Method and system for online analysis and early warning of asynchronous motor faults

Abstract

This invention relates to the field of intelligent monitoring technology, specifically disclosing a method and system for online analysis and early warning of asynchronous motor faults. The method includes acquiring multi-dimensional operating data of the asynchronous motor during operation; constructing an operating data sequence based on the multi-dimensional operating data; segmenting the operating data sequence based on preset time nodes to obtain stage vectors; constructing stage state functions for each corresponding stage based on the stage vectors; concatenating the stage state functions based on time sequence to obtain the state evolution path of the asynchronous motor; and determining and feeding back early warning information for the asynchronous motor based on the state evolution path. This invention acquires data through intelligent modules, extracts segmented data, and constructs a state evolution path containing multiple operating stages as the operating state characteristics of the asynchronous motor, rather than simply judging whether the data of a certain module at a certain moment reaches a threshold, thus providing sufficient advance warning.

Description

Technical Field

[0001] This invention relates to the field of intelligent monitoring technology, specifically to a method and system for online analysis and early warning of asynchronous motor faults. Background Technology

[0002] Asynchronous motors are widely used in industrial production and in fields such as energy, chemical industry, and metallurgy due to their simple structure, reliable operation, and low maintenance costs. They are especially used as the main drive device in key equipment such as pumps, compressors, and fans. During long-term operation, asynchronous motors are susceptible to factors such as power supply fluctuations, load changes, environmental conditions, and mechanical wear, which can gradually lead to potential faults such as winding aging, insulation deterioration, bearing wear, and rotor imbalance.

[0003] Existing asynchronous motor operation status monitoring and fault early warning technologies typically rely on single or limited operating parameters such as current, voltage, temperature, or vibration to perform threshold judgments or feature extraction on operating data to identify abnormal states. These methods mostly focus on judging the operating state at a certain moment or within a short time window, which is relatively simple. They are mostly threshold-based triggering early warning architectures with very low lead time. How to provide a multi-dimensional, full-cycle data monitoring and analysis architecture to identify potential anomalies in advance is the technical problem that this invention aims to solve. Summary of the Invention

[0004] The purpose of this invention is to provide an online analysis and early warning method and system for asynchronous motor faults, so as to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: An online fault analysis and early warning method for asynchronous motors, the method comprising: Acquire multidimensional operating data of the asynchronous motor during operation, and construct an operating data sequence based on the multidimensional operating data; The running data sequence is segmented based on preset time nodes to obtain stage vectors; the segmentation types include at least the startup stage, steady-state operation stage, and abnormal evolution stage, and the stage vectors include at least the startup vector, steady-state vector, and evolution vector; Based on the stage vectors, the corresponding stage state functions are constructed. The stage state functions are then concatenated based on the time sequence to obtain the state evolution path of the asynchronous motor. The early warning information of the asynchronous motor is determined and fed back based on the state evolution path; wherein, the determination process is a segmented comparison process, and each segment of the comparison process is a matrix feature comparison process.

[0006] As a further aspect of the present invention: the step of acquiring multi-dimensional operating data of the asynchronous motor during operation and constructing an operating data sequence based on the multi-dimensional operating data includes: The system acquires time-stamped operating data based on an intelligent module installed on the asynchronous motor, and normalizes the operating data based on its historical maximum and minimum values. Based on the time sequence, the normalized running data is statistically analyzed to obtain the time series corresponding to each intelligent module; Obtain the distance to the monitoring points of the intelligent modules, and cluster the intelligent modules based on the distance; The positional relationships of intelligent modules are determined by class, and data elements are created based on these relationships; the element numbers in the data elements correspond to the intelligent modules. The running data sequence is obtained by reading data from the data element in the data sequence.

[0007] As a further aspect of the present invention: the step of obtaining the distance between the monitoring points of the intelligent modules and clustering the intelligent modules based on the distance includes: Pair the intelligent modules in pairs, and for each pair of intelligent modules, query the corresponding two time series. Perform frequency domain transformation on the two time series to obtain a frequency domain array; Calculate the array distance of the frequency domain array as the distance between monitoring points, and cluster the intelligent modules based on the distance; The step of determining the positional relationships of intelligent modules by class and creating data elements based on the positional relationships includes: Randomly select an intelligent module from the unselected intelligent modules, read the type of intelligent modules corresponding to the selected intelligent module, and determine the sequence number based on the preset sorting rules. The process is repeated in a loop to obtain data elements. The step of reading data from the data sequence based on data elements to obtain the running data sequence includes: Time points are created based on a preset time step. At each time point, the corresponding time series is located sequentially according to the element order of the data elements, and the data closest to the time point is read to obtain the data element at that time. By statistically analyzing data elements in chronological order, a sequence of running data is obtained.

[0008] As a further aspect of the present invention: the step of acquiring multi-dimensional operating data of the asynchronous motor during operation and constructing an operating data sequence based on the multi-dimensional operating data further includes: For any given smart module, query its array distances to all other smart modules; The prediction accuracy of the current intelligent module is determined based on the array distance and the data acquisition frequency of other intelligent modules; Adjust the data acquisition frequency of the intelligent module based on the prediction accuracy; Each intelligent module contains a preset minimum frequency; each intelligent module contains a timed peak frequency adjustment command, which is used to set the data acquisition frequency of the intelligent module to the maximum value.

