Operating Data Analysis Method Based on Motor Control

Abstract

The present invention discloses an operation data analysis method based on motor control, comprising obtaining current motor sensor streaming information, segmenting it, performing characterization information mining on the sensor streaming information segments, obtaining an operation state description vector of the current motor sensor streaming information, identifying a pending motor operation state sequence corresponding to the current motor sensor streaming information, determining at least one state-significant data point in the current motor sensor streaming information, determining at least one positioning interval of a state indication information flow in the current motor sensor streaming information based on the state-significant data point, extracting the operation state description vector according to the positioning interval, obtaining a target operation state description vector for each state indication information flow, optimizing the pending motor operation state sequence based on the target operation state description vector, and obtaining a final motor operation state sequence. The method improves data detection accuracy and facilitates accurate motor control.

Description

Technical Field

[0001] The present invention belongs to the technical field of motor operation data processing, and in particular relates to an operation data analysis method based on motor control. Background Art

[0002] In industrial production systems, motors are key power output devices. Their stable and efficient operation plays a decisive role in the smooth progress of the entire production process. Accurately monitoring the motor's operating status and implementing effective control can significantly reduce equipment failure rates, improve production efficiency, and reduce energy consumption and maintenance costs.

[0003] Existing motor operation status monitoring and control technologies have many limitations:

[0004] 1) Single Data Dimension: Some monitoring methods rely solely on a single type of sensor data, such as analyzing current or voltage data. Motor operation is a complex process, and a single data point cannot fully reflect the motor's actual operating status. For example, in the early stages of a motor bearing failure, the current and voltage may not change significantly, but the temperature and speed may already be fluctuating abnormally. Relying solely on electrical parameter monitoring will miss such a fault.

[0005] 2) Weak emergency handling capabilities: Traditional technologies struggle to accurately grasp the evolution of motor operating conditions when handling sudden motor conditions. When a motor suddenly overloads or experiences abnormal speed, traditional methods are unable to promptly and accurately determine the fault's onset, progression, and end. This results in delayed control decisions and an inability to effectively prevent equipment damage.

[0006] 3) Insufficient data feature mining: When processing sensor data, there is a lack of in-depth mining of data features, resulting in low accuracy and reliability of monitoring results, making it difficult to meet the strict requirements of complex industrial environments for motor operating status monitoring and control. Summary of the Invention

[0007] The purpose of the present invention is to provide an operation data analysis method based on motor control to solve the above technical problems.

[0008] In order to solve the above technical problems, this solution is implemented as follows:

[0009] A method for analyzing operating data based on motor control comprises the following steps:

[0010] Step S1: Acquire motor sensor streaming information and segment it: Based on multiple sensors arranged on the motor, collect motor operation data in real time to form continuous motor sensor streaming information; segment the continuous motor sensor streaming information to obtain multiple sensor streaming information segments.

[0011] Segmented processing is to convert continuous streaming data into small blocks of data that are easier to analyze and process, so as to better capture the changes in the motor's operating status in different time periods and provide a basis for subsequent operating status description and fault diagnosis.

[0012] Step S2: Mining the motor operating state description vector and identifying the pending sequence: performing characterization information mining on each sensor streaming information segment to obtain the operating state description vector of the motor sensor streaming information; based on the operating state description vector, identifying the pending motor operating state sequence corresponding to the motor sensor streaming information.

[0013] Characterization information mining is to extract key features that can reflect the operating status of the motor from each sensor streaming information segment, while the identification of the operating status sequence is to preliminarily determine the operating status of the motor in different time periods.

[0014] Step S3: Determine significant status data points and positioning intervals: Within the sequence of pending motor operating states, search for sensor stream information segments indicating abnormal motor operating states and identify significant status data points within these segments. Based on these significant status data points, the corresponding positioning intervals of the state indication information stream are determined within the motor sensor stream. By determining the positioning intervals, the key locations of sudden changes in the motor operating state can be more accurately located.

[0015] Step S4: Determine and extract the target running state description vector: extract the running state sub-description vectors of all data points in the positioning interval from the running state description vector to obtain the target running state description vector of the state indication information flow.

[0016] Step S5: Optimizing the pending sequence to obtain the final sequence: Based on the target operating state description vector, the pending motor operating state sequence is optimized to obtain the final motor operating state sequence corresponding to the current motor sensor streaming information.

[0017] The optimization improves the accuracy and reliability of the motor operating state sequence, making it closer to the motor's actual operating state. The resulting motor operating state sequence more accurately reflects the evolution of the motor's operating state, providing an important basis for motor control and maintenance.

[0018] Step S6: Based on the final motor operating state sequence, timely control measures are taken for abnormal states during subsequent motor operation to achieve precise control and ensure that the motor resumes normal operation.

[0019] Further optimization, in step S1, motor operation data is collected, including current data, voltage data, temperature data and speed data, and the motor sensor streaming information contains multiple data points, each data point contains a timestamp and corresponding four-dimensional feature data.

[0020] Further optimization, in step S1, the sliding window technology is used to cut the continuous motor sensor stream information into overlapping or non-overlapping sensor stream information segments S j , j ∈[1,N], N means dividing the data stream into N sensor stream information segments.

