Abnormal data detection method and device, electronic equipment and storage medium

Abstract

The invention relates to the technical field of abnormal detection of sensors, and discloses an abnormal data detection method and device, electronic equipment and a storage medium. The method comprises the following steps: acquiring sensor data with the length of N, wherein N is a positive integer; wherein when the length of the data of the test set is N, the average error between the predicted value and the actual value of the test set is the minimum; inputting the sensor data with the length of N into an abnormal data detection model, and obtaining a predicted value of the sensor data at the current moment; determining an abnormal degree value of the sensor data at the current moment based on the predicted value and the actual value of the sensor data at the current moment; and based on the abnormity degree value, adopting a corresponding abnormity processing strategy. According to the invention, the accuracy of anomaly detection can be improved, and potential risks can be early warned.

Description

Technical Field

[0001] The present disclosure relates to the technical field of abnormality detection of sensors, and in particular to a method, device, electronic device and storage medium for detecting abnormal data. Background Art

[0002] The rotary atomizer plays a vital role in waste incineration power plants. Its main function is to neutralize the acidic gases in the flue gas, such as sulfuric acid vapor (H2SO4) and hydrogen chloride (HCL), by injecting alkaline dry powder. If these acidic gases are discharged into the atmosphere without proper treatment, they will seriously harm the environment and affect human health. In this process, the sensors inside the rotary atomizer undertake the key tasks of monitoring and regulation to ensure that the amount of alkaline dry powder is accurately matched with the acidic components in the flue gas, so as to achieve the best neutralization effect.

[0003] However, the rotary atomizer may face equipment failure during long-term operation. By detecting the sensor data of the rotary atomizer, real-time monitoring and fault diagnosis of the rotary atomizer can be achieved. Summary of the invention

[0004] In view of the above situation, the embodiments of the present application provide a method, device, electronic device and storage medium for detecting abnormal data, aiming to solve the above problem or at least partially solve the above problem.

[0005] In a first aspect, an embodiment of the present application provides a method for detecting abnormal data, which includes: obtaining sensor data of a length of N, where N is a positive integer; wherein when the length of the test set data is N, the average error between the predicted value and the actual value of the test set is minimized; inputting the sensor data of length N into an abnormal data detection model to obtain the predicted value of the sensor data at the current moment; determining the abnormality degree value of the sensor data at the current moment based on the predicted value and the actual value of the sensor data at the current moment; and adopting a corresponding abnormality handling strategy based on the abnormality degree value.

[0006] In a second aspect, an embodiment of the present application also provides an abnormal data detection device, the device comprising: an acquisition module, used to acquire sensor data of a length of N, where N is a positive integer; wherein when the length of the test set data is N, the average error between the predicted value and the actual value of the test set is minimized; a processing module, used to input the sensor data of length N into an abnormal data detection model to obtain the predicted value of the sensor data at the current moment; based on the predicted value and the actual value of the sensor data at the current moment, determine the abnormality degree value of the sensor data at the current moment; and based on the abnormality degree value, adopt a corresponding abnormality handling strategy.

[0007] In a third aspect, an embodiment of the present application further provides an electronic device, comprising: a processor; and a memory arranged to store computer executable instructions, wherein the executable instructions, when executed, cause the processor to perform the steps of the first aspect described above.

[0008] In a fourth aspect, an embodiment of the present application further provides a computer-readable storage medium, which stores one or more programs. When the one or more programs are executed by an electronic device including multiple application programs, the electronic device executes the steps of the first aspect above.