[0009] As a further aspect of the present invention: the step of segmenting the running data sequence based on preset time nodes to obtain a stage vector includes: The segmented intervals are determined based on preset identification information; the identification information includes at least the rate of change of vibration information; the segmented intervals include at least a start-up interval, a steady-state interval, and an evolution interval. Within the startup interval, data is extracted from the running data sequence to obtain a startup vector; the startup vector includes at least the startup peak current, startup duration, startup current change rate, voltage drop amplitude, and startup count; Within the steady-state range, data is extracted from the running data sequence to obtain a steady-state vector; the steady-state vector includes at least the steady-state average current, three-phase current imbalance, stator temperature average, temperature rise rate, zero-sequence current average, and vibration characteristic quantities; Within the evolution interval, data is extracted from the running data sequence to obtain an evolution vector; the evolution vector includes at least the zero-sequence current change rate, temperature rise change rate, current deviation, abnormal duration, and alarm trigger count.

[0010] As a further aspect of the present invention: the step of determining and feeding back the early warning information of the asynchronous motor based on the state evolution path includes: Read the fault evolution path sequentially from the preset information table; The state evolution path is compared with the read fault evolution path to determine the path difference. When the path difference is less than the preset difference threshold, the read fault evolution path will be used as the target fault evolution path of the asynchronous motor. Synchronously query and return the fault early warning information corresponding to the target fault evolution path; wherein, the fault evolution path and its fault early warning information are determined in the testing phase and stored in the information table in advance; The calculation process for the path difference is as follows: In the formula, Indicates path difference. and These represent the dimensions of the path. and The left and right endpoints of the preset time period, Indicating the state evolution path The element at that position, Indicating the fault evolution path The element at the specified position.

[0011] The present invention also provides an online fault analysis and early warning system for asynchronous motors, the system comprising: The data acquisition module is used to acquire multi-dimensional operating data of the asynchronous motor during operation and to construct an operating data sequence based on the multi-dimensional operating data. The sequence segmentation module is used to segment the running data sequence based on preset time nodes to obtain stage vectors; the segmentation types include at least the start-up stage, steady-state operation stage, and abnormal evolution stage, and the stage vectors include at least the start-up vector, steady-state vector, and evolution vector; The path generation module is used to construct the stage state function corresponding to each stage based on the stage vector, and to concatenate the stage state functions based on the time sequence to obtain the state evolution path of the asynchronous motor. The path application module is used to determine and feed back early warning information of the asynchronous motor based on the state evolution path; wherein, the determination process is a segmented comparison process, and each segment comparison process is a matrix feature comparison process.

[0012] As a further aspect of the present invention: the data acquisition module includes: The normalization processing unit is used to obtain time-stamped operating data based on the intelligent module installed on the asynchronous motor, and to normalize the operating data based on the historical maximum and minimum values ​​of the operating data. The sequential statistics unit is used to statistically analyze the normalized running data according to the time sequence to obtain the time series corresponding to each intelligent module. Clustering unit, used to obtain the distance of the monitoring points of the intelligent module, and to cluster the intelligent modules according to the distance; A data element creation unit is used to determine the positional relationship of intelligent modules by class and to create data elements according to the positional relationship; the element number in the data element corresponds to the intelligent module. The data element application unit is used to read data from the data sequence based on the data elements to obtain the running data sequence.

[0013] As a further aspect of the present invention: the sequence segmentation module includes: An interval determination unit is used to determine segmented intervals based on preset identification information; the identification information includes at least the rate of change of vibration information; the segmented intervals include at least a start-up interval, a steady-state interval, and an evolution interval; A startup vector generation unit is used to extract data from the running data sequence within the startup interval to obtain a startup vector; the startup vector includes at least the startup peak current, startup duration, startup current change rate, voltage drop amplitude, and startup count; A steady-state vector generation unit is used to extract data from the running data sequence within the steady-state interval to obtain a steady-state vector; the steady-state vector includes at least the steady-state average current, three-phase current imbalance, stator temperature average, temperature rise rate, zero-sequence current average, and vibration characteristic quantity. An evolution vector generation unit is used to extract data from the running data sequence within the evolution interval to obtain an evolution vector; the evolution vector includes at least the zero-sequence current change rate, temperature rise change rate, current deviation, abnormal duration, and alarm trigger count.

[0014] As a further aspect of the present invention: the path application module includes: The path reading unit is used to sequentially read the fault evolution path from a preset information table; The difference calculation unit is used to compare the state evolution path with the read fault evolution path to determine the path difference degree; The path selection unit is used to select the read fault evolution path as the target fault evolution path of the asynchronous motor when the path difference is less than the preset difference threshold. An information query and feedback unit is used to synchronously query and feedback the fault early warning information corresponding to the target fault evolution path; wherein, the fault evolution path and its fault early warning information are determined in the testing phase and stored in the information table in advance; The calculation process for the path difference is as follows: In the formula, Indicates path difference. and These represent the dimensions of the path. and The left and right endpoints of the preset time period, Indicating the state evolution path The element at that position, Indicating the fault evolution path The element at the specified position.