[0021] Further optimization, the step S2 specifically includes:

[0022] Step S2.1, single segment feature extraction: for each sensor stream information segment S j , extract the time series features through the state monitoring algorithm and generate the initial operation state description vector V j =[ v j1 , v j2 , v j3 , v j4 ], v ji Indicates the j The first segment of the sensor stream information i Operational status characteristic values of sensor-like monitoring data; i ∈[1, 4], representing current, voltage, temperature and speed respectively.

[0023] Step S2.2, global state splicing: Splice the initial operating state description vectors of all sensor stream information segments in chronological order to obtain the global operating state description vector V=[V1, V2,…, V N ].

[0024] Step S2.3, initial state sequence identification: determine the motor operating state corresponding to the initial operating state description vector of each sensor stream information segment, thereby obtaining the pending motor operating state sequence Q corresponding to the global operating state description vector V = [ q 1, q 2,…, q N ],in q j For the j The initial operating state description vector of each sensor streaming information segment corresponds to the motor operating state, such as "normal operation", "overload operation" and "fault operation". This sequence is a preliminary inference of the motor operating state and provides a basis for subsequent accurate analysis.

[0025] Further optimization, the step S3 specifically includes:

[0026] Step S3.1, detect sudden working conditions and identify significant data points: for the sensor flow information segment with abnormal motor operating state in the motor operating state sequence to be determined, calculate the operating state descriptor vector of each data point, and estimate the possibility of the motor instantaneous characteristics corresponding to these data points based on the operating state descriptor vector, compare the possibility of each data point with the preset threshold, and identify the data points that exceed the preset threshold. d k Record it as a state-significant data point and record the mutation location corresponding to the data point idx ( k ).

[0027] Step S3.2, forming mutation condition groups and locating intervals: classify all state significant data points into G mutation condition groups; determine the starting mutation corresponding to each mutation condition group s g and end mutation e g , and determine the starting mutations in the motor sensing flow information respectively s g and end mutation e g The corresponding data point will be the starting mutation data point position idx ( s g ) Push forward ΔL1 points and end the mutation data point position idx ( e g ) is extended by ΔL2 points to form the target data point coverage range of this sudden working condition group [ idx ( s g )-ΔL1, idx ( e g )+ΔL2], recorded as the positioning interval of the state indication information flow corresponding to the sudden working condition group; the same method is used to obtain the positioning interval of the state indication information flow corresponding to all sudden working condition groups.

[0028] Further optimization, in step S4, for each positioning interval, [ idx ( s g )-ΔL1, idx ( e g The data points contained in )+ΔL2] are classified as the candidate sensor monitoring data point set, and the operating state descriptor vectors of all data points in the candidate sensor monitoring data point set constitute the target operating state description vector V g .

[0029] Further optimization, step S5 includes:

[0030] Step S5.1, determine the motor operating state corresponding to each operating state description sub-vector in the target operating state description vector, and form a target motor operating state sub-sequence q corresponding to the positioning interval g ′.

[0031] Step S5.2: Using the target motor operating state subsequence q g ′, replace the motor running state corresponding to the positioning area in the sensor stream information segment with abnormal running state in the pending motor running state sequence Q, and obtain the final motor running state sequence Q corresponding to the current motor sensor stream information = [ q 1, q 2,…,q g ′,…, q N ].

[0032] Further optimization, in step S2.1, for the sensor stream information segment S j ={x1,x2,......,x n} to mine the representation information and obtain the initial operation state description vector corresponding to each sensor stream information segment, where the data point x t =[ I ( t ),U( t ),T( t ),w( t )],include t The current, voltage, temperature and speed of the motor at all times; specifically including:

[0033] Step S2.1.1, data preprocessing, data set construction: collect a large amount of motor sensor streaming information in advance, and divide it into several sensor streaming information segments as samples.

[0034] Normalize different types of sensor data in the sensor stream information segment so that each data feature has a similar scale range. For example, normalize current, voltage, temperature, and speed data so that their values fall within the range [0, 1] or (-1, 1). This helps improve the training efficiency and stability of the RNN network.

[0035] The large amount of preprocessed sensor streaming information segment data is divided into training set, validation set and test set according to the ratio of 7:1.5:1.5.

[0036] Step S2.1.2: Build a motor operation status monitoring model based on RNN and CTC network:

[0037] Determine the input layer: Based on the characteristic dimensions of the current, voltage, temperature and speed data of each sensor stream information segment n, the number of neurons in the input layer is determined to be 4×n. Assuming that each data feature in the sensor streaming information segment has n sampling points, the number of neurons in the input layer can be set to 4×n, that is, the number of sampling points corresponding to the four types of sensor data, used to receive the multidimensional data input of each information segment.

[0038] Determine the RNN hidden layer: Select the RNN unit type as Long Short-Term Memory (LSTM) or Gated Recurrent Unit (GRU), and determine the number of hidden layers and the number of neurons in each layer.

[0039] LSTM or GRU units can effectively process long-term dependencies in sequential data and are crucial for capturing time series features in sensory streaming information segments. Hidden layers are typically 3-5, enabling better extraction of deep-level features. The number of neurons in each layer can be adjusted empirically or experimentally; common choices include 64, 128, and 256.

[0040] Determine the CTC network layer: Add a CTC (Connectionist Temporal Classification) network layer after the RNN hidden layer to calculate the probability distribution between the input sequence and the output label sequence. This enables the model to directly classify or describe the state of sensory streaming information segments of variable length without forcibly aligning the input and output sequence lengths, because the length of the initial running state description vector may be different from the length of the input sensor data segment.

