Energy-saving motor health data assessment method and system based on cloud computing

Abstract

The invention relates to the technical field of energy-saving motor health data evaluation based on cloud computing, in particular to an energy-saving motor health data evaluation method and system based on cloud computing, and the method comprises the following steps: recognizing a motor needing to be concerned and weak signal features of a potential hidden fault indicated by the motor needing to be concerned; the cloud computing center issues an instruction to the local server; adjusting a data screening strategy of the local server; the local server uploads the monitoring data to the cloud computing center; the cloud computing center analyzes the monitoring data to confirm potential hidden faults; the local server performs conventional data screening on the motors which are not indicated to be monitored by the instruction; the cloud computing center updates the instruction according to the analysis result of the monitoring data, and the local server adjusts the data screening strategy of the local server according to the updated instruction. The reliability of motor health assessment can be improved.

Description

Technical Field

[0001] This invention relates to the technical field of cloud computing-based energy-saving motor health data assessment, specifically to a cloud computing-based energy-saving motor health data assessment method and system. Background Technology

[0002] In modern industrial production, cloud-based motor health assessment systems are widely used to ensure the stable operation of critical equipment and effectively manage energy consumption. These systems typically install various sensors on the motor to collect real-time operational data such as vibration, temperature, and current, and then upload this data to a remote cloud computing center for analysis. Leveraging its powerful data processing capabilities, the cloud system can establish a standard operating status reference for each motor. If the real-time data deviates from this reference, the system will issue an early warning, thus supporting preventative maintenance and energy consumption optimization.

[0003] However, this evaluation system faces challenges in industrial environments where motor workloads change frequently and drastically. For example, in a large automated production workshop, many motors need to be started, stopped, accelerated, and decelerated frequently according to the production rhythm, or quickly switched between different load levels. This unstable working state causes various operating data of the motors to change drastically and irregularly in a short period of time. These changes are caused by changes in normal operating conditions, but in terms of data performance, they are very similar to the characteristics of some early faults, causing the analysis program in the cloud computing center to issue frequent false alarms. Summary of the Invention

[0004] The purpose of this invention is to address the aforementioned shortcomings by proposing a cloud computing-based method and system for evaluating the health data of energy-saving motors.

[0005] The present invention adopts the following technical solution: A cloud-based method for assessing the health data of energy-efficient motors, comprising the following steps: The cloud computing center receives and analyzes the preliminary screening data uploaded by the local server to identify the motors that need attention and the weak signal characteristics of potential hidden faults in the motor indicators. Based on the weak signal characteristics of potential hidden faults in the motors that need to be monitored, the cloud computing center sends instructions to the local server, which instructs the monitoring of weak signal characteristics of the motors that need to be monitored. After receiving the instruction, the local server adjusts its data filtering strategy for the motors that need attention, including improving the acquisition accuracy of weak signal features and pausing the filtering of weak signal features. The local server uploads the monitoring data to the cloud computing center; The cloud computing center analyzes the monitoring data to identify potential hidden faults; The local server performs routine data filtering on motors that are not monitored as instructed. Based on the analysis results of the monitoring data, the cloud computing center updates the instructions, and the local server adjusts its data filtering strategy according to the updated instructions.

[0006] This technical solution effectively addresses the predicament in existing technologies where the initial screening mechanism of the local server may filter out weak signals of early, hidden faults, leading to the inability of the cloud to detect problems in a timely manner and ultimately causing production accidents. This application, by using the cloud to initially identify potential hidden faults and issue instructions, enables the local server to specifically improve the accuracy of weak signal acquisition and pause filtering. This ensures that crucial early fault evidence can be uploaded to the cloud for confirmation, achieving early and accurate identification of hidden faults and significantly improving the reliability and energy efficiency of motor health assessment.

[0007] This application also discloses a cloud-based energy-saving motor health data assessment system, applied to a cloud-based energy-saving motor health data assessment method. The system includes: The cloud-based analysis module receives and analyzes the preliminary screening data uploaded from the local server to identify the weak signal characteristics of motors that require attention and the potential hidden faults in the motor indicators. The cloud command delivery module, based on the weak signal characteristics of the motors to be monitored for potential hidden faults, sends commands from the cloud computing center to the local server, instructing the monitoring of the weak signal characteristics of the motors to be monitored. The local strategy adjustment module, upon receiving instructions, adjusts the data filtering strategy of the local server for the motors that need attention. The adjustments include improving the acquisition accuracy of weak signal features and pausing the filtering of weak signal features. The local data upload module uploads monitoring data to the cloud computing center; The cloud-based confirmation module analyzes monitoring data to identify potential hidden faults. The local routine filtering module performs routine data filtering on motors that are not monitored by instructions. The cloud-based command update module updates commands based on the analysis results of the monitoring data, and the local server adjusts its data filtering strategy according to the updated commands.

[0008] Through modular design, collaborative work between the cloud and local servers is achieved, ensuring early identification, accurate monitoring and confirmation of hidden faults, thereby significantly improving the efficiency and accuracy of motor health assessment and solving the information barrier problem in the existing system.

[0009] This application not only effectively reduces data transmission volume and computing costs, but also ensures precise monitoring of high-risk motors. Thus, while ensuring system economy, it significantly improves the accuracy and timeliness of fault diagnosis, providing strong support for the stable operation of industrial production.

[0010] To further understand the features and technical content of the present invention, please refer to the following detailed description and drawings of the present invention. However, the drawings provided are for reference and illustration only and are not intended to limit the present invention. Attached Figure Description

[0011] Figure 1 This is a flowchart of a cloud computing-based method for assessing the health data of an energy-saving motor according to the present invention. Figure 2 This is a schematic diagram of the structure of an energy-saving motor health data assessment system based on cloud computing according to the present invention. Detailed Implementation

[0012] The following specific embodiments illustrate the implementation of the present invention. Those skilled in the art can understand the advantages and effects of the present invention from the content disclosed in this specification. The present invention can be implemented or applied through other different specific embodiments, and various details in this specification can also be modified and changed based on different viewpoints and applications without departing from the spirit of the present invention. Furthermore, the accompanying drawings of the present invention are for simple illustrative purposes only and are not depictions of actual dimensions; this is stated in advance. The following embodiments will further describe the relevant technical content of the present invention in detail, but the disclosed content is not intended to limit the scope of protection of the present invention.

[0013] This embodiment provides a cloud computing-based method and system for assessing the health data of energy-saving motors, combined with... Figure 1 and Figure 2 As shown.

[0014] refer to Figure 1 A cloud-based method for assessing the health data of energy-efficient motors, comprising the following steps: The cloud computing center receives and analyzes the preliminary screening data uploaded by the local server to identify the motors that need attention and the weak signal characteristics of potential hidden faults in the motor indicators. Based on the weak signal characteristics of potential hidden faults in the motors that need to be monitored, the cloud computing center sends instructions to the local server, which instructs the monitoring of weak signal characteristics of the motors that need to be monitored. After receiving the instruction, the local server adjusts its data filtering strategy for the motors that need attention, including improving the acquisition accuracy of weak signal features and pausing the filtering of weak signal features. The local server uploads the monitoring data to the cloud computing center; The cloud computing center analyzes the monitoring data to identify potential hidden faults; The local server performs routine data filtering on motors that are not monitored as instructed. Based on the analysis results of the monitoring data, the cloud computing center updates the instructions, and the local server adjusts its data filtering strategy according to the updated instructions.

[0015] Among them, "cloud computing center" refers to one or more remote server clusters that have powerful data storage, processing and analysis capabilities, and can run complex machine learning algorithms and big data analysis models to perform in-depth evaluation and fault diagnosis of received motor health data.