[0015] Compared with the prior art, the beneficial effects of the present invention are: the present invention acquires data through intelligent modules, extracts the data in segments, and constructs a state evolution path containing multiple operating stages as the operating state characteristics of the asynchronous motor. The time span of the data is very large, and it is not simply a matter of judging whether the data of a certain module at a certain moment has reached the threshold. The lead time is sufficient, and it can discover most of the risks that have not yet occurred but may occur. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention.

[0017] Figure 1 This is a flowchart of an online fault analysis and early warning method for asynchronous motors.

[0018] Figure 2 This is the first sub-flowchart of the method for online analysis and early warning of asynchronous motor faults.

[0019] Figure 3 This is the second sub-flowchart of the method for online analysis and early warning of asynchronous motor faults.

[0020] Figure 4 This is the third sub-flowchart of the method for online analysis and early warning of asynchronous motor faults.

[0021] Figure 5 This is a block diagram of the structure of an online fault analysis and early warning system for asynchronous motors. Detailed Implementation

[0022] To make the technical problems to be solved, the technical solutions, and the beneficial effects of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present invention and are not intended to limit the present invention.

[0023] Figure 1 The flowchart illustrates an online fault analysis and early warning method for asynchronous motors. In this embodiment of the invention, an online fault analysis and early warning method for asynchronous motors includes: Step S100: Obtain multi-dimensional operating data of the asynchronous motor during operation, and construct an operating data sequence based on the multi-dimensional operating data; The multidimensional operating data is a holistic, overarching concept representing the data that can be monitored during the operation of an asynchronous motor. In practical applications, common operating data includes: electrical parameters: current, voltage, active power, reactive power, and starting inrush current; mechanical parameters: vibration amplitude and vibration spectrum; thermal parameters: stator temperature, bearing temperature, and terminal temperature; insulation and grounding parameters: insulation resistance, zero-sequence current, and grounding current changes; and operating event data: trip records, alarm records, number of starts, and start-up time intervals. Furthermore, each parameter can be acquired from multiple points, resulting in a comprehensive data structure with a wide variety of data types. Additionally, each acquired operating data point contains time information; statistically analyzing the acquired operating data according to time sequence yields a data sequence, known as the operating data sequence.

[0024] Step S200: Segment the running data sequence based on preset time nodes to obtain stage vectors; The operational data sequence is actually a comprehensive data structure with a very large data volume and a very wide time span. The working process of an asynchronous motor is segmented, for example, the startup phase and the running phase are obviously different, and the data to be analyzed in different phases are also different. Therefore, it is necessary to extract some data from the operational data sequence for analysis. The extracted data is called the phase vector. In the technical solution of this invention, the segmentation type includes at least the startup phase, the steady-state running phase, and the abnormal evolution phase. The phase vector includes at least the startup vector, the steady-state vector, and the evolution vector. The meanings of different elements in different vectors are actually different. Among them, the startup phase and the steady-state phase are easy to understand. When the motor starts working, the data fluctuates greatly, which belongs to the startup phase. When the data runs smoothly, it is called the steady-state phase. The evolution phase refers to the operating phase in which the operating state parameters of the asynchronous motor have begun to continuously deviate from the steady-state characteristics and show a directional trend of change before the motor enters a significant fault or protection action. When the data runs no longer smoothly, it can be regarded as entering the evolution phase.

[0025] Step S300: Construct the stage state function corresponding to each stage based on the stage vector, and concatenate the stage state functions according to the time sequence to obtain the state evolution path of the asynchronous motor; The stage vector is essentially a matrix, with one dimension representing time and the other representing the data class. Data is then extracted from the stage vector using time as the independent variable to obtain the data set corresponding to each time point. This data set is then used as elements to fit a function, resulting in the stage state function for each stage. It's important to note that this process actually involves fitting each element in the data set. In other words, it involves sequentially reading the value of each element in the stage vector at different times, fitting the values, and obtaining the fitting function for that element. The order of the elements in the stage vector is preserved, and the fitting function is statistically analyzed to obtain the stage state function. The process of fitting the function from the values ​​is existing technology and will not be elaborated upon here.

[0026] Since the types of elements counted at different stages are different, and the actual devices and meanings corresponding to each function in the stage state function are also different at different stages, they need to be concatenated to obtain a sequence composed of stage state functions, called the state evolution path. Since the stage state function is actually a one-dimensional function array, the obtained state evolution path is a two-dimensional function array composed of multiple one-dimensional function arrays. In the technical solution of this invention, it is divided into only three stages, so one dimension of the dynamic evolution path is three. Of course, in practical applications, other stages can be added, as long as the architecture of the technical solution of this invention is applied, which is not complicated for those skilled in the art.