[0041] Determine the output layer: According to the dimension of the initial running state description vector, the number of neurons in the output layer is determined to be 4. Each neuron corresponds to the predicted value of a state feature and is used to output the initial running state description vector.

[0042] Step S2.1.3, model training: Use the training set in step S2.1.1 to train the model, use the CTC loss function as the training objective function of the model, optimize the model parameters by minimizing the CTC loss, so that the model can accurately predict the corresponding initial operating state description vector based on the input sensor streaming information segment; use the stochastic gradient descent SGD or Adam optimization algorithm to update the network parameters.

[0043] The Adam optimization algorithm has the characteristic of adaptive learning rate adjustment, which can better balance the convergence speed and stability. It usually shows good results when training RNN-based models and is the preferred choice.

[0044] During the training process, the training data is input into the constructed model in batches, the output results are calculated through forward propagation, and then the loss value is calculated using the CTC loss function. The network parameters are updated through back propagation and optimization algorithms. After each training cycle, the model is verified using the validation set to evaluate the performance indicators of the model on the validation set, such as loss value, accuracy, etc. The hyperparameters of the model (such as learning rate, number of hidden layer neurons, etc.) and the parameter update strategy during training are adjusted according to the verification results to prevent overfitting and underfitting, until the performance of the model on the validation set reaches the preset training stop condition.

[0045] Step S2.1.4, real-time output of the motor's initial operating state description vector: The motor's current current, voltage, temperature, and speed data collected by the sensor are segmented and processed according to the same preprocessing process as the training data. The real-time preprocessed sensor streaming information segment data is input into the trained state monitoring model based on RNN and CTC networks, and the model outputs the corresponding initial operating state description vector.

[0046] Based on the initial operating state description vector, the model can determine in real time whether the motor's current operating state is normal and whether any abnormalities or faults are present, enabling timely implementation of appropriate control measures to ensure stable motor operation. Furthermore, as new data is continuously input, the model can update its operating state monitoring results in real time, enabling continuous tracking and monitoring of the motor's operating status.

[0047] Further optimization, in step S2.3, the motor operating state corresponding to the initial operating state description vector of each sensor stream information segment is calculated, thereby obtaining the pending motor operating state sequence Q corresponding to the global operating state description vector V, specifically including:

[0048] Step S2.3.1, setting mapping rules: according to the physical meaning of the motor operating state and expert knowledge, set rules for mapping the initial operating state description vector to the specific motor operating state.

[0049] Step S2.3.2, state determination: for each sensor stream information segment, the initial running state description vector V j , according to the above mapping rules, the corresponding motor operating state is obtained q j ;

[0050] Step S2.3.3, motor operation state sequence arrangement: Arrange the motor operation states corresponding to the initial operation state description vector in the operation state description vector V in chronological order to obtain the pending motor operation state sequence Q = [ q 1, q 2,…, q N ].

[0051] Further optimization is performed in step S3.1. For the sensor stream information segment with abnormal motor operating state in the pending motor operating state sequence, the operating state descriptor vector of each data point in the corresponding sensor stream information segment is obtained, and based on the operating state descriptor vector, the possibility of the motor instantaneous characteristics corresponding to these data points is estimated, specifically including:

[0052] Step S3.1.1, extracting the operating state descriptor vector, includes:

[0053] Step S3.1.1.1. Adjust the model and train: Based on the RNN-CTC network state monitoring model described above, adjust the model output to meet the need for obtaining the operating state descriptor vector. Specifically, add a sub-vector generation layer between the RNN hidden layer and the CTC network layer. The number of neurons in this layer is set to 4 based on the expected dimensionality of the operating state descriptor vector.

[0054] The adjusted model is trained using pre-collected sensor stream information segments of normal and abnormal motor operating conditions. This training enables the adjusted model to accurately output an initial operating state description vector while also generating meaningful sub-vectors of the operating state for each data point. For example, a multi-task learning approach can be employed to combine the initial operating state description vector prediction task with the sub-vector generation task for each data point. Through joint training, the model simultaneously learns global operating state characteristics and the characteristics of local data points.

[0055] Step S3.1.1.2, sub-vector extraction: After the training of the adjusted model is completed, for the sensor streaming information segment with abnormal motor operating status in the pending motor operating status sequence, the information segment data is preprocessed according to the model input requirements, and then input into the adjusted model; during the forward propagation process, after the RNN hidden layer extracts the sequence features, the newly added sub-vector generation layer is used to obtain the operating status description sub-vector of each data point in the sensor streaming information segment.

[0056] Step S3.1.2, estimating the transient characteristics of the motor, includes:

[0057] Step S3.1.2.1, Characteristic Category Definition and Labeling: Clarify the four types of situations included in the transient characteristics of the motor, including current mutation, voltage fluctuation, temperature surge and abnormal speed mutation, and label and define these four situations.

[0058] Based on historical data or expert experience, the specific characteristics and judgment criteria of each characteristic category are determined. For example, a sudden change in current may be manifested as the current value exceeding a certain threshold in a short period of time and changing at a rapid rate.

[0059] Step S3.1.2.2, classification model construction and training:

[0060] Pre-labeled motor operation data samples containing different transient characteristic category information are collected. These samples cover various normal and abnormal transient characteristic conditions of the motor and are used as the dataset for training the classification model. Each sample element includes the operating state descriptor vector of the corresponding data point and the corresponding transient characteristic category label.

