomaly caused by electromagnetic interference. In this case, the first verification port can correct or mark the data for subsequent processing.

[0051] Next, data structure conversion and encapsulation are performed based on the second verification port. This involves converting the data, aggregated and screened by the first verification port, from its original acquisition format into a unified data structure more suitable for subsequent storage, transmission, and analysis, and converting it into a feature vector format. Encapsulation, based on this data structure conversion, adds specific identifiers and descriptive information to the data, transforming it into a complete, easily identifiable and processable data unit.

[0052] After completing the aforementioned processing, the data is then transmitted back to the data center. As the core of the entire electromechanical equipment diagnostic system, the data center possesses powerful data storage, computing, and analysis capabilities. It is responsible for receiving processed electromechanical current data transmitted from the diagnostic network architecture and conducting subsequent in-depth analysis. For example, the data center receives the electromechanical data packets encapsulated and transmitted back by the second verification port via a high-speed network interface and stores them in a dedicated database for further processing and analysis.

[0053] Finally, based on the built-in operation and maintenance diagnostics, operation and maintenance failure decisions and trend predictions are made, and a fault diagnosis sheet is output. The operation and maintenance diagnostics tool is an intelligent diagnostic tool trained based on specific algorithms and models, and is built into the data center. It uses received electromechanical current data, historical data, and failure cases, and through machine learning, data analysis, and other technical means to evaluate and diagnose the operating status of electromechanical equipment, determine whether the equipment is currently faulty, and determine the type and severity of the fault, and then provide corresponding treatment recommendations. For example, if the operation and maintenance diagnostics analyzes the electromechanical current data and finds that the motor current fluctuations are abnormal and exceed the normal range, combined with historical fault data, it may be due to a short circuit fault in the motor winding. In this case, the operation and maintenance diagnostics tool will provide a decision recommendation to stop the equipment operation and inspect and repair the motor winding.

[0054] At the same time, the diagnostics system also performs trend predictions. Based on the changing trends of motor current data and the equipment's operating history, the diagnostics system predicts potential equipment failures within a certain period of time. For example, by analyzing the trend of gradually increasing motor current over a period of time, the diagnostics system predicts a possible motor failure due to overload, alerting maintenance personnel to conduct equipment inspections and load adjustments in advance.

[0055] Ultimately, the O&M diagnostics system compiles these diagnostic results and decision-making recommendations into a fault diagnosis sheet, providing a robust basis for maintenance and management of electromechanical equipment. These sheets typically include detailed information such as the equipment name, fault type, time of occurrence, cause analysis, and recommended solutions, enabling maintenance personnel to quickly understand equipment status and take appropriate action.

[0056] Furthermore, performing aggregate screening based on the first verification port, step S3 of this application includes: The electromechanical current data flow is aggregated to the first verification port for data preprocessing and verification filtering to determine the verification data, wherein data processing calibration or resampling compensation is used as the preprocessing method; the verification data is grouped according to the data correlation based on the fault characteristics to determine the aggregated flow data.

[0057] Specifically, electromechanical current data encompasses multidimensional sensory information from the motor's operation, collected by sensors distributed throughout the device. Within the multidimensional sensor network's transmission system, data flows along a predetermined path to the first verification port, where it is aggregated.

[0058] Data preprocessing and verification filtering were then carried out to improve data quality and provide a reliable basis for subsequent analysis. Two common methods for data preprocessing are data processing calibration and resampling compensation. Data processing calibration uses methods such as interpolation to correct for data acquisition errors, such as missing data. For example, due to long-term use of sensors, zero-point drift may occur, resulting in deviations in the collected current data. Data processing calibration can bring the data closer to the true value.

[0059] Resampling compensation addresses issues such as inconsistent sampling frequencies that may arise during data collection by performing targeted sampling and adding them to the electromechanical current data. For example, if some electromechanical current data sampling points are lost due to network fluctuations over a period of time, resampling compensation can be performed based on the changing trends of the preceding and following data to fill in the missing data points, ensuring a complete and smooth data sequence.