[0027] Step S400: Determine and feed back the early warning information of the asynchronous motor based on the state evolution path; wherein, the determination process is a segmented comparison process, and each segment of the comparison process is a matrix feature comparison process; The dynamic evolution path is equivalent to extracting core information from the data acquired by all intelligent modules. Identifying the dynamic evolution path can reveal the operating status of the asynchronous motor and generate early warning information. There are many specific identification processes. One can be threshold-based identification methods, such as generating early warning information when certain characteristics (values ​​or rates of change) of certain parameters reach a threshold. Another approach is comparison-based, which has higher accuracy. The same method is used to pre-acquire the state evolution path under certain fault conditions and then use it as a reference. Under actual operating conditions, the state evolution path within a preset time period is captured and compared with the reference to determine the degree of matching. When the degree of matching is high enough, the early warning information corresponding to the reference is read and fed back to the management.

[0028] It should be noted that since the state evolution path is actually a matrix, the comparison process can directly apply the matrix comparison process. The matrix elements are functions, and the comparison of functions can be done by normalizing the range of the functions and calculating the absolute value of the integral difference within the same time period (corresponding to the numerical difference). In general, we can directly establish the difference function of the two functions, and then add the absolute value to the difference function to obtain the absolute value function, and calculate the integral of the absolute value function over a period of time.

[0029] Figure 2 This is a first sub-flowchart of the method for online fault analysis and early warning of asynchronous motors. The step of acquiring multi-dimensional operating data of the asynchronous motor during operation and constructing an operating data sequence based on the multi-dimensional operating data includes: Step S101: Obtain time-stamped operating data based on the intelligent module installed on the asynchronous motor, and normalize the operating data based on the historical extreme values ​​of the operating data; Step S102: Based on the time sequence, statistically normalize the running data to obtain the time series corresponding to each intelligent module; Step S103: Obtain the distance of the monitoring points of the intelligent module, and cluster the intelligent modules according to the distance; Step S104: Determine the positional relationship of intelligent modules by class, and create data elements according to the positional relationship; the element number in the data element corresponds to the intelligent module; Step S105: Read data from the data sequence according to the data elements to obtain the running data sequence.

[0030] In one example of the technical solution of this invention, the process of acquiring operating data of an asynchronous motor is described. Operating data with time tags is acquired based on an intelligent module installed on the asynchronous motor. The intelligent module is a higher-level concept; it only requires data acquisition devices with acquisition and transmission functions, such as sensors and signal transceiver modules. The data acquired by these devices is collectively referred to as operating data. During data acquisition, the acquisition time is recorded synchronously, called a time tag. Because different data have different value ranges and significant differences in dimensions, processing is very inconvenient. Therefore, after acquiring the operating data, normalization processing is required. The normalization process uses historical extreme values. For example, the difference between the maximum value and the current value is calculated, then the difference between the maximum value and the minimum value is calculated, the absolute value of the two differences is calculated, and then the ratio is calculated to obtain the normalized value. The absolute value of the difference between the maximum value and the minimum value is used as the denominator.

[0031] Then, based on the normalized operational data processed according to the time sequence, a time series corresponding to each intelligent module is obtained. The time series is actually an array. Further, the distance between the monitoring points of the intelligent modules is obtained, and the intelligent modules are clustered according to the distance. Here, distance refers to spatial distance. Generally, the monitoring points corresponding to nearby intelligent modules have a certain correlation, such as vibration information and temperature of adjacent monitoring points. The purpose of clustering is to make the operational data of nearby intelligent modules more similar when statistically analyzing them. In the array, the serial numbers of nearby intelligent modules are similar. In summary, the data of intelligent modules of the same type are grouped and inserted into the array. The positional relationship of intelligent modules is determined by class as unit, and data elements are created according to the positional relationship. The data element is an array, and each element serial number corresponds to an intelligent module. The data element is used as a data statistical template to read data from the data sequence of each intelligent module to obtain the operational data sequence. Generally, a reading period is set to obtain the operational data sequence within the reading period.

[0032] As a preferred embodiment of the technical solution of the present invention, the step of obtaining the distance of the monitoring points of the intelligent module and clustering the intelligent modules according to the distance includes: Pair the intelligent modules in pairs, and for each pair of intelligent modules, query the corresponding two time series. Perform frequency domain transformation on the two time series to obtain a frequency domain array; Calculate the array distance of the frequency domain array as the distance between monitoring points, and cluster the intelligent modules based on the distance.

[0033] The above solution is a preferred embodiment that expands the concept of "distance." The distance in the original solution refers to spatial distance. In this invention, the relationship between spatial distance and data correlation is only potentially strong; in other words, the correlation is weak, and the resulting clustering results may lack practical significance. For example, a location with high temperature may not necessarily have high amplitude. Therefore, the above provides a data-based distance calculation method. This method involves comparing intelligent modules pairwise. The comparison method involves querying the time series of two intelligent modules. The time series consists of normalized data at different times. The two time series are then transformed into a frequency domain array, representing the frequency distribution and reflecting the periodic characteristics of the running data. Calculating the array distance of the frequency domain arrays measures the similarity in data changes between the two intelligent modules. At this point, the influence of spatial distance is no longer considered; the focus is directly on the data. Clustering of intelligent modules based on array distance shows that intelligent modules of the same type have similar periods of data change and higher correlation.