[0061] Select random forest, support vector machine, logistic regression, or deep neural network classification algorithms to build a classification model; use the operating state descriptor vector in the sample as the input feature and the corresponding instantaneous characteristic category label as the output target, train the classification model, and minimize the loss function (such as cross entropy loss) by adjusting the classification model parameters, so that the classification model can accurately predict the corresponding motor instantaneous characteristic category based on the operating state descriptor vector.

[0062] Step S3.1.2.3, likelihood estimation: for each data point in the sensor stream information segment to be predicted, the operating state descriptor vector is input into the trained classification model;

[0063] The classification model outputs the probability that the subvector corresponds to each instantaneous feature category, that is, the possibility.

[0064] Compared with the prior art, the present invention has the following beneficial effects:

[0065] 1) This invention comprehensively captures motor operating status information by comprehensively analyzing multi-dimensional data such as current, voltage, temperature, and speed. Compared with methods that rely solely on a single type of sensor data, it can more accurately detect the motor's operating status.

[0066] 2) When processing sudden motor states, by determining the positioning intervals of significant state data points and state indication information flows, it is possible to accurately locate the beginning and end of a complete sudden state, precisely grasp the evolution of the motor's operating state, and provide strong support for timely and accurate control decisions.

[0067] 3) Controlling the motor based on the final motor operating state sequence enables precise management of the motor's operating state. Timely adjustment of the control strategy based on the motor's actual operating state can improve the motor's safety, stability, and efficiency, reduce failures, and extend its service life, possessing significant practical application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 Schematic diagram of the implementation flow of the motor control-based operation data analysis method of the invention;

[0069] Figure 2 The figure is a hardware schematic diagram of a computer system described in the present invention. DETAILED DESCRIPTION

[0070] In order to make the purpose, technical solutions and advantages of this application clearer, the technical solutions of this application are further elaborated in detail below with reference to the accompanying drawings and embodiments. The described embodiments should not be regarded as limiting this application. All other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0071] Example 1: Figure 1 As shown, a method for analyzing operation data based on motor control includes the following steps:

[0072] Step S100: Acquire motor sensor streaming information and segment it: Based on multiple sensors arranged on the motor, collect motor operation data in real time to form continuous motor sensor streaming information; segment the continuous motor sensor streaming information to obtain multiple sensor streaming information segments.

[0073] In this implementation, sensor streaming data was collected from a motor on an industrial production line during 100 seconds of continuous operation. Table 1 shows this data. Due to space limitations, only a portion of the data is provided. This motor primarily drives a conveyor belt. During this 100-second period, the conveyor belt experienced various operating conditions, including startup, stable operation, brief overload, recovery from stable operation, and shutdown. Sensors installed on the motor collected current, voltage, temperature, and speed data at a rate of once per second.

[0074] Table 1 Sensor flow information of the motor during continuous 100 seconds of operation

[0075]

[0076] The current motor sensor stream information is divided into time periods of 10 seconds each, and 10 sensor stream information segments are obtained, as shown in Table 2.

[0077] Table 2 Sensor flow information segments of motor operation

[0078]

[0079] Step S200: Mining the motor operating state description vector and identifying the pending sequence: performing characterization information mining on each sensor stream information segment to obtain the operating state description vector of the motor sensor stream information; based on the operating state description vector, identifying the pending motor operating state sequence corresponding to the motor sensor stream information.

[0080] In this embodiment, an RNN-based state monitoring algorithm is used, which includes a CTC network. The input layer receives the current, voltage, temperature and speed data of each sensor streaming information segment, the hidden layer extracts and converts the features of these data, and the output layer outputs the initial operating state description vector.

[0081] Taking the first sensor stream information segment (1-10 seconds) as an example, its data is shown in Table 3, including 10 data points, each of which includes four values: current, voltage, temperature, and speed.

[0082] Table 3 Data of the first sensor stream information segment

[0083]

[0084] These data are input into the state monitoring algorithm. After calculation, the initial operating state description vector corresponding to the sensor streaming information segment is obtained as V1=[0.2, 0.8, 0.1, 0.9], where each element represents the operating state characteristic value of the motor in terms of current, voltage, temperature and speed.

[0085] Using the same method, the other nine sensor stream information segments are processed to obtain their corresponding initial operating state description vectors, as shown in Table 4.

[0086] Table 4 Initial operating state description vectors corresponding to the 10 sensor stream information segments

[0087]

[0088] Combine the above 10 initial operating state description vectors in order to obtain the operating state description vector V corresponding to the current motor sensor streaming information: [

[0090] [0.2, 0.8, 0.1, 0.9],

[0091] [0.22, 0.81, 0.11, 0.92],

[0092] [0.23, 0.82, 0.12, 0.93], …

[0093] [0.19, 0.8, 0.1, 0.88] ]

[0095] According to the physical meaning of the motor operating state and expert knowledge, a rule for mapping the initial operating state description vector to the specific motor operating state is set.

[0096] In this embodiment, three motor operating states are predefined: normal operation, overload operation, and fault operation, corresponding to three standard vectors. The standard vector corresponding to normal operation is [0.2, 0.8, 0.1, 0.9]; the standard vector corresponding to overload operation is [0.5, 0.7, 0.3, 0.6]; and the standard vector corresponding to fault operation is [0.8, 0.2, 0.6, 0.2].