[0060] After data preprocessing, verification and filtering are performed. This involves filtering the preprocessed data according to specific rules to determine whether it represents a fault in an electromechanical device. Specifically, a threshold is set based on the critical value of the data that indicates a fault. If the threshold is met, indicating a potential fault risk in the current data state, data filtering is performed, data is transmitted back, and diagnostic decisions are made. If the current data state represents normal operation, it is intercepted and excluded from subsequent diagnostic decisions.

[0061] After data verification is complete, the verified data is grouped based on the data correlation degree of the fault characteristics to determine the aggregated flow data. Data correlation indicates the degree of correlation between different fault characteristics. For example, a fault caused by a motor winding short circuit is characterized by a sharp increase in current and a decrease in voltage. When grouping the verified data, data showing an increase in current and a decrease in voltage are grouped together. This data has a high correlation with the fault characteristics of a winding short circuit.

[0062] Through this data correlation grouping method based on fault characteristics, the verification data is divided into different sets. Each set is an aggregated stream data, so that different types of data can be subsequently analyzed and processed in depth, improving the accuracy and efficiency of fault diagnosis.

[0063] Further, verification and filtering are performed to determine the verification data. Step S3 of this application includes: According to the fault characteristics, a data checkpoint based on a rigid data threshold is set, wherein the setting is performed based on a non-fault data threshold; according to the data checkpoint, data screening is performed on the pre-processed data to determine the verification data, wherein the data portion that does not meet the rigid data threshold is passed, and the data portion that meets the rigid data threshold is intercepted.

[0064] First, based on the fault characteristics, a data checkpoint based on a rigid data threshold is set. The rigid data threshold is a numerical standard with clear boundaries determined based on the data range during normal operation of the equipment. It is set based on a non-fault data threshold, that is, a reasonable critical value is determined based on the fluctuation range of various data under normal operating conditions of the equipment. As long as it is within this critical value, there is no probability of electromechanical operation failure. For example, for a motor with a rated operating current of 5-8 amperes, after a large amount of statistical analysis of normal operating data, it is determined that the current fluctuation range during normal operation is between 6-7.5 amperes. In this way, 6 amperes can be set as the lower limit rigid data threshold, and 7.5 amperes can be set as the upper limit rigid data threshold, thereby constructing a checkpoint for the motor current data.

[0065] Next, after the rigid data threshold level is set, it is applied to the electromechanical current data after preprocessing, such as calibration or resampling compensation. The data screening process acts as a rigorous quality control procedure, examining each preprocessed data individually. For example, if the current data being processed is motor current data, and the current value collected at a certain moment is 5.5 amps, which is lower than the set lower rigid data threshold of 6 amps, this data does not meet the rigid data threshold. According to the set rules, this data will be judged as data with a fault risk and passed, allowing it to proceed to further analysis, as it may indicate potential motor operational issues that require further investigation. Conversely, if the collected current value is 7 amps, which falls within the rigid data threshold range of 6-7.5 amps, thus meeting the rigid data threshold, this data is considered normal data and intercepted, avoiding the subsequent in-depth analysis process specifically targeting abnormal data, as it aligns with the data characteristics of normal equipment operation.

[0066] Through meticulous screening operations, verification data that may be related to the fault can be accurately screened out from a large amount of pre-processed data, providing a reliable data basis for subsequent in-depth analysis and accurate diagnosis based on fault characteristics, effectively improving the efficiency and accuracy of fault diagnosis, and avoiding the influence of interference from a large amount of normal data on the judgment of the actual operating status of the equipment.

[0067] Furthermore, data structure conversion and encapsulation based on the second verification port are performed. Step S3 of this application includes: determining a unified data structure; transferring the aggregated flow data to the second verification port, determining the conversion characteristic state by identifying the characteristic state, and performing conversion based on the unified data structure; for the conversion characteristic state, performing encapsulation and routing coupling based on the characteristic vector to determine the encapsulated data.