[0016] A "local server" refers to an edge computing device deployed in an industrial site. Its main function is to collect motor sensor data in real time, perform preliminary data filtering and preprocessing, and adjust data processing strategies according to instructions from the cloud computing center. Local servers typically have some computing and storage capabilities, but their resources are limited, and they mainly play the role of data relay and local decision-making.

[0017] "Preliminary screening data" refers to a subset of raw sensor data that the local server, operating under normal conditions, initially processes and deems representative or potentially anomaly-prone. This preliminary filtering aims to reduce the amount of data transmitted to the cloud.

[0018] "Weak signal characteristics" refer to signal patterns that are low in intensity and easily masked by background noise, but indicate potential hidden faults in the motor (such as early insulation aging, bearing micro-cracks, etc.). These signals are often ignored under traditional coarse screening.

[0019] A "data filtering strategy" refers to a set of rules and algorithms used by a local server to process raw sensor data, including data sampling rate, filtering parameters, anomaly thresholds, and feature extraction methods. Adjusting the data filtering strategy means that these parameters can be dynamically changed to adapt to different monitoring needs.

[0020] Specifically, in the step of receiving and analyzing the preliminary screening data uploaded from the local server by the cloud computing center, the cloud computing center can employ various methods to receive and analyze the data. For example, the local server can push the preliminary screening data to the cloud computing center's data receiving interface in real time via the encrypted MQTT protocol. The cloud computing center can then utilize big data processing frameworks (such as Apache Spark) to perform parallel analysis on this data. During the analysis process, a deep learning-based anomaly detection model can be deployed, such as a long short-term memory network model, which, after training, can identify weak signal features associated with known motor fault patterns. Another implementation is that the local server periodically packages the preliminary screening data into CSV files and uploads them to a bucket at the cloud computing center via the SFTP protocol. The cloud computing center can then utilize batch processing jobs, such as Apache Flink, to process these files and use statistical methods, such as principal component analysis or independent component analysis, to identify potential anomaly patterns and weak signal features in the data.

[0021] Based on the weak signal characteristics of potential hidden faults in the motors of interest, the cloud computing center sends instructions to the local server to monitor these weak signal characteristics. The cloud computing center can generate a JSON-formatted instruction file containing information such as the motor ID, monitoring task type (e.g., "high-precision weak signal monitoring"), the frequency range of the target signal characteristics, and the monitoring duration. This instruction file can be sent to the local server via a secure API interface. The local server periodically polls the cloud computing center's instruction queue and parses its contents upon receiving a new instruction. Alternatively, the cloud computing center can publish monitoring instructions via a message queue service (such as RabbitMQ). The instruction can include a binary-coded monitoring configuration file that defines the monitoring parameters the local server needs to execute. The local server subscribes to this message queue, and upon receiving an instruction, directly loads and executes the configuration file.

[0022] After receiving instructions, the local server adjusts its data filtering strategy for the motors of interest, including improving the acquisition accuracy of weak signal features and pausing filtering of weak signal features. The local server can dynamically modify the configuration parameters of its data acquisition modules. For example, it can increase the sampling rate of the analog-to-digital converter from the usual 1kHz to 10kHz to capture weak signals at higher frequencies. Simultaneously, the local server can temporarily disable its internal low-pass or band-pass filters to ensure that signals across the entire frequency range are fully acquired, preventing the false filtering of weak signals. Another implementation approach is for the local server to maintain a programmable logic controller (PLC) interface, through which it can dynamically adjust the gain settings of the sensor front-end to improve the acquisition sensitivity of specific signals. Furthermore, the local server can update the parameters of its data preprocessing algorithms, for example, setting the threshold of the noise suppression algorithm to an extremely low value, or completely bypassing certain signal filtering modules.

[0023] The local server uploads monitoring data to the cloud computing center. The local server can use real-time streaming, for example, using a Kafka producer to stream high-precision monitoring data to the Kafka cluster in the cloud computing center. Lightweight compression and encryption can be performed on the data before transmission. Alternatively, the local server can cache the monitoring data in local storage, and periodically (e.g., hourly) package the cached data into encrypted files and upload them to the cloud computing center's cloud storage service (such as Amazon S3 or Alibaba Cloud OSS) via HTTPS.

[0024] The cloud computing center analyzes monitoring data to identify potential hidden faults. It can utilize more sophisticated signal processing algorithms, such as wavelet transform or Hilbert-Huang transform, to perform time-frequency analysis on the uploaded high-precision monitoring data to identify precise patterns of weak signal characteristics. Simultaneously, it can combine historical fault databases and expert knowledge systems to match and diagnose the identified features, thereby confirming potential hidden faults. Another approach is to deploy a fault diagnosis engine integrating multiple machine learning models in the cloud computing center. This engine can include support vector machines for classification and neural networks for pattern recognition. By extracting multi-dimensional features from high-precision monitoring data and inputting them into these models, the accuracy of identifying hidden faults can be improved.

[0025] For motors not instructed to be monitored, the local server performs routine data filtering. For motors not specifically monitored by the cloud computing center, the local server continues to use its preset, relatively lenient data filtering strategy. For example, maintaining a low sampling rate, enabling routine noise filtering and anomaly threshold judgment to reduce the burden of data processing and transmission. Another implementation is that the local server maintains a list of motor statuses. For motors marked as "routine monitoring" in the list, the data processing flow will follow the default energy-saving mode, that is, only uploading data that has undergone significant compression and preliminary feature extraction.

[0026] Based on the analysis of the monitoring data, the cloud computing center updates its instructions, and the local server adjusts its data filtering strategy accordingly. If the cloud computing center identifies a potential hidden fault, it may update its instructions, for example, instructing the local server to perform continuous monitoring of the motor for a longer period or adjusting monitoring parameters to focus on specific stages of the fault's development. If the fault is resolved, the cloud computing center may instruct the local server to revert to normal monitoring mode. These updated instructions are also sent to the local server via API or message queues, and the local server adjusts its data filtering strategy accordingly. For example, if a fault is confirmed, the local server may further increase the sampling rate and enable a more refined signal separation algorithm; if the fault is resolved, it will restore the sampling rate and filtering strategy to energy-saving mode.

[0027] The cloud-based energy-saving motor health data assessment method proposed in this application works by constructing a dynamic, intelligent, and energy-efficient closed-loop system for motor health monitoring and assessment. Traditional motor health assessment systems, when dealing with industrial environments where motor workloads change frequently and drastically, suffer from overly simplistic initial screening mechanisms on local servers. These systems are prone to misinterpreting weak signals indicating potential, hidden faults as normal background noise and filtering them out. This results in the cloud computing center failing to receive crucial early fault evidence, ultimately leading to production accidents.

[0028] To address the aforementioned issues, this application first receives and analyzes preliminary screening data uploaded from the local server via a cloud computing center. This step aims to leverage the powerful computing capabilities of the cloud to identify weak signal characteristics that may indicate potential hidden faults from the preliminary screening data, even if these signals might be overlooked in the local server's initial screening. Once the motor requiring attention and its weak signal characteristics are identified, the cloud computing center issues an instruction to the local server, directing it to monitor the motor's weak signal characteristics. This instruction is one of the key innovations of this application; it breaks away from the traditional fixed screening mode of local servers, enabling dynamic intervention of edge monitoring strategies by the cloud.

[0029] Upon receiving the instruction, the local server adjusts its data filtering strategy accordingly, specifically by increasing the accuracy of weak signal feature acquisition and pausing the filtering of weak signal features. This means the local server enters a "high-precision monitoring mode" to ensure that weak signals are captured completely and accurately, rather than simply filtered out. Subsequently, the local server uploads this high-precision monitoring data to the cloud computing center. Upon receiving this data, the cloud computing center performs further analysis to ultimately identify potential hidden faults.