[0097] In this embodiment, the set rule is: by calculating the initial running state description vector V corresponding to each segment in the running state description vector j The similarity with these three standard vectors (using Euclidean distance) is used to determine the operating state sequence of the pending motor.

[0098] Taking the initial operating state description vector [0.2, 0.8, 0.1, 0.9] corresponding to the first sensor stream information segment as an example, calculate its Euclidean distance with the three standard vectors:

[0099] Euclidean distance from the normal operating standard vector:

[0100] ;

[0101] Euclidean distance from the overload operation standard vector:

[0102] ;

[0103] Euclidean distance from the faulty running standard vector:

[0104] .

[0105] Since the Euclidean distance to the normal operating standard vector is the smallest, the undetermined motor operating state corresponding to this segment is normal operation.

[0106] Using the same method, the other nine sensor stream information segments are calculated to obtain the undetermined motor operation state sequence Q: [normal operation, normal operation, normal operation, overload operation, ..., normal operation].

[0107] In other embodiments, the rule may be set as follows: when the values of each dimension of the motor initial operating state description vector are within the normal range, the current dimension value is within [I norm_low ,I norm_high ], the voltage dimension value is between [U norm_low ,U norm_high ], the temperature dimension value is between [T norm_low ,T norm_high ], the speed dimension value is between [R norm_low ,R norm_high ]between.

[0108] Correspondingly, the overload state is: the current dimension value exceeds I norm_high , and the speed dimension value shows a downward trend. The stall state is: the speed dimension value is close to zero and the current dimension value rises sharply. The undervoltage state is: the voltage dimension value is lower than U low The initial operating state description vector V corresponding to each sensor stream information segment is j Compare with the above range to obtain the corresponding motor operating status.

[0109] Step S300: Determine the status-significant data points and positioning intervals: Search for sensor stream information segments with abnormal motor operating states in the pending motor operating state sequence, and find status-significant data points in the sensor stream information segments; based on the status-significant data points, determine the corresponding positioning intervals of the status indication information flow in the motor sensor stream information.

[0110] In this embodiment, based on the above-mentioned undetermined motor operating state sequence, it is determined that the operating state of the motor in the 5th sensor streaming information segment is abnormal. The adjusted model is used to obtain the operating state descriptor vector of each data point in the corresponding sensor streaming information segment. A logistic regression classification model is used to estimate the likelihood value of the instantaneous characteristics of the motor corresponding to each sensor monitoring data point. Specifically, 10 operating state descriptor vectors in the 5th sensor streaming information segment are input, and the probability of the motor corresponding to each sensor monitoring data point being in normal operation, overload operation, and fault operation is output. In other embodiments, a random forest, support vector machine, or deep neural network classification algorithm can be selected to construct a classification model.

[0111] In this embodiment, taking the sensor monitoring data point at the 51st second as an example, its corresponding operating state descriptor vector [0.4, 0.75, 0.2, 0.7] is input into the logistic regression model, and the probabilities of the motor corresponding to the data point being in normal operation, overload operation, and fault operation are obtained as follows: normal operation probability P1 = 0.3; overload operation probability P2 = 0.6; fault operation probability P3 = 0.1.

[0112] The threshold of the overload probability is set to 0.5. When the overload probability corresponding to a sensor monitoring data point exceeds 0.5, it is considered that a sudden change has occurred at the location of the data point.

[0113] By evaluating the probability values of all sensor monitoring data points, we found that the overload probability corresponding to the sensor monitoring data points from 51 to 53 seconds exceeded 0.5. Therefore, these three time points were identified as the locations where the sudden change occurred. The corresponding sensor monitoring data points were recorded as significant state data points. Since only the consecutive sudden change locations from 51 to 53 seconds occurred, we obtained a sudden change operating condition group.

[0114] According to the location where the mutation occurs, the start mutation condition is determined to be the 51st second and the end mutation condition is determined to be the 53rd second in the mutation condition group.

[0115] Determine the sequence number of the state significant data point corresponding to the start of the sudden change working condition in the current motor sensor flow information idx ( s g ) is 51.

[0116] Determine the sequence number of the state significant data point corresponding to the end of the sudden change working condition in the current motor sensor flow information idx ( e g ) is 53.

[0117] Set the first preset value ΔL1 to 1, and the second preset value ΔL2 to 1. Subtract the sequence number of the starting mutation operating condition data point from the first preset value to obtain the target starting data point sequence number of 50. Add the sequence number of the ending mutation operating condition data point to the second preset value to obtain the target ending data point sequence number of 54. Therefore, the positioning interval of the status indication information flow corresponding to the mutation operating condition group is obtained from the 50th to 54th seconds, that is, the target data point coverage range is from the 50th to 54th seconds. Therefore, the candidate sensor monitoring data point set is the sensor monitoring data points of the 50th, 51st, 52nd, 53rd, and 54th seconds. The corresponding sensor monitoring data points are shown in Table 5.

[0118] Table 5 Sensor monitoring data from 50 to 54 seconds

[0119]

[0120] Step S400: Determine and extract a target running state description vector: extract the running state sub-description vectors of all data points within the positioning interval from the running state description vector to obtain a target running state description vector of the state indication information flow.

[0121] In this embodiment, the running status sub-description vectors corresponding to seconds 50-54 in the running status description vector are: [0.3, 0.78, 0.15, 0.85] (50th second); [0.4, 0.75, 0.2, 0.7] (51st second); [0.45, 0.72, 0.25, 0.65] (52nd second); [0.42, 0.73, 0.22, 0.68] (53rd second); and [0.35, 0.76, 0.18, 0.8] (54th second). These sub-description vectors constitute the running status sub-description vector set.