[0068] First, a unified data structure must be established. During the operation of electromechanical equipment, the electromechanical current data collected by the multidimensional sensor network initially varies in format. This data comes from sensors of different types and manufacturers, and each sensor generates data in a unique format based on the characteristics of the physical quantity it monitors. For example, a current sensor outputs a sequence of values ​​reflecting the time-varying current intensity, while a vibration sensor outputs complex waveform data related to the equipment's vibration amplitude and frequency.

[0069] In order to facilitate the subsequent unified processing, storage and transmission of data, a unified data structure needs to be determined. This structure specifies the organization of data, the types of data elements and their relationships.

[0070] Next, the aggregated data stream—preprocessed and filtered by the first verification port, grouped according to the data relevance of the fault characteristics—is transferred and aggregated to the second verification port. This data flows along the transmission path of the diagnostic network architecture to the second verification port, which is located in the middle of the transmission process.

[0071] After reaching the second verification port, the data is analyzed by identifying characteristic states, which are the various operational characteristics of the electromechanical equipment reflected in the aggregated flow data. These include the values ​​and trends of equipment operating parameters. For example, the current value, voltage stability, and frequency of current fluctuations in electromechanical current data are all considered characteristic states. Identifying characteristic states involves extracting this key characteristic information from the aggregated flow data.

[0072] Afterwards, based on the determined unified data structure, the identified characteristic states are converted, and the converted data are reorganized to obtain converted characteristic states that conform to the unified data structure.

[0073] Then, for the conversion characteristic state, feature vector-based encapsulation and routing coupling are performed. Specifically, the feature vector quantifies and vectorizes the key characteristic information in the conversion characteristic state to facilitate efficient computer processing and analysis. For example, the device's current value, voltage value, vibration amplitude and other characteristics are quantified into a multi-dimensional vector, with each dimension corresponding to an eigenvalue. Encapsulation is to add the necessary identification and description information to these feature vectors to make them an independent and complete data unit. For example, a timestamp is added to the feature vector to indicate the time when the device corresponding to this group of data is running; a device number is added to clarify the device to which the data belongs.

[0074] After encapsulating the data vector, multiple encapsulation units are routed and coupled to each other. A unified data system is determined as the encapsulated data, which is then efficiently and accurately transmitted to the data center in the diagnostic network architecture, providing reliable data support for subsequent equipment fault diagnosis and operation and maintenance decisions.

[0075] Furthermore, for the conversion characteristic state, encapsulation and routing coupling based on the characteristic vector is performed to determine the encapsulation data. Step S3 of the present application includes: The conversion feature state is traversed, the feature vector is decomposed and determined, and multiple feature layers are determined based on the feature progressive relationship, wherein each feature vector performs independent encapsulation; based on the feature transfer relationship between the multiple feature layers, a directed transfer method is determined and routing coupling is performed to determine the encapsulated data.

[0076] First, the conversion feature state is traversed, that is, the data after conversion to a unified data structure. Next, the feature vector is decomposed and determined, and multiple feature layers are determined based on the feature progression relationship. The feature vector quantifies and vectorizes the various characteristics of the device operation so that the computer can process and analyze this data more efficiently. Decomposing and determining the feature vector is to extract representative features from the conversion feature state and convert them into vector form. For example, from the operating data of the motor, the three features of current value, voltage value, and vibration amplitude are extracted to form a three-dimensional feature vector, with each dimension corresponding to an eigenvalue.

[0077] A feature progression relationship refers to a data logic relationship between different features, characterized by logical precedence, causal relationships, and differences in importance. For example, in motor fault diagnosis, a change in current may be the first sign of abnormal motor operation, followed by fluctuations in voltage and then increases in vibration amplitude, forming a feature progression relationship.

[0078] Based on this progressive relationship of features, multiple feature layers are determined. Features that are closely related to motor operation and play a key role in fault diagnosis are divided into a feature layer. For example, features that directly reflect the electrical performance of the motor, such as current and voltage values, are classified as the electrical feature layer; features that reflect the mechanical state of the motor, such as vibration amplitude and temperature, are classified as the mechanical feature layer. In this process, each feature vector is independently encapsulated, that is, each feature vector is processed separately, and the necessary identification and descriptive information is added to it, making it an independent data unit. For example, for the current value feature vector, a timestamp is added to indicate the time corresponding to the current value, and a device number is added to clearly indicate the motor to which the data belongs.