[0009] At least one of the above technical solutions adopted in the embodiments of the present application can achieve the following beneficial effects: sensor data of length N is obtained. Since the average error corresponding to the value of N is the smallest, the predicted value output by the anomaly detection model based on the sensor data of length N will be more accurate, so that the abnormality degree value determined by the predicted value and the actual value can more accurately reflect the abnormality of the sensor data, thereby improving the accuracy of anomaly detection. Furthermore, the corresponding abnormality handling strategy is adopted through the abnormality degree value, and potential risks are warned in advance, helping enterprises to discover and solve problems in a timely manner, and effectively extending the service life of equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0011] Figure 1 A schematic diagram of the structure of an anomaly detection system provided in an embodiment of the present application is shown;

[0012] Figure 2 A schematic diagram of the layout of the sensor provided in the embodiment of the present application is shown;

[0013] Figure 3 A schematic diagram showing a flow chart of a method for detecting abnormal data provided in an embodiment of the present application is shown;

[0014] Figure 4 A model architecture diagram of an abnormal data detection model provided in an embodiment of the present application is shown;

[0015] Figure 5 A schematic diagram showing a flow chart of a method for determining data length provided in an embodiment of the present application;

[0016] Figure 6 A diagram showing the prediction results of the abnormal data detection model provided in an embodiment of the present application is shown;

[0017] Figure 7 shows a time domain waveform diagram of sensor data provided by an embodiment of the present application;

[0018] Figure 8 shows a frequency domain waveform diagram of sensor data provided by an embodiment of the present application;

[0019] Fig. 9 A flow chart showing a method for determining an abnormality degree value provided in an embodiment of the present application is shown;

[0020] Fig.10 A flow chart showing a method for determining an abnormality degree value provided by another embodiment of the present application;

[0021] Fig.11 A structural diagram of an abnormal data detection device provided in an embodiment of the present application is shown;

[0022] Fig.12 A schematic structural diagram of an electronic device provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0023] In order to make the purpose, technical solution and advantages of the present application clearer, the technical solution of the present application will be clearly and completely described below in combination with the specific embodiments of the present application and the corresponding drawings. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application.

[0024] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that such use is interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the term "including" and its variants are to be interpreted as open-ended terms meaning "including but not limited to".

[0025] The rotary atomizer plays a vital role in waste incineration power plants. Its main function is to neutralize the acidic gases in the flue gas, such as sulfuric acid vapor (H2SO4) and hydrogen chloride (HCL), by spraying alkaline dry powder. If these acidic gases are discharged into the atmosphere without proper treatment, they will seriously harm the environment and affect human health. In this process, the sensors inside the rotary atomizer undertake the key tasks of monitoring and regulation to ensure that the amount of alkaline dry powder is accurately matched with the acidic components in the flue gas, so as to achieve the best neutralization effect. However, the rotary atomizer may face equipment failure during long-term operation. By detecting the sensor data of the rotary atomizer, real-time monitoring and fault diagnosis of the rotary atomizer can be achieved.

[0026] In terms of detecting abnormal sensor data, there are three mainstream methods in related technologies, namely unsupervised clustering algorithm, binary classification-based method, and reconstruction-based method. Among them, the unsupervised clustering-based method and the reconstruction-based method have the problems of slow algorithm operation speed and poor real-time performance. The binary classification-based method relies on a large number of abnormal samples for training, but the frequency of equipment failure is not high, and the failure forms are diverse. Therefore, it is difficult to collect abnormal data samples under all abnormal conditions, and the practicality is not high. At the same time, the research on the above three algorithms only focuses on the detection of abnormal data points, without in-depth research on the degree of data abnormality, and has little effect on equipment fault diagnosis and real-time status monitoring.

[0027] Based on this, the embodiment of the present application proposes a sensor anomaly detection method, which obtains sensor data of length N. Since the average error corresponding to the N value is the smallest, the predicted value output by the anomaly detection model based on the sensor data of length N will be more accurate, so that the abnormality degree value determined by the predicted value and the actual value can more accurately reflect the abnormality of the sensor data, thereby improving the accuracy of anomaly detection. Furthermore, the corresponding abnormality handling strategy is adopted through the abnormality degree value, and potential risks are warned in advance, helping enterprises to discover and solve problems in a timely manner, and effectively extending the service life of equipment.

[0028] The present application is described in detail below through specific embodiments.