[0030] Meanwhile, to maintain the system's energy efficiency, the local server still performs routine data filtering for motors not instructed to be monitored. This ensures that system resources are not wasted on unnecessary high-precision monitoring. Finally, the cloud computing center updates the instructions based on the analysis of the monitoring data, and the local server then adjusts its data filtering strategy according to the updated instructions. This feedback loop mechanism enables the entire system to self-optimize and adjust based on actual monitoring results, thereby achieving continuous, accurate, and energy-efficient assessment of motor health. Through this collaborative working mode between the cloud and local servers, this application effectively solves the problem of traditional systems being "invisible" to hidden faults, significantly improves the accuracy and reliability of motor health assessment, and optimizes the energy consumption of data transmission and processing.

[0031] This application further proposes that when the local server receives an instruction and enters a high-precision monitoring mode for the motor of interest, it further includes the following steps: The cloud computing center receives interference event reports uploaded by the local server. The interference event reports contain the characteristics of abnormal electromagnetic signals detected by the local server in daily operation or on-demand detailed inspection mode. The cloud computing center maintains an industrial environment interference map based on interference event reports; When the local server is instructed to enter the on-demand fine inspection mode to monitor the weak signal characteristics of the motors that need to be monitored, or when new strong interference is detected in the electromagnetic environment around the local server during the on-demand fine inspection mode, the local server generates an environmental interference snapshot and reports it to the cloud computing center. After receiving the environmental interference snapshot, the cloud computing center combines the industrial environmental interference map with the actual environmental information of the local server to generate a customized interference suppression configuration scheme. After receiving the customized interference suppression configuration scheme, the local server dynamically loads and executes the interference suppression strategy specified in the customized interference suppression configuration scheme. The interference suppression strategy includes filter parameter adjustment, signal separation algorithm selection, or sensor physical shielding suggestions. After executing the interference suppression strategy, the local server continuously monitors the effectiveness of the interference suppression strategy and generates an interference suppression effect evaluation report, and feeds back the processed data and the interference suppression effect evaluation report to the cloud computing center. Based on feedback, the cloud computing center optimizes the industrial environment interference map and interference strategy decision engine.

[0032] Specifically, interference event reporting refers to any abnormal electromagnetic signal characteristics detected by the local server during routine data screening or high-precision monitoring that may affect the acquisition and analysis of weak target signals. These abnormal signal characteristics may include, but are not limited to, sudden spike signals, broadband noise, and harmonic interference at specific frequencies. The purpose is to promptly report potential interference sources to the cloud computing center for centralized management and analysis.

[0033] The industrial environment interference map can be understood as a dynamically updated database that records the characteristics, frequency, intensity, location information, and historical suppression effects of various known interference sources in a specific industrial environment. Maintaining this map aims to provide comprehensive background knowledge and decision-making basis for subsequent interference identification and suppression.

[0034] In practical applications, an environmental interference snapshot is specifically a momentary sampling and feature extraction of the current electromagnetic environment performed by a local server when strong interference is detected. This sampling may include information such as the spectrum, time-domain waveform, intensity, duration, and location of the interference signal. Its purpose is to provide real-time, accurate interference information to assist the cloud computing center in generating targeted suppression solutions.

[0035] A customized interference suppression configuration scheme refers to a set of interference suppression strategies tailored to a local server by the cloud computing center based on real-time interference snapshots, historical interference maps, and the actual environmental information of the local server. This scheme can include various suppression methods, such as filter parameter adjustment (dynamically adjusting the cutoff frequency, bandwidth, or notch depth of digital or analog filters according to the frequency characteristics of the interference signal to filter out the interference signal); signal separation algorithm selection (selecting appropriate blind source separation, independent component analysis, or wavelet denoising algorithms based on the differences in characteristics between the interference and the target signal to separate the interference from the target signal); or sensor physical shielding recommendations (recommending physical shielding or adjusting the installation position of sensors in extreme interference conditions to reduce interference coupling). Its purpose is to provide efficient and flexible interference suppression methods to ensure the quality of weak signal acquisition.

[0036] After implementing interference suppression strategies, the local server continuously monitors their effectiveness. Specifically, this involves comparing indicators such as the signal-to-noise ratio, feature integrity, and clarity of fault indications of the target frequency band signals before and after processing to evaluate the effectiveness of the implemented strategies. The interference suppression effectiveness evaluation report then feeds these evaluation results and the processed data back to the cloud computing center. Its purpose is to provide empirical data for the cloud computing center to optimize the industrial environment interference map and interference strategy decision engine.

[0037] In some preferred embodiments, suppose a local server is performing high-precision monitoring of a critical motor to capture weak vibration signals caused by early bearing wear. During monitoring, the local server suddenly detects the startup of a nearby high-power frequency inverter, generating broadband electromagnetic interference whose frequency range overlaps with the harmonic frequencies of the target vibration signal. At this point, the local server generates an interference event report, including the inverter's startup time, the spectral characteristics and intensity of the interference signal, and uploads it to a cloud computing center. Upon receiving the report, the cloud computing center compares it with existing industrial environment interference maps, identifies the inverter as a known interference source, and updates its activity status.

[0038] Subsequently, when the local server is instructed to enter on-demand detailed inspection mode, or when strong interference from the frequency converter is detected again during the detailed inspection process, the local server immediately generates an environmental interference snapshot, recording in detail the real-time spectrum, intensity, and duration of the current interference, and reports it to the cloud computing center. The cloud computing center combines this real-time snapshot, historical data about the frequency converter in the industrial environmental interference map, and the current operating status of the local server to generate a customized interference suppression configuration scheme. This scheme may instruct the local server to: precisely adjust the notch filter frequency of the digital filter to the dominant frequency of the frequency converter interference, and select a wavelet transform-based signal separation algorithm to further extract the target vibration signal from the interfered signal.

[0039] After receiving the proposed solution, the local server dynamically loads and executes these strategies. During execution, the local server continuously monitors the quality of the processed vibration signal, such as calculating the improvement in signal-to-noise ratio, and generates an interference suppression effectiveness evaluation report. This report, along with the processed vibration data, is fed back to the cloud computing center. Based on this feedback, the cloud computing center further optimizes the interference suppression strategy parameters for the inverter in the industrial environment interference map and updates the interference strategy decision engine, enabling it to generate more effective suppression solutions faster and more accurately when encountering similar interference in the future. Through this series of steps, even in environments with strong interference, accurate capture and evaluation of weak motor fault signals can be ensured.

[0040] The steps for maintaining industrial environment interference maps based on interference event reports in the cloud computing center can be further refined as follows: The cloud computing center receives interference event reports uploaded from the local server; The cloud computing center performs pattern recognition on interference event reports to identify new interference signal patterns; The cloud computing center has identified the new interference signal pattern as a new source of interference. Based on the spatiotemporal correlation of the new interference sources, the cloud computing center classifies the new interference sources into temporary interference sources or mobile interference sources. The cloud computing center integrates the key features of new interference sources into the industrial environment interference map; The cloud computing center continuously monitors subsequent interference event reports of new interference sources and updates the status of new interference sources in the industrial environment interference map based on the monitoring results.

[0041] The cloud computing center receives interference event reports uploaded by local servers. These reports typically contain anomalous electromagnetic signal characteristics detected by the local servers during routine operation or on-demand detailed inspection. The cloud computing center performs pattern recognition on these interference event reports. The aim is to extract unknown patterns with specific frequencies, amplitudes, durations, or modulation characteristics from complex electromagnetic signal data, thereby identifying novel interference signal patterns. Specifically, pattern recognition can employ machine learning algorithms, such as cluster analysis, anomaly detection, or deep learning models, to distinguish known interference from potential novel interference.