[0079] Finally, based on the feature transfer relationship between the multiple feature layers, a directed transfer method is determined and route coupling is performed. Route coupling is a directed transfer method that couples and associates different package units to obtain a standardized data system with an intuitive data structure as the packaged data. The feature transfer relationship between multiple feature layers describes the flow direction and mutual influence of information between different feature layers. For example, during motor operation, changes in the current and voltage values ​​of the electrical feature layer may trigger changes in the vibration amplitude and temperature of the mechanical feature layer. Determining a directed transfer method is to clarify the transmission direction of data between different feature layers based on this feature transfer relationship. For example, it is stipulated that data is transferred from the electrical feature layer to the mechanical feature layer, and then to other related feature layers; for example, how the lower layer is transferred to the upper layer, under multi-source data, the lower layer may be the cause and the upper layer may be the result, and the relevant package units of the lower layer are weighted to obtain the upper layer.

[0080] The encapsulated data not only contains the key characteristic information of the operation of electromechanical equipment, but also has the necessary properties for accurate transmission in the network, providing reliable data support for equipment fault diagnosis and operation and maintenance decision-making in data centers.

[0081] Furthermore, the fault diagnosis sheet is outputted. Step S3 of this application includes: Historical fault records are called up, and the historical fault records are adjusted based on an architecture encapsulated by data vectors to determine sample data, wherein the sample data includes encapsulated samples, diagnostic samples, and predicted samples; based on the sample data, the operation and maintenance diagnostic device is constructed through supervised training until convergence and built into the data center; the operation and maintenance diagnostic device receives the encapsulated data and decides to output the fault diagnosis form; the fault diagnosis form is displayed on the terminal as an early warning, and the source sensor is located, and reverse transmission and on-site early warning based on the source sensor are combined with the transmission path.

[0082] Specifically, historical fault records are collected as a collection of information related to past operational failures of electromechanical equipment, covering the time of occurrence, symptoms, corresponding operating parameters, and resolution measures. These records are adjusted using an architecture based on data vector encapsulation. This architecture organizes and represents data according to specific rules, converting complex fault information into a vector form that is easily processed by computers. This process filters, extracts, and converts various pieces of information from historical fault records, such as operating parameters closely related to fault diagnosis, such as current, voltage, and temperature. These are then encapsulated into vectors in a sequential format, thereby converting the historical fault records into sample data.

[0083] Sample data includes packaged samples, diagnostic samples, and prediction samples. Packaged samples are historical fault records that have been encapsulated into data vectors, storing key information about equipment failures in vector form. Diagnostic samples are data labeled with fault types and diagnostic results based on packaged samples, providing clear targets for supervised training. Prediction samples are used to verify the performance of operation and maintenance diagnostic tools and predict future failures.

[0084] Through the above sample processing, it is converted into a data system that is compatible with this technical architecture to ensure that the subsequent training results are highly consistent with the technical solution.

[0085] Based on sample data, with the help of supervised training, the model is trained using diagnostic samples with labeled information to allow the model to learn the mapping relationship between fault characteristics and fault types. During training, the model continuously adjusts parameters to minimize the error between the prediction results and the labeled information. When the error reaches a minimum value and no longer changes significantly, the training converges. For example, a neural network model is used to train sample data, predict the fault type of the output package sample, compare it with the labeled information of the diagnostic sample, and adjust the weights and bias through the backpropagation algorithm until the error meets the preset convergence conditions. The trained operation and maintenance diagnostic device is built into the data center. As the core hub of the electromechanical equipment diagnosis system, the data center has powerful storage and computing capabilities, providing a stable operating environment and efficient data processing support for the operation and maintenance diagnostic device, enabling it to receive package data from the multi-dimensional sensor network in real time and diagnose faults quickly and accurately.