[0034] The step of determining the positional relationships of intelligent modules by class and creating data elements based on the positional relationships includes: Randomly select an intelligent module from the unselected intelligent modules, read the type of intelligent modules corresponding to the selected intelligent module, and determine the sequence number based on the preset sorting rules. The process is repeated in a loop to obtain data elements.

[0035] The above content describes the creation process of data elements. This process is not complicated and only requires setting serial numbers one by one. The simplest sorting rule is the positional order of the intelligent modules, such as continuously acquiring intelligent modules along a certain direction.

[0036] The step of reading data from the data sequence based on data elements to obtain the running data sequence includes: Time points are created based on a preset time step. At each time point, the corresponding time series is located sequentially according to the element order of the data elements, and the data closest to the time point is read to obtain the data element at that time. By statistically analyzing data elements in chronological order, a sequence of running data is obtained.

[0037] In one example of the technical solution of this invention, the process of acquiring the running data sequence is described. Time points are created based on a preset time step. Then, at each time point, the corresponding time series is located sequentially according to the element order of the data elements, and the data closest to the time point is read to obtain the data elements at that time. The function of this process is actually time domain registration. If the data acquisition frequency of different intelligent modules is different, then the number of elements in each time series is actually different. When statistical data is collected, the time scale is different. By setting time points in advance and reading the closest order sequentially based on the time points, it is equivalent to copying and filling the data in the time series with less data to obtain the running data sequence after time domain registration.

[0038] As a preferred embodiment of the technical solution of the present invention, the step of obtaining multi-dimensional operating data of the asynchronous motor during operation and constructing an operating data sequence based on the multi-dimensional operating data further includes: For any given smart module, query its array distances to all other smart modules; The prediction accuracy of the current intelligent module is determined based on the array distance and the data acquisition frequency of other intelligent modules; The data acquisition frequency of the intelligent module is adjusted based on the prediction accuracy.

[0039] In one example of the technical solution of this invention, the process of constructing the running data sequence is optimized. The optimization objective is the data acquisition frequency of each intelligent module. For any intelligent module, its array distance with other intelligent modules is queried. As mentioned above, the array distance represents the correlation of the running data of the intelligent module. Based on this, the prediction accuracy of the current intelligent module is determined according to the array distance and the data acquisition frequency of other intelligent modules. The data acquisition frequency of the intelligent module is adjusted based on the prediction accuracy. Each intelligent module has a preset minimum frequency to ensure that the frequency is not too low. Each intelligent module has a timed peak frequency adjustment command, which is used to set the data acquisition frequency of the intelligent module to the maximum value. In a conventional architecture, the data acquisition frequency is a fixed value, which is fixed to the maximum value. The above content actually provides a frequency reduction architecture.

[0040] Regarding the calculation process of prediction accuracy, one feasible approach is as follows: In the formula, Indicates prediction accuracy. This represents the total number of all other intelligent modules. Indicates the first Data acquisition frequency of other intelligent modules Indicates the first The array distance between other intelligent modules and the current intelligent module; the above means that if an intelligent module has a high correlation with the current intelligent module, then the higher its data acquisition frequency and the more data it acquires, the higher the accuracy of its predictions based on the data of the current intelligent module; regarding In practical applications, a constraint is usually added to the value, which is a distance condition, such as the total number of other intelligent modules whose data distance is less than a preset distance threshold.

[0041] Figure 3 This is the second sub-flowchart of the online fault analysis and early warning method for asynchronous motors. The step of segmenting the running data sequence based on preset time nodes to obtain a stage vector includes: Step S201: Determine the segmented intervals based on preset identification information; the identification information includes at least the rate of change of vibration information; the segmented intervals include at least the initiation interval, the steady-state interval, and the evolution interval; Step S202: Within the startup interval, extract data from the running data sequence to obtain a startup vector; the startup vector includes at least the startup peak current, startup duration, startup current change rate, voltage drop amplitude, and startup count; Step S203: Within the steady-state range, extract data from the running data sequence to obtain a steady-state vector; the steady-state vector includes at least the steady-state average current, three-phase current imbalance, stator temperature average, temperature rise rate, zero-sequence current average, and vibration characteristic quantity; Step S204: Within the evolution interval, extract data from the running data sequence to obtain an evolution vector; the evolution vector includes at least the zero-sequence current change rate, temperature rise change rate, current deviation, abnormal duration, and alarm trigger count.

[0042] In one example of the technical solution of this invention, the process of determining the stage vector is described. Segmented intervals are determined based on preset identification information; the identification information includes at least the rate of change of vibration information; a start-up interval, a steady-state interval, and an evolution interval are obtained. Different types of data are extracted from different intervals to obtain a start-up vector, a steady-state vector, and an evolution vector. The start-up vector includes at least the start-up peak current, start-up duration, start-up current change rate, voltage drop amplitude, and number of starts; the steady-state vector includes at least the steady-state average current, three-phase current imbalance, stator temperature average, temperature rise change rate, zero-sequence current average, and vibration characteristic quantities; the evolution vector includes at least the zero-sequence current change rate, temperature rise change rate, current deviation, abnormal duration, and number of alarm triggers.