[0042] Furthermore, once a new interference signal pattern is identified, the cloud computing center assigns it a unique identifier and initially marks it as a new interference source for subsequent tracking and management. The cloud computing center classifies new interference sources based on their spatiotemporal correlation. Spatiotemporal correlation refers to analyzing the frequency, duration, and propagation path of the new interference source at different times, in different local server regions, or in different geographical locations. For example, if an interference source appears only in a specific local server region within a certain time period, it may be classified as a temporary interference source; if the interference source moves between different local server regions, it may be classified as a mobile interference source. This classification helps in developing more targeted interference suppression strategies.

[0043] Subsequently, the cloud computing center integrates key characteristics of the novel interference source into the industrial environment interference map. These key characteristics include, but are not limited to, its frequency range, energy distribution, frequency of occurrence, duration, area of ​​influence, and its classification as a temporary or mobile interference source. These characteristics are structured and added to the industrial environment interference map to enrich its knowledge base. The cloud computing center continuously monitors subsequent interference event reports from the novel interference source, such as changes in its intensity, increases or decreases in its frequency of occurrence, or changes in its spatiotemporal correlation. Based on these monitoring results, the status of the novel interference source in the industrial environment interference map (e.g., active, inactive, intensity changes, etc.) is dynamically updated to ensure the real-time performance and accuracy of the map.

[0044] This application further proposes another method for optimizing industrial environment disturbance maps and disturbance strategy decision engines, which includes the following steps: The cloud computing center receives interference event reports uploaded from the local server; The cloud computing center performs real-time analysis of the electromagnetic signal characteristics in the interference event reports and compares the electromagnetic signal characteristics with known interference patterns in the industrial environment interference spectrum. Based on the comparison results, the cloud computing center marks electromagnetic signal features that match known interference patterns below a preset threshold as potential new types of interference. The cloud computing center extracts key parameters of potential new types of interference. Based on the key parameters of potential new types of interference, the cloud computing center analyzes the frequency and duration of the occurrence of potential new types of interference at different time points and in different local server regions to determine whether potential new types of interference have temporary or mobile characteristics. Based on the judgment results, the cloud computing center marked the potential new interference as an interference source to be verified and assigned a temporary identifier, and integrated the key characteristics and classification information of the interference source to be verified into the new interference area to be verified in the industrial environment interference map. The cloud computing center sends an interference verification command to the relevant local servers. The interference verification command requires special attention and high-priority reporting of the interference sources to be verified with temporary identifiers. The cloud computing center performs aggregated analysis and pattern confirmation on subsequent high-priority reports of the interference sources to be verified, and obtains the confirmed interference sources. The cloud computing center incorporates the identified interference sources into the industrial environment interference map and updates the interference strategy decision engine based on the characteristics of the identified interference sources.

[0045] Specifically, an interference event report can be understood as a detailed record of abnormal electromagnetic signal characteristics detected by a local server during routine operation or on-demand detailed inspection. Its purpose is to provide real-time information about the interference environment to the cloud computing center. Electromagnetic signal characteristics refer to quantifiable parameters such as the frequency, amplitude, phase, duration, and occurrence pattern of the interference signal. The industrial environment interference map is a dynamically updated database that stores historical and currently known electromagnetic interference patterns and their characteristics. Preset thresholds can be set according to actual application scenarios and sensitivity requirements for identifying new types of interference. For example, matching degrees can be defined based on signal correlation, spectral similarity, or feature vector distance. Potential new interference refers to signals that do not fully conform to existing interference patterns but exhibit certain abnormal characteristics. Its purpose is to identify emerging interference sources that may require further attention. Key parameter extraction may include, but is not limited to, center frequency, bandwidth, peak power, modulation type, and pulse repetition frequency, aiming to provide a basis for subsequent analysis and classification. Temporary interference sources may refer to interference that occurs briefly or randomly, while mobile interference sources may refer to interference sources whose location or range of influence changes, such as interference generated by mobile devices or temporary construction equipment. Interference sources to be verified refer to those initially identified as novel interferences but whose nature and impact have not yet been fully confirmed. Temporary identifiers are used for unique identification and tracking during the verification process. Interference verification instructions aim to guide local servers to perform more refined monitoring and reporting of specific signals within a specific area or time period to collect more data for verification. Aggregation analysis involves summarizing and statistically analyzing reports from multiple local servers or different time points to discover commonalities and patterns in potential novel interferences. Pattern confirmation involves using data analysis and expert experience to ultimately determine the nature, source, and scope of impact of the interference source to be verified and formally incorporate it into the industrial environment interference map.

[0046] As a specific implementation, suppose a local server, during its daily operation, has its sensors detect a persistent, narrowband anomalous electromagnetic signal with a frequency of approximately 150 kHz. When the cloud computing center performs real-time analysis of interference event reports, this signal characteristic is compared with known interference patterns in the industrial environment interference map. If the match is found to be below a preset threshold (e.g., correlation coefficient below 0.6), the cloud computing center marks the signal as a potential new type of interference. Subsequently, the system extracts key parameters of this potential new interference, such as its center frequency, bandwidth, and peak power, and analyzes its frequency and duration of occurrence in the local server area over the past 24 hours. If the signal is found to occur frequently within a specific time period and is limited to the local server area, it may be judged to have temporary or localized characteristics. Based on this judgment, the cloud computing center assigns a temporary identifier "NEW_INT_20231027_001" to this potential new interference and integrates its key characteristics and classification information into the new interference region to be verified in the industrial environment interference map.

[0047] To verify the nature of this potential new type of interference, the cloud computing center issued an interference verification command to the local server, requiring it to perform high-precision, unfiltered monitoring of signals in the 140kHz-160kHz frequency band over the next 48 hours and report detailed electromagnetic signal data hourly. Upon receiving the command, the local server adjusted its data filtering strategy, improving the acquisition accuracy for this frequency band and pausing filtering. After receiving subsequent high-priority reports from the local server, the cloud computing center aggregated and analyzed this data, finding that the signal characteristics remained stable at different time points and under different load conditions, and that its spectral characteristics highly matched the operating frequency of a newly installed high-frequency induction heating device nearby. Through pattern confirmation, the cloud computing center ultimately identified the signal as a "high-frequency induction heating interference source" and officially included it in the industrial environment interference map. Simultaneously, based on the characteristics of this interference source, the interference strategy decision engine was updated; for example, notch filter parameters for the 150kHz frequency band were generated for the local server to effectively suppress the interference in subsequent motor health data assessments.

[0048] Specifically, the steps for implementing the interference suppression strategy include: The local server performs spectral analysis on the weak signal characteristics of the motors of interest to obtain the frequency range and energy distribution of the weak signal characteristics; The local server adjusts the cutoff frequency or notch width of the filter based on the frequency range of the weak signal characteristics. The local server presets signal integrity protection parameters based on the energy distribution of weak signal characteristics. These parameters are used to limit the minimum retention strength or maximum allowable distortion of weak signals during signal separation. The signal separation algorithm adjusts the number of iterations or convergence conditions based on the signal integrity protection parameters. The local server performs a self-check on the data processed by the interference suppression strategy. The self-check includes comparing the relative changes and feature integrity of the target frequency band signal before and after processing, and obtaining the evaluation results of the impact of the interference suppression strategy on weak signals. The local server uploads the impact assessment results of the weak signal to the cloud computing center.

[0049] The local server performs spectral analysis on the weak signal characteristics of the motors of interest. This involves using signal processing techniques such as Fourier transform to convert the time-domain signal into a frequency-domain signal, thereby revealing the energy distribution of the weak signal characteristics at different frequencies. This allows for accurate acquisition of the frequency range and energy distribution of the weak signal characteristics, providing precise parameters for subsequent interference suppression.