[0086] The operation and maintenance diagnostics device receives real-time encapsulated data from electromechanical equipment, processed by the second verification port and organized according to the data vector encapsulation architecture. Based on the learned mapping between fault characteristics and fault types, it analyzes and determines the equipment's current operating status and possible fault types. For example, if it receives motor encapsulation data and finds an abnormally high current value and temperature outside the normal range, it determines that the motor may be overloaded and generates a fault diagnosis form containing information such as the fault type, severity, and possible cause.

[0087] Furthermore, the fault diagnosis sheet is displayed to maintenance personnel via terminal devices such as display screens, enabling terminal display warnings. The source sensor collecting abnormal data is then located to pinpoint the specific location of the fault. For example, by analyzing the source of the fault diagnosis sheet data, it can be determined that a temperature sensor near the motor winding is collecting abnormally high temperature data, locating the fault in the motor winding. This fault diagnosis information is then transmitted in the reverse direction of data transmission to the source sensor for reverse transmission. This provides timely on-site warnings, which are implemented through audible and visual alarms, indicator lights, and other indicators based on the source sensor.

[0088] For example, after the operation and maintenance diagnostic device determines that the motor is faulty, it transmits the fault information back to the source sensor near the motor via the transmission path. The source sensor triggers an audible and visual alarm to remind on-site staff to handle the fault.

[0089] Through the above process, historical fault records are used to build an operation and maintenance diagnostic tool to diagnose faults in real-time operation data of electromechanical equipment and provide timely warnings, which can effectively improve equipment reliability and operating efficiency and reduce equipment failure losses.

[0090] The present application provides a method for diagnosing electromechanical equipment based on a multi-dimensional operation and maintenance data training model, which has the following technical effects: 1. Reasonably deploy various types of sensors on electromechanical equipment, perform staggered cross-sampling according to preset sampling frequencies and rules, and continuously and in real time collect the operating data of the equipment to ensure the timeliness and accuracy of the data. At the same time, an on-site early warning mechanism is introduced to perform early warning based on the corresponding sensor on the diagnostic information, which can quickly lock the on-site fault area. Data processing and transmission 2. The first verification port is introduced to perform preliminary corrections and level screening on the data to ensure valid data related to potential faults. The second verification port is introduced for unified format conversion and feature vector encapsulation to improve the systematicity and intuitiveness of the data. At the same time, based on the data features processed above, an operation and maintenance diagnostic device is constructed and deployed in the data center. A systematic diagnostic network architecture is formed to adapt to continuous sampling and diagnosis in complex scenarios.

[0091] Example 2: Based on the same inventive concept as the electromechanical equipment diagnosis method of a multi-dimensional operation and maintenance data training model in the aforementioned embodiment, the present application provides an operation and maintenance device, which is composed of a multi-dimensional sensor network, a diagnostic network architecture and an operation and maintenance diagnostic device; wherein, the multi-dimensional sensor network performs front-end sampling and on-site early warning, and the diagnostic network architecture and the operation and maintenance diagnostic device perform data interaction and processing, and jointly constitute a diagnostic system for electromechanical equipment.

[0092] Specifically, multidimensional sensor networks play a key role in front-end sampling and on-site early warning within operation and maintenance equipment. In terms of front-end sampling, they use a variety of sensors, such as temperature sensors, pressure sensors, and current sensors, to extensively and accurately collect various data from the operation of electromechanical equipment. This data covers multiple dimensions, including temperature, pressure, current, and vibration, providing the raw data foundation for subsequent equipment status analysis. Furthermore, during on-site early warning, once abnormal data reaches a preset warning threshold, the multidimensional sensor network can immediately activate the on-site early warning mechanism. This alerts personnel directly at the equipment site through audible and visual alarms, flashing indicator lights, and other means, prompting them to promptly address potential issues.

[0093] The multi-dimensional sensor network is connected to the diagnostic network architecture via wired or wireless transmission links. Data collected by the sensors is continuously transmitted through these transmission links to the first verification port of the diagnostic network architecture, initiating subsequent data processing.

[0094] Various types of sensors work based on different physical principles and periodically collect equipment operation data according to established sampling frequencies and rules. The diagnostic network architecture is mainly responsible for data interaction and processing.