[0122] The set of running state sub-description vectors is used as the target running state description vector corresponding to the state indication information flow. Then the target running state description vector corresponding to the state indication information flow is: [

[0124] [0.3, 0.78, 0.15, 0.85],

[0125] [0.4, 0.75, 0.2, 0.7],

[0126] [0.45, 0.72, 0.25, 0.65],

[0127] [0.42, 0.73, 0.22, 0.68],

[0128] [0.35, 0.76, 0.18, 0.8]

[0129] ].

[0130] Step S500: Optimizing the pending sequence to obtain the final sequence: Based on the target operating state description vector, optimizing the pending motor operating state sequence to obtain the final motor operating state sequence corresponding to the current motor sensor streaming information.

[0131] In this embodiment, the Euclidean distance between each sub-description vector in the target operating state description vector and three standard vectors (normal operation, overload operation, and fault operation) is calculated to determine the information flow motor operating state corresponding to each operating state sub-description vector.

[0132] Take the sub-description vector [0.4, 0.75, 0.2, 0.7] corresponding to the 51st second as an example:

[0133] Euclidean distance from the normal operating standard vector:

[0134] ;

[0135] Euclidean distance from the overload operation standard vector:

[0136] ;

[0137] Euclidean distance from the faulty running standard vector:

[0138] .

[0139] Since the Euclidean distance to the overload operation standard vector is the smallest, the motor operation state corresponding to the 51st second in the continuous motor sensor stream information is overload operation.

[0140] According to the same method, the motor operating states corresponding to seconds 50-54 are determined as follows: {50th second: normal operation; 51st second: overload operation; 52nd second: overload operation; 53rd second: overload operation; 54th second: normal operation}.

[0141] According to the positioning interval, the motor operating states corresponding to the 50th to 54th seconds are combined to form the target motor operating state subsequence q g ′: {normal operation (50th second), overload operation (51st-53th second), normal operation (54th second)}.

[0142] Using the target motor operating state subsequence q g ′, replace the motor running state corresponding to the positioning area in the sensor stream information segment with abnormal running state in the pending motor running state sequence Q, and obtain the final motor running state sequence Q′=[ q 1, q 2,…,q g ′,…, q N ].

[0143] In this embodiment, the motor operating state corresponding to the 50th to 54th seconds in the 5th sensor stream information segment in the pending motor operating state sequence Q is replaced with the target motor operating state subsequence q g ′, the final motor operation state sequence Q′ is: [normal operation, normal operation, normal operation, ..., normal operation (50th second), overload operation (51st-53rd second), normal operation (54th second), normal operation, ..., normal operation].

[0144] Step S600: Based on the final motor operating state sequence Q', it is discovered that the motor entered an overloaded state between seconds 51 and 53. To prevent damage to the motor due to prolonged overload, the following control measures are implemented: When the motor is detected to be overloaded (at the 51st second), the conveyor belt load is immediately reduced, for example, by reducing the number of items on the conveyor belt. Furthermore, additional cooling measures are implemented to reduce the motor temperature, such as activating an additional cooling fan. The motor's operating status is continuously monitored, and when the motor returns to normal operation (at the 54th second), the conveyor belt load is gradually restored to normal.

[0145] Embodiment 2: A computer system comprises a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and when the processor executes the program, some or all of the steps in the above method are implemented.

[0146] Figure 2 A schematic diagram of the hardware entities of a computer system provided for this embodiment is provided. The hardware entities of the computer system 1000 include: a processor 1001 and a memory 1002, wherein the memory 1002 stores a computer program that can be run on the processor 1001, and when the processor 1001 executes the program, the steps in the method of any of the above embodiments are implemented.

[0147] The memory 1002 stores computer programs that can be run on the processor. The memory 1002 is configured to store instructions and applications executable by the processor 1001. It can also cache data to be processed or processed by the processor 1001 and various modules in the computer system 1000 (for example, image data, audio data, voice communication data, and video communication data). This can be achieved through flash memory (FLASH) or random access memory (RAM).

[0148] When the processor 1001 executes the program, the steps of any of the above-mentioned motor control-based operation data analysis methods are implemented. The processor 1001 generally controls the overall operation of the computer system 1000.

[0149] An embodiment of the present application provides a computer storage medium, which stores one or more programs. The one or more programs can be executed by one or more processors to implement the steps of the motor control-based operation data analysis method of Example 1.

[0150] The above is only an implementation method of the present application, but the scope of protection of the present application is not limited thereto. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed in this application, which should be covered by the scope of protection of the present application.