[0050] Furthermore, the local server adjusts the cutoff frequency or notch width of the filter based on the frequency range of the weak signal characteristics. Specifically, when an interfering signal is detected to overlap with the weak signal characteristics within a specific frequency range, the cutoff frequency of the digital or analog filter can be dynamically adjusted to filter out the interfering signal while preserving the weak signal characteristics as much as possible. Alternatively, when the interfering signal manifests as narrowband interference at a specific frequency, the notch width and center frequency of the notch filter can be adjusted to precisely suppress the interference.

[0051] In addition, the local server presets signal integrity protection parameters based on the energy distribution of weak signal characteristics. Signal integrity protection parameters can be understood as thresholds set during signal separation to ensure that weak signal characteristics are not excessively weakened or distorted. For example, a minimum retention strength of weak signal characteristics can be set to ensure that they can still be effectively detected after separation; or a maximum permissible distortion level can be set to control the degree to which the signal separation algorithm modifies weak signal characteristics.

[0052] The signal separation algorithm adjusts its iteration count or convergence criteria based on the aforementioned signal integrity protection parameters. For example, when the signal integrity protection parameters require a higher degree of signal preservation, the signal separation algorithm can increase the number of iterations to separate the signal more finely, or relax the convergence criteria to avoid prematurely stopping the separation and causing the loss of weak signal characteristics.

[0053] After implementing the interference suppression strategy, the local server performs a self-check on the processed data. The purpose of this self-check is to evaluate the actual effectiveness of the interference suppression strategy, including comparing the relative changes in the target frequency band signal before and after processing to confirm whether the interference has been effectively suppressed, and assessing feature integrity to ensure that weak signal features are not mistakenly impaired. From this, the impact assessment results of the interference suppression strategy on weak signals can be obtained.

[0054] Finally, the local server uploads the impact assessment results of the weak signal to the cloud computing center so that the cloud computing center can further optimize and adjust the interference suppression strategy.

[0055] In some embodiments described above, after receiving a customized interference suppression configuration scheme, the local server dynamically loads and executes the customized interference suppression strategy, and performs a self-check on the data processed by the interference suppression strategy to evaluate the impact of the interference suppression strategy on weak signals. However, in actual industrial environments, the operating load of motors is dynamically changing, and this load change directly affects the signal characteristics generated by the motor. If the self-check process fails to fully consider the instantaneous changes in the motor's workload, it may lead to a deviation in the evaluation of the effectiveness of the interference suppression strategy. For example, signal fluctuations caused by load changes may be misjudged as an over- or under-performance of the interference suppression strategy, thereby affecting the optimization of subsequent strategies and the accurate identification of faults.

[0056] This application further proposes the following steps for a local server to perform a self-check on the data processed by the interference suppression strategy: The local server retrieves the workload data of the motors that need to be monitored. The local server predicts the normal fluctuation range of the target frequency band signal under the current load based on workload data; The local server compares the data processed by the interference suppression strategy with the normal fluctuation range to obtain an indication that the signal change exceeds the normal fluctuation range. The local server analyzes whether the signal changes that exceed the normal fluctuation range are consistent with the direction of parameter adjustment of the interference suppression strategy, based on the indication that the signal changes exceed the normal fluctuation range. Based on the analysis results, the local server determines whether the interference suppression strategy is over-acting or has other anomalies. Based on the judgment results, the local server continuously monitors the instantaneous changes in the load of the motors that require attention; When the load changes drastically, the local server pauses the evaluation of the relative changes in the target frequency band signal before and after processing, and focuses on the evaluation of the characteristic integrity of the target frequency band signal before and after processing. Once the load stabilizes, the local server resumes its assessment of the relative changes in the target frequency band signal before and after processing.

[0057] Specifically, the local server acquires the workload data of the motors it is interested in. This involves obtaining real-time parameters such as power consumption, speed, torque, or current from sensors connected to the motor or from the motor control system. These parameters reflect the motor's current operating status and load conditions. Based on this workload data, the local server predicts the normal fluctuation range of the target frequency band signal under the current load. This can be understood as using historical data, machine learning models, or preset physical models, combined with the current real-time load data, to estimate the expected performance range of weak signals within the target frequency band under the current load conditions, assuming no interference or effective suppression of interference. The purpose is to provide a dynamic benchmark consistent with actual operating conditions for subsequent signal comparisons. In practical applications, the local server compares the data processed by the interference suppression strategy with the normal fluctuation range to obtain indications of signal changes exceeding the normal fluctuation range. For example, statistical methods (such as Z-score, T-test) or setting thresholds can be used to determine whether the processed signal falls within the predicted normal fluctuation range. The purpose is to identify abnormal signal changes that may be caused by the interference suppression strategy.

[0058] Furthermore, based on indications that signal changes exceed the normal fluctuation range, the local server analyzes whether these changes are consistent with the parameter adjustment direction of the interference suppression strategy. For example, if the interference suppression strategy aims to reduce noise in a certain frequency band, but the processed data shows a significant increase in energy in that band, then an inconsistency exists. The purpose is to initially determine the effectiveness of the strategy and whether there is any adverse effect. Based on the analysis results, the local server then determines whether the interference suppression strategy is over-acting or exhibiting other anomalies. For instance, if the strategy adjustment leads to excessive suppression of the target weak signal itself, or introduces new abnormal signals, it is considered over-acting or abnormal. The goal is to ensure that the interference suppression strategy removes interference without compromising the integrity of the target weak signal.

[0059] Furthermore, the local server continuously monitors the instantaneous changes in the motor load that require attention. This involves real-time collection and analysis of load data to determine whether the motor load is stable or experiencing drastic fluctuations. The purpose is to provide dynamic contextual information for self-assessment. Specifically, when the load changes drastically, the local server pauses the evaluation of the relative changes in the target frequency band signal before and after processing, focusing instead on assessing the integrity of the signal characteristics. This means that when drastic load changes cause significant signal fluctuations, the comparison of relative changes may no longer be accurate. In this case, more attention is paid to whether the core characteristics such as the waveform and frequency components of weak signals are preserved. The purpose is to avoid misjudgments caused by load changes and ensure effective strategy evaluation even under complex operating conditions. Once the load stabilizes, the local server resumes the evaluation of the relative changes in the target frequency band signal before and after processing. This means that after the motor operation returns to a stable state, the comparison of relative changes is reactivated to obtain a more comprehensive evaluation of the strategy's effectiveness. The purpose is to conduct a more refined evaluation under appropriate conditions.

[0060] The steps by which a local server predicts the normal fluctuation range of a target frequency band signal under the current load, based on workload data, include: The local server performs real-time verification of the workload data of the motors that require attention. The local server identifies and removes transient outliers that exceed preset physical limits or short-term rate of change thresholds; The local server marks the workload data of motors that require attention as potential data drift if they appear continuously and change slowly but are outside the normal range. After real-time verification, identification, elimination and tagging, the local server starts cross-comparison between the workload data of the motors that need to be monitored and the load data of adjacent or similar motors. The local server corrects the workload data of motors that need attention due to deviations based on the cross-comparison results, so as to improve the accuracy of subsequent predictions. The local server dynamically adjusts the trust weights of the workload data of the motors that need attention, collected by different data acquisition sensors, based on the historical performance records of the data acquisition sensors. Based on trust weights, the local server weakens abnormal data samples in the workload data of motors that need attention, which are collected by low-trust data acquisition sensors. The local server predicts the normal fluctuation range of the target frequency band signal under the current load based on the processed workload data of the motors that need to be monitored.