[0043] In addition, for the interval segmentation process, one or more intelligent modules can be pre-selected as identification modules, generally vibration sensors. The vibration information they acquire is queried, the rate of change of vibration information (vibration frequency, etc.) is calculated, and the rate of change of vibration information is used as the segmentation standard to determine different stages. This is somewhat similar to the edge-jumping detection method. Vibration generated from 0 is regarded as the start-up stage. When the vibration is stable (the rate of change is close to 0), it is regarded as the steady-state stage. When the vibration becomes unstable again (the rate of change increases or decreases again), it is regarded as the evolution stage. It should be noted that this detection process is continuous, and the steady-state stage and the evolution stage can alternate. Therefore, in step S300, it is necessary to splice the stage state function based on the time sequence, that is, to splice different stages (several steady-state stages and evolution stages).

[0044] Figure 4 The third sub-flowchart of the online fault analysis and early warning method for asynchronous motors includes the following steps: determining and feeding back early warning information for the asynchronous motor based on the state evolution path. Step S401: Read the fault evolution path sequentially from the preset information table; Step S402: Compare the state evolution path with the read fault evolution path to determine the path difference. Step S403: When the path difference is less than the preset difference threshold, the read fault evolution path is taken as the target fault evolution path of the asynchronous motor. Step S404: Synchronously query and return the fault warning information corresponding to the target fault evolution path; wherein, the fault evolution path and its fault warning information are determined in the testing phase and stored in the information table in advance.

[0045] In one example of the technical solution of this invention, the process of determining the early warning information is described. The fault evolution path is read sequentially from a preset information table. The state evolution path is compared with the read fault evolution path to determine the path difference degree. When the path difference degree is less than the preset difference degree threshold, it means that the path difference degree is small enough. At this time, the read fault evolution path is taken as the target fault evolution path of the asynchronous motor. The fault early warning information corresponding to the target fault evolution path is queried synchronously as the feedback result. The above process is actually a traversal matching process. The two sides of the comparison in the traversal process are the feature data body provided by the technical solution of this invention, which is called the fault evolution path.

[0046] Specifically, the calculation process for the path difference is as follows: In the formula, Indicates path difference. and These represent the dimensions of the path. and The left and right endpoints of the preset time period, Indicating the state evolution path The element at that position, Indicating the fault evolution path The element at the specified position.

[0047] For the above calculation process, since the state evolution path is actually a matrix, the comparison process can directly apply the matrix comparison procedure. The matrix elements are functions, and the absolute value of the integral difference (corresponding to the numerical difference) within the same time period is calculated. Generally, a difference function of the two functions can be directly established, and then the absolute value is added to the difference function to obtain the absolute value function. The integral of the absolute value function over a time period is then calculated. and The actual meanings of these correspond to the total number of stages and the dimension of the stage vector, respectively.

[0048] Figure 5 This is a block diagram of the structure of an online fault analysis and early warning system for asynchronous motors. In this embodiment of the invention, an online fault analysis and early warning system for asynchronous motors is provided, the system 10 comprising: Data acquisition module 11 is used to acquire multi-dimensional operating data of the asynchronous motor during operation and construct an operating data sequence based on the multi-dimensional operating data; The sequence segmentation module 12 is used to segment the running data sequence based on preset time nodes to obtain stage vectors; the segmentation types include at least the start-up stage, steady-state operation stage and abnormal evolution stage, and the stage vectors include at least the start-up vector, steady-state vector and evolution vector; The path generation module 13 is used to construct the stage state function corresponding to each stage based on the stage vector, and to splice the stage state function according to the time sequence to obtain the state evolution path of the asynchronous motor. The path application module 14 is used to determine and feed back early warning information of the asynchronous motor based on the state evolution path; wherein, the determination process is a segmented comparison process, and each segment comparison process is a matrix feature comparison process.

[0049] Furthermore, the data acquisition module 11 includes: The normalization processing unit is used to obtain time-stamped operating data based on the intelligent module installed on the asynchronous motor, and to normalize the operating data based on the historical maximum and minimum values ​​of the operating data. The sequential statistics unit is used to statistically analyze the normalized running data according to the time sequence to obtain the time series corresponding to each intelligent module. Clustering unit, used to obtain the distance of the monitoring points of the intelligent module, and to cluster the intelligent modules according to the distance; A data element creation unit is used to determine the positional relationship of intelligent modules by class and to create data elements according to the positional relationship; the element number in the data element corresponds to the intelligent module. The data element application unit is used to read data from the data sequence based on the data elements to obtain the running data sequence.