Claims

1. An operation data analysis method based on motor control, characterized in that: The steps include: Step S1: Acquire motor sensor stream information and segment it: Based on multiple sensors arranged on the motor, collect motor operation data in real time to form continuous motor sensor stream information; Segmenting the continuous motor sensor stream information to obtain a plurality of sensor stream information segments; Step S2: Mining the motor operating state description vector and identifying the pending sequence: Mining the representation information of each sensor stream information segment to obtain the operating state description vector of the motor sensor stream information; based on the operating state description vector, identifying the pending motor operating state sequence corresponding to the motor sensor stream information; Step S3: Determine the state-significant data points and the positioning interval: Search for the sensor stream information segment with abnormal motor operation state in the pending motor operation state sequence, and find the state-significant data points in the sensor stream information segment; Based on the state-significant data points, a location interval of the corresponding state indication information flow is determined in the motor sensor flow information; Step S4: Determine and extract the target running state description vector: extract the running state sub-description vectors of all data points within the positioning interval from the running state description vector to obtain the target running state description vector of the state indication information flow; Step S5: Optimizing the pending sequence to obtain the final sequence: Based on the target operating state description vector, optimizing the pending motor operating state sequence to obtain the final motor operating state sequence corresponding to the current motor sensor streaming information; Step S6: According to the final motor operating state sequence, control measures are taken in a timely manner for abnormal states during subsequent motor operation to ensure that the motor returns to normal operation.

2. The method for analyzing operation data based on motor control according to claim 1, characterized in that: In step S1, motor operation data is collected, including current data, voltage data, temperature data and speed data. The motor sensor streaming information contains multiple data points, and each data point contains a timestamp and corresponding four-dimensional feature data.

3. The method for analyzing operation data based on motor control according to claim 2, characterized in that: In step S1, a sliding window technique is used to cut the continuous motor sensor stream information flow into overlapping or non-overlapping sensor stream information segments S. j , j ∈[1,N], where N represents dividing the continuous motor sensor stream information into N sensor stream information segments.

4. The method for analyzing operation data based on motor control according to claim 3, characterized in that: The step S2 specifically includes: Step S2.1, single segment feature extraction: for each sensor stream information segment S j , extract the time series features through the state monitoring algorithm and generate the initial operation state description vector V j =[ v j1 , v j2 , v j3 , v j4 ], v ji Indicates the j The sensor stream information segment should be the first i Operational status characteristic values of sensor-like monitoring data; i ∈[1, 4], representing current, voltage, temperature and speed respectively; Step S2.2, global state splicing: Splice the initial operating state description vectors of all sensor stream information segments in chronological order to obtain the global operating state description vector V=[V1, V2,…, V N ]; Step S2.3, initial state sequence identification: determine the motor operating state corresponding to the initial operating state description vector of each sensor stream information segment, thereby obtaining the pending motor operating state sequence Q corresponding to the global operating state description vector V = [ q 1, q 2,…, q N ],in q j For the j The initial operating state description vector of each sensor stream information segment corresponds to the motor operating state.

5. The method for analyzing operation data based on motor control according to claim 4, characterized in that: The step S3 specifically includes: Step S3.1, detect sudden working conditions and identify significant data points: for the sensor flow information segment with abnormal motor operating state in the motor operating state sequence to be determined, calculate the operating state descriptor vector of each data point, and estimate the possibility of the motor instantaneous characteristics corresponding to these data points based on the operating state descriptor vector, compare the possibility of each data point with the preset threshold, and identify the data points that exceed the preset threshold. d k Record it as a state-significant data point and record the mutation location corresponding to the data point idx ( k ); Step S3.2, forming mutation condition groups and locating intervals: classify all state significant data points into G mutation condition groups; determine the starting mutation corresponding to each mutation condition group s g and end mutation e g , and determine the starting mutations in the motor sensing flow information respectively s g and end mutation e g The corresponding data point will be the starting mutation data point position idx ( s g ) Push forward ΔL1 points and end the mutation data point position idx ( e g ) is extended by ΔL2 points to form the target data point coverage range of this sudden working condition group [ idx ( s g )-ΔL1, idx ( e g )+ΔL2], recorded as the positioning interval of the state indication information flow corresponding to the sudden working condition group; The same method is used to obtain the positioning intervals of the state indication information flows corresponding to all sudden working condition groups.

6. The method for analyzing operation data based on motor control according to claim 5, characterized in that: In step S4, for each positioning interval, [ idx ( s g )-ΔL1, idx ( e g The data points contained in )+ΔL2] are classified as the candidate sensor monitoring data point set, and the operating state descriptor vectors of all data points in the candidate sensor monitoring data point set constitute the target operating state description vector V g .

7. The method for analyzing operation data based on motor control according to claim 6, characterized in that: The step S5 comprises: Step S5.1, determine the motor operating state corresponding to each operating state description sub-vector in the target operating state description vector, and form a target motor operating state sub-sequence q corresponding to the positioning interval g '; Step S5.2: Using the target motor operating state subsequence q g ′, replace the motor running state corresponding to the positioning area in the sensor stream information segment with abnormal running state in the pending motor running state sequence Q, and obtain the final motor running state sequence Q′=[ q 1, q 2,…,q g ′,…, q N ].

8. The method for analyzing operation data based on motor control according to claim 7, characterized in that: In the step S2.1, for the sensor stream information segment S j ={x1,x2,......,x n } to mine the representation information and obtain the initial operation state description vector corresponding to each sensor stream information segment, where the data point x t =[ I ( t ),U( t ),T( t ),w( t )],include t The current, voltage, temperature and speed of the motor at all times; Specifically include: Step S2.1.