[0061] Specifically, real-time verification refers to the immediate checking of workload data to ensure it conforms to basic physical laws and operational logic. For example, the motor load value should not exceed its rated maximum load, and there should be no drastic and physically unfounded changes within a short period. Preset physical limits can include motor design parameters, safe operating ranges, etc., while short-term change rate thresholds are used to limit the fluctuation range of data within a very short time, preventing instantaneous spikes or drops from being misjudged as real load changes. Instantaneous outliers refer to data points that suddenly appear within a very short time and significantly deviate from the normal range. These data points are usually caused by momentary sensor malfunctions, electromagnetic interference, or other transient external factors. Removing these outliers helps to clean up the data and prevent it from interfering with subsequent analysis. In practical applications, potential data drift refers to the phenomenon where workload data continuously deviates from the normal range over a period of time, but the rate of change is relatively slow. This may indicate a decline in sensor performance, calibration deviations, or long-term changes in motor operating status. Marking these outliers helps the system perform deeper diagnostics or trigger maintenance alarms.

[0062] Furthermore, cross-comparison refers to comparing the workload data of the motor currently of interest with the workload data of other adjacent motors or motors of the same type with similar operating characteristics on the same production line. This method can identify systematic biases or anomalies in the data of individual motors. For example, if the load data of all motors of the same type show a similar trend, but the data of a certain motor deviates significantly, then there may be a problem with that motor's data. Therefore, correction operations can include, but are not limited to, linear adjustments based on the comparison results, nonlinear mapping, or corrections using statistical methods to make the biased data closer to reality. In addition, the historical performance records of data acquisition sensors can include information such as sensor calibration dates, maintenance records, failure rates, accuracy levels, and performance under different environmental conditions. Trust weights are reliability scores assigned to different sensor data based on these historical records; data from high-trust sensors will have a larger weight in the prediction. Based on trust weights, weakening processing can involve smoothing low-trust data samples, downsampling, reducing their weight in the prediction model, or even temporarily ignoring their influence in extreme cases to reduce their negative impact on overall prediction accuracy.

[0063] This application further proposes a step for predicting the normal fluctuation range of a target frequency band signal under current load, which includes: The cloud computing center receives long-term motor operation data uploaded from the local server; The cloud computing center analyzes long-term motor operation data to identify motor aging trends, the influence of ambient temperature on signal characteristics, and signal change patterns caused by production process adjustments. Based on the analysis results, the cloud computing center adjusts the prediction parameters for predicting the normal fluctuation range of the target frequency band signal under the current load. The cloud computing center will send the adjusted prediction parameters to the local server; After receiving the adjusted prediction parameters, the local server updates the prediction parameters for predicting the normal fluctuation range of the target frequency band signal under the current load. The local server predicts the normal fluctuation range of the target frequency band signal under the current load based on the updated prediction parameters. The local server compares the prediction results with the actual monitoring data to obtain the prediction deviation; The local server uploads the prediction bias to the cloud computing center; Based on the prediction deviation, the cloud computing center optimizes the prediction parameters for the normal fluctuation range of the target frequency band signal under the current load.

[0064] Specifically, the cloud computing center is configured to receive long-term motor operating data uploaded from local servers. This long-term operating data can include historical load data, vibration data, temperature data, current and voltage data, and other relevant environmental parameters and production logs of the motor over several months or even years. The cloud computing center utilizes its powerful computing capabilities and storage resources to perform in-depth analysis of this massive amount of long-term operating data. The analysis aims to identify motor aging trends, such as by monitoring changes in energy decay or harmonic distortion of specific frequency components over time; identify the influence of ambient temperature on signal characteristics, such as by establishing a correlation model between ambient temperature and motor vibration spectrum characteristics; and identify signal change patterns caused by adjustments in production processes, such as by comparing motor operating data before and after adjustments to different production batches or process parameters.

[0065] Based on the above analysis, the cloud computing center can dynamically adjust the prediction parameters used to predict the normal fluctuation range of the target frequency band signal under the current load. These prediction parameters can be weights, thresholds, baseline values, or coefficients of a statistical model, etc. The adjusted prediction parameters are then sent to the corresponding local servers. Upon receiving these updated prediction parameters, the local servers will promptly update the model parameters used to predict the normal fluctuation range within their systems. Subsequently, based on these prediction parameters optimized and updated by the cloud computing center, the local servers will more accurately predict the normal fluctuation range of the target frequency band signal under the current load.

[0066] To continuously optimize prediction accuracy, the local server compares the prediction results with actual monitoring data, calculates the prediction deviation, and uploads this deviation to the cloud computing center. The cloud computing center then uses these feedback deviations to further optimize the corresponding prediction parameters, forming a closed-loop, continuously learning and improving prediction system.

[0067] As a specific implementation, suppose a local server monitors a critical production motor. Initially, the local server predicts the normal fluctuation range of the target frequency band signal based on short-term data it collects. However, as the motor operates for longer periods, its internal bearings gradually wear down, causing the baseline of the vibration signal at a specific frequency to slowly rise. Simultaneously, seasonal variations in the factory's ambient temperature also periodically affect the motor signal. Relying solely on the local server's local model may not accurately distinguish these slowly changing normal signal drifts from genuine fault signals.

[0068] At this point, the cloud computing center receives several years' worth of long-term operating data for the motor, including historical vibration, temperature, and load. Using time series analysis, the cloud computing center identifies aging-related trends and periodic fluctuation patterns related to ambient temperature in the motor's vibration signal. For example, regression analysis is used to establish a quantitative relationship between temperature and vibration amplitude. Based on these analysis results, the cloud computing center adjusts the aging compensation coefficient and temperature correction parameters in the local server's prediction model and sends these optimized prediction parameters back to the local server. After receiving and updating these parameters, the local server can automatically consider the motor's current aging level and the influence of ambient temperature when predicting the normal fluctuation range of the target frequency band signal under the current load, thus obtaining a more accurate normal fluctuation range. For example, if the actual monitored vibration signal is slightly higher than the original prediction range, a false alarm is avoided if the new prediction range (considering aging and temperature) can include it; conversely, if the signal still exceeds the new prediction range even after considering these factors, it can more accurately indicate potential faults. The local server feeds back the prediction bias to the cloud, and the cloud computing center further fine-tunes the prediction parameters based on these biases to ensure that the prediction model always maintains optimal performance.

[0069] The steps involved in analyzing long-term motor operation data by the cloud computing center include: The cloud computing center performs time series decomposition on long-term running data to separate trend terms and residual terms; The cloud computing center will use the decomposed trend items to identify motor aging trends; The cloud computing center performs outlier detection on the decomposed residual terms and identifies abnormal signals that exceed the set threshold. The cloud computing center compares the abnormal signals with the characteristics of known fault signals; The cloud computing center distinguishes between early, hidden aging signals and normal operating condition fluctuations based on the comparison results.

[0070] Time series decomposition of long-term operating data involves breaking down data collected during long-term motor operation, such as vibration, current, and temperature signals, into different components. Specifically, these components typically include trend terms, seasonal terms, and residual terms. Trend terms reflect the long-term direction or trend of data changes over time, such as a slow decline in motor performance or aging. Residual terms represent random fluctuations or noise in the data after removing trend and seasonal terms, and may contain weak signals of abnormal events or early faults. This decomposition allows for a clearer observation and analysis of the data's inherent structure.

[0071] Furthermore, using the decomposed trend terms to identify motor aging trends means determining whether motor performance is gradually declining by analyzing the slope, rate of change, or specific patterns of the trend terms. For example, if the trend terms show a sustained, slow increase in motor vibration levels or a sustained, slow decrease in efficiency, these can be identified as signs of motor aging.