[0050] Specifically, the sequence segmentation module 12 includes: An interval determination unit is used to determine segmented intervals based on preset identification information; the identification information includes at least the rate of change of vibration information; the segmented intervals include at least a start-up interval, a steady-state interval, and an evolution interval; A startup vector generation unit is used to extract data from the running data sequence within the startup interval to obtain a startup vector; the startup vector includes at least the startup peak current, startup duration, startup current change rate, voltage drop amplitude, and startup count; A steady-state vector generation unit is used to extract data from the running data sequence within the steady-state interval to obtain a steady-state vector; the steady-state vector includes at least the steady-state average current, three-phase current imbalance, stator temperature average, temperature rise rate, zero-sequence current average, and vibration characteristic quantity. An evolution vector generation unit is used to extract data from the running data sequence within the evolution interval to obtain an evolution vector; the evolution vector includes at least the zero-sequence current change rate, temperature rise change rate, current deviation, abnormal duration, and alarm trigger count.

[0051] Furthermore, the path application module 14 includes: The path reading unit is used to sequentially read the fault evolution path from a preset information table; The difference calculation unit is used to compare the state evolution path with the read fault evolution path to determine the path difference degree; The path selection unit is used to select the read fault evolution path as the target fault evolution path of the asynchronous motor when the path difference is less than the preset difference threshold. An information query and feedback unit is used to synchronously query and feedback the fault early warning information corresponding to the target fault evolution path; wherein, the fault evolution path and its fault early warning information are determined in the testing phase and stored in the information table in advance; The calculation process for the path difference is as follows: In the formula, Indicates path difference. and These represent the dimensions of the path. and The left and right endpoints of the preset time period, Indicating the state evolution path The element at that position, Indicating the fault evolution path The element at the specified position.

[0052] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for online analysis and early warning of asynchronous motor faults, characterized in that, The method includes: Acquire multidimensional operating data of the asynchronous motor during operation, and construct an operating data sequence based on the multidimensional operating data; The running data sequence is segmented based on preset time nodes to obtain stage vectors; the segmentation types include at least the startup stage, steady-state operation stage, and abnormal evolution stage, and the stage vectors include at least the startup vector, steady-state vector, and evolution vector; Based on the stage vectors, the corresponding stage state functions are constructed. The stage state functions are then concatenated based on the time sequence to obtain the state evolution path of the asynchronous motor. The early warning information of the asynchronous motor is determined and fed back based on the state evolution path; wherein, the determination process is a segmented comparison process, and each segment of the comparison process is a matrix feature comparison process.

2. The method of online analysis and early warning of asynchronous motor faults according to claim 1, characterized in that, The steps of acquiring multi-dimensional operating data of the asynchronous motor during operation and constructing an operating data sequence based on the multi-dimensional operating data include: The system acquires time-stamped operating data based on an intelligent module installed on the asynchronous motor, and normalizes the operating data based on its historical maximum and minimum values. Based on the time sequence, the normalized running data is statistically analyzed to obtain the time series corresponding to each intelligent module; Obtain the distance to the monitoring points of the intelligent modules, and cluster the intelligent modules based on the distance; The positional relationship of intelligent modules is determined by class, and data elements are created based on the positional relationship; the element number in the data element corresponds to the intelligent module. The running data sequence is obtained by reading data from the data element in the data sequence.

3. The method of online fault analysis and early warning of asynchronous motor according to claim 2, characterized in that, The step of obtaining the distance to the monitoring points of the intelligent modules and clustering the intelligent modules based on the distance includes: Pair the intelligent modules in pairs, and for each pair of intelligent modules, query the corresponding two time series. Perform frequency domain transformation on the two time series to obtain a frequency domain array; Calculate the array distance of the frequency domain array as the distance between monitoring points, and cluster the intelligent modules based on the distance; The step of determining the positional relationship of intelligent modules by class and creating data elements based on the positional relationship includes: Randomly select an intelligent module from the unselected intelligent modules, read the type of intelligent modules corresponding to the selected intelligent module, and determine the sequence number based on the preset sorting rules. The process is repeated in a loop to obtain data elements. The step of reading data from the data sequence based on data elements to obtain the running data sequence includes: Time points are created based on a preset time step. At each time point, the corresponding time series is located sequentially according to the element order of the data elements, and the data closest to the time point is read to obtain the data element at that time. By statistically analyzing data elements in chronological order, a sequence of running data is obtained.

4. The method for online fault analysis and early warning of asynchronous motors according to claim 3, characterized in that, The step of acquiring multi-dimensional operating data of the asynchronous motor during operation and constructing an operating data sequence based on the multi-dimensional operating data further includes: For any given smart module, query its array distances to all other smart modules; The prediction accuracy of the current intelligent module is determined based on the array distance and the data acquisition frequency of other intelligent modules; Adjust the data acquisition frequency of the intelligent module based on the prediction accuracy; Each intelligent module contains a preset minimum frequency; each intelligent module contains a timed peak frequency adjustment command, which is used to set the data acquisition frequency of the intelligent module to the maximum value.