1. Data preprocessing and dataset construction: Collect a large amount of motor sensor streaming information in advance and segment it into several sensor streaming information segments as samples; standardize the different types of sensor data in the sensor streaming information segments so that the data features have similar scale ranges; then divide the preprocessed sensor streaming information segments into training, validation, and test sets in a 7:1.5:1.5 ratio; Step S2.1.2: Build a motor operation status monitoring model based on RNN and CTC network: Determine the input layer: Based on the characteristic dimensions of the current, voltage, temperature and speed data of each sensor stream information segment n , determine the number of neurons in the input layer to be 4×n; Determine the RNN hidden layer: Select the RNN unit type as LSTM or Gated Recurrent Unit (GRU), and determine the number of hidden layers and the number of neurons in each layer. Determine the CTC network layer: Add a CTC network layer after the RNN hidden layer to calculate the probability distribution between the input sequence and the output label sequence. This allows the model to directly classify or describe the state of sensory stream information segments of variable length without forcibly aligning the input and output sequence lengths. Determine the output layer: According to the dimension of the initial running state description vector, the number of neurons in the output layer is determined to be 4; Step S2.1.3, model training: Use the training set in step S2.1.1 to train the model, using the CTC loss function as the training objective function of the model. Optimize the model parameters by minimizing the CTC loss so that the model can accurately predict the corresponding initial operating state description vector based on the input sensor streaming information segment; use the stochastic gradient descent (SGD) or Adam optimization algorithm to update the network parameters; During the training process, the training data is input into the constructed model in batches, the output results are calculated through forward propagation, and then the loss value is calculated using the CTC loss function. The network parameters are then updated through backpropagation and optimization algorithms. After each training cycle, the model is verified using the validation set to evaluate the performance indicators of the model on the validation set until the performance of the model on the validation set reaches the preset training stop condition. Step S2.1.4, real-time output of the motor's initial operating state description vector: The motor's current current, voltage, temperature, and speed data collected by the sensor are segmented and processed according to the same preprocessing process as the training data. The real-time preprocessed sensor streaming information segment data is input into the trained state monitoring model based on RNN and CTC networks, and the model outputs the corresponding initial operating state description vector.

9. The method for analyzing operation data based on motor control according to claim 8, characterized in that: In step S2.3, the motor operating state corresponding to the initial operating state description vector of each sensor stream information segment is determined, thereby obtaining the pending motor operating state sequence Q corresponding to the global operating state description vector V, specifically including: Step S2.3.1, setting mapping rules: according to the physical meaning of the motor operating state and expert knowledge, setting rules for mapping the initial operating state description vector to the specific motor operating state; Step S2.3.2, state determination: for each sensor stream information segment, the initial running state description vector V j , according to the above mapping rules, the corresponding motor operating state is obtained q j ; Step S2.3.3, motor operation state sequence arrangement: Arrange the motor operation state corresponding to each initial operation state description vector in the operation state description vector V in chronological order to obtain the pending motor operation state sequence Q = [ q 1, q 2,…, q N ].

10. The method for analyzing operation data based on motor control according to claim 9, characterized in that: In step S3.1, for the sensor stream information segment with abnormal motor operating state in the pending motor operating state sequence, the operating state descriptor vector of each data point in the corresponding sensor stream information segment is obtained, and based on the operating state descriptor vector, the likelihood of the motor transient characteristics corresponding to these data points is estimated, specifically including: Step S3.1.1, extracting the operating state descriptor vector, includes: Step S3.1.1.

1. Adjust the model and train: Based on the RNN-based state monitoring model including the CTC network constructed in claim 8, add a sub-vector generation layer between the RNN hidden layer and the CTC network layer; the number of neurons in this layer is determined to be 4 based on the expected dimension of the operating state description sub-vector; The adjusted model is trained using pre-collected sensor stream information segments of the motor's normal and abnormal operating states. After training, the adjusted model can accurately output not only the initial operating state description vector but also the operating state descriptor subvector for each data point. Step S3.1.1.2, Extracting Subvectors of Sensory Streaming Information Segments with Abnormal Motor Operating Status: After the adjusted model is trained, the sensory streaming information segment data with abnormal motor operating status in the pending motor operating status sequence is pre-processed and input into the adjusted model. During the forward propagation process, the adjusted model extracts sequence features through the RNN hidden layer, and then outputs the operating status description subvector of each data point in the sensory streaming information segment through the newly added subvector generation layer. Step S3.1.2, estimating the transient characteristics of the motor, includes: Step S3.1.2.

1. Define and label characteristic categories: Identify the four types of conditions encompassed by the motor's transient characteristics, including current surges, voltage fluctuations, temperature surges, and abnormal speed surges, and label and define these four conditions. Based on historical data or expert experience, determine the specific characteristics and criteria for each characteristic category. Step S3.1.2.2, classification model construction and training: Collect pre-labeled motor operation data samples containing different transient characteristic category information. This sample covers a variety of normal and abnormal transient characteristics of the motor and uses this sample as the dataset for training the classification model. Each sample element includes the operating state descriptor vector of the corresponding data point and the corresponding transient characteristic category label. Select a random forest, support vector machine, logistic regression, or deep neural network classification algorithm to build a classification model; use the operating state descriptor vector in the sample as the input feature and the corresponding instantaneous characteristic category label as the output target, and train the classification model so that the classification model can accurately predict the corresponding motor instantaneous characteristic category based on the operating state descriptor vector; Step S3.1.2.3, possibility estimation: The operating state descriptor vector of each data point in the sensor streaming information segment of the abnormal motor operating state in the pending motor operating state sequence is input into the trained classification model, and the classification model outputs the probability that the operating state descriptor vector corresponds to each instantaneous characteristic category, that is, the possibility.

Images