[0072] Furthermore, outlier detection of the decomposed residuals, identifying anomalous signals exceeding a set threshold, involves searching for data points in the residuals that significantly deviate from the normal fluctuation range. These outliers may indicate sudden events, transient failures, or potential hidden faults. The set threshold can be based on statistical methods, such as the three-standard-deviation principle, or on the empirical range of normal fluctuations in historical data.

[0073] Specifically, comparing abnormal signals with known fault signal characteristics involves matching the frequency, amplitude, waveform, and other features of the detected abnormal signal with a pre-stored, expert-verified database of typical signal features for various motor faults (such as bearing wear, winding short circuits, and rotor bar breakage). This comparison process can employ pattern recognition algorithms, machine learning models, or rule-based expert systems.

[0074] Therefore, distinguishing between early, latent aging signals and normal operating condition fluctuations based on the comparison results means determining, through the above comparison, whether the abnormal signal is caused by early, latent aging within the motor (such as bearing micro-cracks, partial discharge in insulation, etc.) or simply by instantaneous operating condition fluctuations caused by factors such as load changes and ambient temperature fluctuations during normal motor operation. This distinction is crucial for avoiding false alarms and promptly identifying potential faults that truly require attention.

[0075] refer to Figure 2 A cloud-based energy-saving motor health data assessment system, applied to a cloud-based energy-saving motor health data assessment method, the system comprising: The cloud-based analysis module receives and analyzes the preliminary screening data uploaded from the local server to identify the weak signal characteristics of motors that require attention and the potential hidden faults in the motor indicators. The cloud command delivery module, based on the weak signal characteristics of the motors to be monitored for potential hidden faults, sends commands from the cloud computing center to the local server, instructing the monitoring of the weak signal characteristics of the motors to be monitored. The local strategy adjustment module, upon receiving instructions, adjusts the data filtering strategy of the local server for the motors that need attention. The adjustments include improving the acquisition accuracy of weak signal features and pausing the filtering of weak signal features. The local data upload module uploads monitoring data to the cloud computing center; The cloud-based confirmation module analyzes monitoring data to identify potential hidden faults. The local routine filtering module performs routine data filtering on motors that are not monitored by instructions. The cloud-based command update module updates commands based on the analysis results of the monitoring data, and the local server adjusts its data filtering strategy according to the updated commands.

[0076] Specifically, the cloud analytics module can be understood as a software or hardware component deployed in a cloud computing center. Its main function is to receive preliminary screening data from the local server and use preset analysis algorithms (such as machine learning models and statistical analysis tools) to perform in-depth analysis on this data to identify motors that may have potential hidden faults and extract their weak signal characteristics. This module aims to quickly locate anomalies from massive amounts of data through efficient data processing capabilities.

[0077] The cloud-based command delivery module is responsible for generating and sending specific monitoring commands to the local server based on the weak signal characteristics identified by the cloud-based analysis module. These commands detail the motors that need attention and the specific requirements for monitoring their weak signal characteristics, such as monitoring frequency and data acquisition parameters. This module ensures that cloud-based decisions are accurately and promptly transmitted to the local execution end.

[0078] In practical applications, the local policy adjustment module is deployed on a local server. Its function is to receive and parse instructions from the cloud computing center. Once an instruction is received, the module dynamically adjusts the data filtering strategy of the local server. Specifically, this includes improving the acquisition accuracy of specific weak signal features and temporarily stopping the filtering of these weak signal features to ensure that more refined and complete raw data can be captured at the local level.

[0079] The local data upload module is responsible for securely and reliably uploading high-precision monitoring data collected from the local server after strategy adjustments to the cloud computing center. This module ensures that local monitoring results can be promptly transmitted back to the cloud for further centralized analysis.

[0080] The cloud-based confirmation module is one of the core analytical components of the cloud computing center. After receiving monitoring data uploaded from the local server, it uses higher-level analytical models or expert systems to conduct in-depth analysis of this data to ultimately confirm whether there are any potential hidden faults. This module improves the accuracy of fault diagnosis through multi-dimensional and multi-level analysis.

[0081] The local routine screening module is responsible for continuously executing routine data screening strategies for motors that are not subject to high-precision monitoring as instructed by the cloud. This means that for most normally operating motors, the local server can still process data in an energy-efficient manner, avoiding unnecessary resource consumption.

[0082] The cloud-based command update module dynamically updates monitoring commands based on the analysis results of the monitoring data from the cloud-based confirmation module. These updated commands are then resent to the local server, guiding it to further adjust its data filtering strategy, forming a closed-loop optimization control process. This module ensures that the entire evaluation system can adaptively adjust and optimize according to actual conditions.

[0083] The content disclosed above is only a preferred and feasible embodiment of the present invention, and is not intended to limit the scope of protection of the present invention. Therefore, all equivalent technical changes made based on the content of the present invention specification and drawings are included within the scope of protection of the present invention. Furthermore, the elements therein can be updated as technology develops.

Claims

1. A cloud computing-based energy-saving motor health data evaluation method, characterized in that, The method comprises the following steps: The cloud computing center receives the preliminary screening data uploaded by the local server and analyzes to identify the motor that needs attention and the weak signal features of potential hidden faults indicated by the motor that needs attention; According to the weak signal features of potential hidden faults indicated by the motor that needs attention, the cloud computing center issues an instruction to the local server, which instructs to monitor the weak signal features of the motor that needs attention; After receiving the instruction, the local server adjusts the data screening strategy of the local server for the motor that needs attention, including improving the collection accuracy of the weak signal features and suspending the filtering of the weak signal features; The local server uploads the monitoring data to the cloud computing center; The cloud computing center analyzes the monitoring data to confirm the potential hidden faults; The local server performs regular data screening on the motors that are not instructed to monitor; The cloud computing center updates the instruction according to the analysis result of the monitoring data, and the local server adjusts the data screening strategy of the local server according to the updated instruction.

2. The cloud-computing-based energy-saving motor health data evaluation method according to claim 1, wherein When the local server receives the instruction and enters the high-precision monitoring mode for the motor that needs attention, the following steps are further included: The cloud computing center receives the interference event report uploaded by the local server, which contains abnormal electromagnetic signal features detected by the local server in daily operation or on-demand inspection mode; The cloud computing center maintains an industrial environment interference map according to the interference event report; When the local server is instructed to enter the on-demand inspection mode to monitor the weak signal features of the motor that needs attention, or detects new strong interference in the electromagnetic environment around the local server during the on-demand inspection mode, the local server generates an environment interference snapshot and reports it to the cloud computing center; After receiving the environment interference snapshot, the cloud computing center generates a customized interference suppression configuration scheme in combination with the industrial environment interference map and the actual environment information of the local server; After receiving the customized interference suppression configuration scheme, the local server dynamically loads and executes the interference suppression strategy specified in the customized interference suppression configuration scheme, which includes filter parameter adjustment, signal separation algorithm selection, or sensor physical shielding suggestion; After executing the interference suppression strategy, the local server continuously monitors the effect of the interference suppression strategy and generates an interference suppression effect evaluation report, and feeds back the processed data and the interference suppression effect evaluation report to the cloud computing center; The cloud computing center optimizes the industrial environment interference map and the interference strategy decision engine according to the feedback.

3. The cloud-computing-based energy-saving motor health data evaluation method of claim 2, wherein The steps of maintaining the industrial environment interference map by the cloud computing center include: The cloud computing center receives the interference event report uploaded by the local server; The cloud computing center performs pattern recognition on the interference event report to identify new interference signal patterns; The cloud computing center marks the new interference signal patterns as new interference sources; The cloud computing center classifies the new interference sources as temporary interference sources or mobile interference sources according to their spatio-temporal correlation; The cloud computing center integrates the key features of the new interference sources into the industrial environment interference map. The cloud computing center continuously monitors subsequent interference event reports of the new interference source and updates the status of the new interference source in the industrial environment interference map according to the monitoring results.