[0095] On the one hand, it receives raw data from the multi-dimensional sensor network and aggregates this data through the first verification port, integrating data collected by sensors of different types and locations. It then performs preliminary data preprocessing and verification filtering to improve data quality. On the other hand, the second verification port encapsulates the data processed by the first verification port into a data vector, converting the data into a format suitable for subsequent analysis and transmission. The processed data is then transmitted to the operation and maintenance diagnostic device, while also receiving control information or instructions from the operation and maintenance diagnostic device, enabling two-way data interaction.

[0096] The maintenance diagnostics unit is the core decision-making component of the entire diagnostic system. It receives processed, encapsulated data from the diagnostic network architecture and performs in-depth analysis to determine the current operating status of the electromechanical equipment, identifying the presence of faults, their type, and their severity. Furthermore, based on historical fault records and real-time data, it predicts future equipment operating trends and ultimately generates a fault diagnosis form, providing a basis for decision-making in equipment maintenance and management.

[0097] In terms of overall operational logic, the multi-dimensional sensor network first collects comprehensive data from electromechanical equipment at the front end and transmits the collected raw data to the diagnostic network architecture. The diagnostic network then performs a series of processing steps, including data aggregation, preprocessing, verification and filtering, and data vector encapsulation, before transmitting high-quality data to the operation and maintenance diagnostics unit. The operation and maintenance diagnostics unit uses built-in diagnostic models to analyze the data, determine the equipment's operating status, generate a fault diagnosis report, and feed the results back to the diagnostic network architecture and terminal devices. This forms a closed-loop, efficient electromechanical equipment diagnostic system, ensuring stable equipment operation and timely maintenance.

[0098] Through the above-mentioned detailed description of the electromechanical equipment diagnosis method based on a multi-dimensional operation and maintenance data training model in this specification, those skilled in the art can clearly understand the electromechanical equipment diagnosis method and operation and maintenance equipment based on a multi-dimensional operation and maintenance data training model in this embodiment. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description.

[0099] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present application. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application is not limited to the embodiments shown herein, but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for diagnosing electromechanical equipment based on a multi-dimensional operation and maintenance data training model, characterized in that: The method comprises: Deploy a multi-dimensional sensor network in the electromechanical equipment area and acquire electromechanical current data by adopting staggered cross sampling. The electromechanical current data is the spatiotemporal flow data of the electromechanical equipment. A first verification port and a second verification port are introduced to construct a diagnostic network architecture. The first verification port is deployed at the transmission start segment of the sensor side to perform multi-dimensional data aggregation, and the second verification port is deployed at the transmission mid-segment to perform data vector encapsulation. Based on the diagnostic network architecture, the electromechanical current data is connected to the network, and is transmitted back to the data center by performing aggregation screening based on the first verification port and data structure conversion and encapsulation based on the second verification port. Based on the built-in operation and maintenance diagnostic device, operation and maintenance fault decisions and trend predictions are made, and a fault diagnosis sheet is output.

2. The electromechanical equipment diagnosis method based on a multi-dimensional operation and maintenance data training model according to claim 1, characterized in that: Adopt staggered cross sampling method, including: For the multi-dimensional sensor sources in the electromechanical equipment area, determining data characteristics, wherein the sensor source type and deployment location are identified; calling the historical fault records of the electromechanical equipment area, mining the fault characteristics, wherein the fault source is identified; configuring a sampling mode according to the data characteristics and the fault characteristics; The data characteristics include linear and nonlinear types, and the fault characteristics include periodic stable fault types, periodic non-stationary fault types, non-periodic stable fault types, and non-periodic non-stationary fault types.

3. The electromechanical equipment diagnosis method based on a multi-dimensional operation and maintenance data training model according to claim 2, characterized in that: Configure the sampling mode, including: Combining the data characteristics and the fault characteristics to determine a first sampling class, wherein the first sampling class is any combination of characteristics; For the first sampling class, positioning a first sensing portion and setting a first sampling condition, wherein the sampling condition at least includes a sampling frequency; Identify the first sampling condition, cross-fuse the sampling conditions in a sampling time sequence, and determine a sampling mode; According to the sampling mode, data collection and driving of the multi-dimensional sensor network are performed.