5. The method for online analysis and early warning of asynchronous motor faults according to claim 1, characterized in that, The step of segmenting the running data sequence based on preset time nodes to obtain the stage vector includes: The segmented intervals are determined based on preset identification information; the identification information includes at least the rate of change of vibration information; the segmented intervals include at least a start-up interval, a steady-state interval, and an evolution interval. Within the startup interval, data is extracted from the running data sequence to obtain a startup vector; the startup vector includes at least the startup peak current, startup duration, startup current change rate, voltage drop amplitude, and startup count; Within the steady-state range, data is extracted from the running data sequence to obtain a steady-state vector; the steady-state vector includes at least the steady-state average current, three-phase current imbalance, stator temperature average, temperature rise rate, zero-sequence current average, and vibration characteristic quantities; Within the evolution interval, data is extracted from the running data sequence to obtain an evolution vector; the evolution vector includes at least the zero-sequence current change rate, temperature rise change rate, current deviation, abnormal duration, and alarm trigger count.

6. The method for online analysis and early warning of asynchronous motor faults according to claim 1, characterized in that, The step of determining and feeding back the early warning information of the asynchronous motor based on the state evolution path includes: Read the fault evolution path sequentially from the preset information table; The state evolution path is compared with the read fault evolution path to determine the path difference. When the path difference is less than the preset difference threshold, the read fault evolution path will be used as the target fault evolution path of the asynchronous motor. Synchronously query and return the fault early warning information corresponding to the target fault evolution path; wherein, the fault evolution path and its fault early warning information are determined in the testing phase and stored in the information table in advance; The calculation process for the path difference is as follows: In the formula, Indicates path difference. and These represent the dimensions of the path. and The left and right endpoints of the preset time period, Indicating the state evolution path The element at that position, Indicating the fault evolution path The element at the specified position.

7. An online fault analysis and early warning system for asynchronous motors, characterized in that, The system includes: The data acquisition module is used to acquire multi-dimensional operating data of the asynchronous motor during operation and to construct an operating data sequence based on the multi-dimensional operating data. The sequence segmentation module is used to segment the running data sequence based on preset time nodes to obtain stage vectors; the segmentation types include at least the start-up stage, steady-state operation stage, and abnormal evolution stage, and the stage vectors include at least the start-up vector, steady-state vector, and evolution vector; The path generation module is used to construct the stage state function corresponding to each stage based on the stage vector, and to concatenate the stage state functions according to the time sequence to obtain the state evolution path of the asynchronous motor. The path application module is used to determine and feed back early warning information of the asynchronous motor based on the state evolution path; wherein, the determination process is a segmented comparison process, and each segment comparison process is a matrix feature comparison process.

8. The online fault analysis and early warning system for asynchronous motors according to claim 7, characterized in that, The data acquisition module includes: The normalization processing unit is used to obtain time-stamped operating data based on the intelligent module installed on the asynchronous motor, and to normalize the operating data based on the historical maximum and minimum values ​​of the operating data. The sequential statistics unit is used to statistically analyze the normalized running data according to the time sequence to obtain the time series corresponding to each intelligent module. Clustering unit, used to obtain the distance of the monitoring points of the intelligent module, and to cluster the intelligent modules according to the distance; A data element creation unit is used to determine the positional relationship of intelligent modules by class and to create data elements according to the positional relationship; the element number in the data element corresponds to the intelligent module. The data element application unit is used to read data from the data sequence based on the data elements to obtain the running data sequence.

9. The online fault analysis and early warning system for asynchronous motors according to claim 7, characterized in that, The sequence segmentation module includes: An interval determination unit is used to determine segmented intervals based on preset identification information; the identification information includes at least the rate of change of vibration information; the segmented intervals include at least a start-up interval, a steady-state interval, and an evolution interval; A startup vector generation unit is used to extract data from the running data sequence within the startup interval to obtain a startup vector; the startup vector includes at least the startup peak current, startup duration, startup current change rate, voltage drop amplitude, and startup count; A steady-state vector generation unit is used to extract data from the running data sequence within the steady-state interval to obtain a steady-state vector; the steady-state vector includes at least the steady-state average current, three-phase current imbalance, stator temperature average, temperature rise rate, zero-sequence current average, and vibration characteristic quantity. An evolution vector generation unit is used to extract data from the running data sequence within the evolution interval to obtain an evolution vector; the evolution vector includes at least the zero-sequence current change rate, temperature rise change rate, current deviation, abnormal duration, and alarm trigger count.

10. The online fault analysis and early warning system for asynchronous motors according to claim 7, characterized in that, The path application module includes: The path reading unit is used to sequentially read the fault evolution path from a preset information table; The difference calculation unit is used to compare the state evolution path with the read fault evolution path to determine the path difference degree; The path selection unit is used to select the read fault evolution path as the target fault evolution path of the asynchronous motor when the path difference is less than the preset difference threshold. An information query and feedback unit is used to synchronously query and feedback the fault early warning information corresponding to the target fault evolution path; wherein, the fault evolution path and its fault early warning information are determined in the testing phase and stored in the information table in advance; The calculation process for the path difference is as follows: In the formula, Indicates path difference. and These represent the dimensions of the path. and The left and right endpoints of the preset time period, Indicating the state evolution path The element at that position, Indicating the fault evolution path The element at the specified position.

Images