4. The cloud-computing-based energy-saving motor health data evaluation method of claim 2, wherein Another way to optimize the industrial environment interference map and the interference strategy decision engine includes the following steps: The cloud computing center receives the interference event reports uploaded by the local server; The cloud computing center performs real-time analysis on the electromagnetic signal features in the interference event reports and compares the electromagnetic signal features with known interference patterns in the industrial environment interference map; According to the comparison results, the cloud computing center marks the electromagnetic signal features with a matching degree lower than a preset threshold with the known interference patterns as potential new interference; The cloud computing center extracts the key parameters of the potential new interference; According to the key parameters of the potential new interference, the cloud computing center analyzes the frequency and duration of the potential new interference at different time points and in different local server regions to determine whether the potential new interference has temporary or mobile characteristics; According to the determination result, the cloud computing center marks the potential new interference as a to-be-verified interference source and assigns a temporary identifier to it, and integrates the key features and classification information of the to-be-verified interference source into the to-be-verified new interference region of the industrial environment interference map; The cloud computing center issues interference verification instructions to the relevant local servers, which require special attention and high-priority reporting of the to-be-verified interference source with the temporary identifier; The cloud computing center aggregates and analyzes subsequent high-priority reports of the to-be-verified interference source and confirms the confirmed interference source; The cloud computing center incorporates the confirmed interference source into the industrial environment interference map and updates the interference strategy decision engine according to the characteristics of the confirmed interference source.

5. The cloud-computing-based energy-saving motor health data evaluation method of claim 2, wherein The steps of executing the interference suppression strategy include: The local server performs frequency spectrum analysis on the weak signal features of the motor that needs attention to obtain the frequency range and energy distribution of the weak signal features; The local server adjusts the cutoff frequency or notch width of the filter according to the frequency range of the weak signal features; The local server presets signal integrity protection parameters according to the energy distribution of the weak signal features, which are used to limit the minimum reserved strength or maximum allowed distortion of the weak signal during signal separation; The signal separation algorithm adjusts the iteration number or convergence condition of the signal separation algorithm according to the signal integrity protection parameters; The local server performs self-checking on the data processed by the interference suppression strategy, which includes comparing the relative changes and feature integrity of the target frequency band signals before and after processing to obtain the influence evaluation result of the interference suppression strategy on the weak signal; The local server uploads the influence evaluation result of the weak signal to the cloud computing center.

6. The cloud-computing-based energy-saving motor health data evaluation method of claim 5, wherein, The steps of the local server performing self-checking on the data processed by the interference suppression strategy include: The local server obtains the workload data of the motor that needs attention; The local server predicts the normal fluctuation range of the target frequency band signal under the current load according to the workload data; The local server compares the data processed by the interference suppression strategy with the normal fluctuation range to obtain an indication that the signal change exceeds the normal fluctuation range; The local server analyzes whether the signal change beyond the normal fluctuation range is consistent with the parameter adjustment direction of the interference suppression strategy according to the indication that the signal change exceeds the normal fluctuation range; The local server judges whether the interference suppression strategy has over-reaction or other abnormalities according to the analysis result; According to the judgment result, the local server continuously monitors the instantaneous change of the motor load that needs attention; When the load changes sharply, the local server suspends the evaluation of the relative change of the target frequency band signal before and after processing, and focuses on the evaluation of the feature integrity of the target frequency band signal before and after processing; When the load stabilizes, the local server resumes the evaluation of the relative change of the target frequency band signal before and after processing.

7. The cloud-computing-based energy-saving motor health data evaluation method of claim 6, wherein, The step of predicting the normal fluctuation range of the target frequency band signal under the current load by the local server according to the working load data includes: The local server performs real-time verification on the working load data of the motor that needs attention; The local server identifies and eliminates instantaneous outliers that exceed the preset physical limit or short-term change rate threshold; The local server marks the working load data of the motor that needs attention which appears continuously and has a slow change trend but exceeds the normal range as potential data drift; After real-time verification, identification, elimination and marking, the local server starts cross comparison of the working load data of the motor that needs attention with the load data of adjacent or same type motors; The local server corrects the working load data of the motor that needs attention with deviation according to the cross comparison result to improve the accuracy of subsequent prediction; The local server dynamically adjusts the trust weight corresponding to the working load data of the motor that needs attention collected by different data collection sensors according to the historical performance record of the data collection sensors; Based on the trust weight, the local server weakens the data samples with abnormalities in the working load data of the motor that needs attention collected by the low-trust data collection sensor; The local server predicts the normal fluctuation range of the target frequency band signal under the current load according to the processed working load data of the motor that needs attention.

8. The cloud-computing-based energy-saving motor health data evaluation method of claim 7, wherein, The step of predicting the normal fluctuation range of the target frequency band signal under the current load includes: The cloud computing center receives the long-term running data of the motor uploaded by the local server; The cloud computing center analyzes the long-term running data of the motor, identifies the aging trend of the motor, the influence law of environmental temperature on signal characteristics, and the signal change mode caused by production process adjustment; The cloud computing center adjusts the prediction parameters for predicting the normal fluctuation range of the target frequency band signal under the current load according to the analysis result; The cloud computing center distributes the adjusted prediction parameters to the local server; After receiving the adjusted prediction parameters, the local server updates the prediction parameters for predicting the normal fluctuation range of the target frequency band signal under the current load by the local server; The local server predicts the normal fluctuation range of the target frequency band signal under the current load according to the updated prediction parameters; The local server compares the prediction result with the actual monitoring data to obtain the prediction deviation; The local server uploads the prediction deviation to the cloud computing center; The cloud computing center optimizes the prediction parameters for the normal fluctuation range of the target frequency band signal under the corresponding current load according to the prediction deviation.

9. The cloud-computing-based energy-saving motor health data evaluation method of claim 8, wherein, The steps of the cloud computing center analyzing the long-term operation data of the motor include: The cloud computing center performs time series decomposition on the long-term operation data to separate the trend item and the residual item; The cloud computing center uses the decomposed trend item to identify the motor aging trend; The cloud computing center performs anomaly point detection on the decomposed residual item to identify abnormal signals that exceed the set threshold; The cloud computing center compares the abnormal signals with known fault signal characteristics; The cloud computing center distinguishes early hidden aging signals from normal operating condition fluctuations based on the comparison results.

10. A cloud computing-based energy-saving motor health data evaluation system applied to the cloud computing-based energy-saving motor health data evaluation method of claim 1, characterized in that, The system includes: A cloud analysis module that receives the preliminary screening data uploaded by the local server and performs analysis to identify the motors that need attention and the weak signal characteristics of potential hidden faults indicated by the motors that need attention; A cloud instruction issuing module that, based on the weak signal characteristics of potential hidden faults indicated by the motors that need attention, issues instructions from the cloud computing center to the local server, instructing the monitoring of the weak signal characteristics of the motors that need attention; A local strategy adjustment module that, after receiving the instructions, adjusts the data screening strategy of the local server for the motors that need attention, including improving the collection accuracy of the weak signal characteristics and suspending the filtering of the weak signal characteristics; A local data uploading module that uploads the monitoring data to the cloud computing center; A cloud confirmation module that analyzes the monitoring data to confirm potential hidden faults; A local routine screening module that performs routine data screening on the motors that are not instructed to be monitored; A cloud instruction updating module that updates the instructions based on the analysis results of the monitoring data, and the local server adjusts the data screening strategy of the local server based on the updated instructions.

Images