4. The electromechanical equipment diagnosis method based on a multi-dimensional operation and maintenance data training model according to claim 1, characterized in that: Connecting the multi-dimensional sensor network to the data center of the electromechanical diagnostic system via a transmission path, deploying a first verification port and a second verification port on the transmission path to build a diagnostic network architecture; Among them, building a diagnostic network architecture includes: For the multi-dimensional sensor network, a transmission path is established with the sensor deployment location as the starting location and the data center of the electromechanical diagnostic system as the end location; Perform multi-threaded aggregation of sensor sampling return and deploy a first verification port in the transmission path; A second verification port is deployed in the transmission path, wherein the second verification port is placed after the first verification port.

5. The electromechanical equipment diagnosis method based on a multi-dimensional operation and maintenance data training model according to claim 1, characterized in that: Perform aggregate screening based on the first verification port, including: The electromechanical current data is transferred and aggregated to the first verification port, and data preprocessing and verification filtering are performed to determine verification data, wherein data processing calibration or resampling compensation is used as a preprocessing method; The verification data is grouped according to the data correlation degree based on the fault characteristics to determine the aggregated flow data.

6. The electromechanical equipment diagnosis method based on a multi-dimensional operation and maintenance data training model according to claim 5, characterized in that: Perform verification filtering to determine verification data, including: According to the fault characteristics, a data checkpoint based on a rigid data threshold is set, wherein the setting is performed based on a non-fault data threshold; According to the data checkpoint, the pre-processed data is screened to determine the verification data, wherein the data portion that does not meet the rigid data threshold is passed, and the data portion that meets the rigid data threshold is intercepted.

7. The electromechanical equipment diagnosis method based on a multi-dimensional operation and maintenance data training model according to claim 5, characterized in that: Execute data structure conversion and encapsulation based on the second verification port, including: Determine a unified data structure; Transferring the aggregated flow data to the second verification port, identifying a characteristic state, and performing a conversion based on a unified data structure to determine a conversion characteristic state; For the conversion characteristic state, encapsulation and routing coupling based on the characteristic vector is performed to determine encapsulation data.

8. The electromechanical equipment diagnosis method based on a multi-dimensional operation and maintenance data training model according to claim 7, characterized in that: For the conversion characteristic state, performing encapsulation and routing coupling based on the characteristic vector to determine encapsulation data includes: Traversing the conversion feature state, decomposing and determining feature vectors and determining a plurality of feature layers according to a feature progression relationship, wherein each feature vector is independently encapsulated; Based on the feature transfer relationship between the multiple feature layers, a directed transfer mode is determined and routing coupling is performed to determine the encapsulated data.

9. The electromechanical equipment diagnosis method based on a multi-dimensional operation and maintenance data training model according to claim 8, characterized in that: Output troubleshooting ticket, including: Calling historical fault records, adjusting the historical fault records based on a data vector encapsulation architecture, and determining sample data, wherein the sample data includes encapsulation samples, diagnosis samples, and prediction samples; Based on the sample data, the operation and maintenance diagnostic tool is constructed through supervised training until convergence and is built into the data center; The operation and maintenance diagnostic device receives the encapsulated data and decides to output the fault diagnosis form; The terminal displays an early warning for the fault diagnosis sheet, locates the source sensor, and combines the transmission path for reverse transmission and on-site early warning based on the source sensor.

10. An operation and maintenance device, characterized in that: A method for diagnosing electromechanical equipment using a multidimensional operation and maintenance data training model according to any one of claims 1 to 9, wherein the operation and maintenance equipment comprises a multidimensional sensor network, a diagnostic network architecture, and an operation and maintenance diagnostic device; Among them, the multi-dimensional sensor network performs front-end sampling and on-site early warning, and the diagnostic network architecture and operation and maintenance diagnostic device perform data interaction and processing, and jointly constitute a diagnostic system for electromechanical equipment.

Images