[0029] The sensor abnormality detection method proposed in this application can be applied to Figure 1 In the abnormal detection system shown in the figure, each sensor data is transmitted to a high-speed data collector, which collects the sensor data and sends it to the gateway, and the gateway uniformly sends the sensor data to the server ( Figure 1 The data is processed by a public cloud server (shown), through which the operation of the rotary atomizer can be viewed remotely in real time through electronic equipment.

[0030] In some embodiments, in view of the complex working environment of the rotary atomizer, the difficulty in collecting abnormal sensor data samples, and the difficulty in evaluating the degree of sensor data abnormality, the sensor is installed on the rotary atomizer in an array manner. The rotary atomizer is installed on the top platform of the boiler room of the waste incineration plant. The rotary atomizer is responsible for spraying alkaline powder ink to neutralize the acidic substances produced by combustion. Figure 2 As shown, according to the equipment mechanism and structure of the rotary atomizer, temperature-vibration integrated sensors, acceleration sensors, etc. are installed on the bearing, input, and output ends of the rotary atomizer respectively to collect relevant information of the rotary atomizer in real time and realize real-time monitoring of all components of the rotary atomizer without blind spots.

[0031] It should be noted that this application is not limited to Figure 1The anomaly detection system shown may be any system, device, framework, etc. that can implement the business logic of the present application. Figure 1 This is for exemplary purposes only.

[0032] Figure 3 The schematic diagram of the process of the sensor abnormality detection method provided in the embodiment of the present application is shown. Figure 3 It can be seen that the present application at least includes steps S101 to S104:

[0033] Step S101: Acquire sensor data of length N, where N is a positive integer.

[0034] Among them, when the length of the test set data is N, the average error between the predicted value and the actual value of the test set is the smallest. It is worth noting that the N value is predetermined based on the test set and the abnormal data detection model.

[0035] In some embodiments, the current moment is taken as a starting point, and N sensor data are acquired from the current moment forward.

[0036] In some embodiments, the sensor data includes, but is not limited to, vibration signals, temperature, voltage, current.

[0037] In some embodiments, the sensors are mounted on the device in an array.

[0038] Step S102: Input the sensor data of length N into the abnormal data detection model to obtain the predicted value of the sensor data at the current moment.

[0039] In some embodiments, the abnormal data detection model is used to predict sensor data at the current moment based on sensor data from a historical time period.

[0040] In some embodiments, the abnormal data detection model may be a convolutional neural network or other neural network that can implement the solution of the present application.

[0041] Step S103: Determine the abnormality level of the sensor data at the current moment based on the predicted value and the actual value of the sensor data at the current moment.

[0042] In some embodiments, the predicted value of the sensor data at the current moment is obtained through the abnormal data detection model, the actual value of the sensor data at the current moment is obtained through the sensor, and the abnormality degree value of the current sensor data is determined by the predicted value and the actual value, so as to determine whether the sensor data is abnormal.

[0043] Step S104: Based on the abnormality level value, adopt a corresponding abnormality handling strategy.

[0044] In the embodiment of the present application, sensor data of length N is obtained. Since the average error corresponding to the value of N is the smallest, the predicted value output by the anomaly detection model based on the sensor data of length N will be more accurate, so that the abnormality degree value determined by the predicted value and the actual value can more accurately reflect the abnormality of the sensor data, thereby improving the accuracy of anomaly detection. Furthermore, the corresponding abnormality handling strategy is adopted through the abnormality degree value, and potential risks are warned in advance, helping enterprises to discover and solve problems in a timely manner, and effectively extending the service life of equipment.

[0045] In some embodiments of the present application, before performing abnormal data detection, an abnormal data detection model is pre-trained. The training process of the abnormal data detection model is described below.

[0046] First, historical sensor data is collected, selected and classified, and sensor data under normal operation of the rotary atomizer is screened out. The reason for using normal data for training during the training process is that the changes in normal data are within a certain range, while the changes in abnormal data are difficult to predict. At the same time, equipment abnormalities are sporadic, and it is difficult to obtain enough samples in a short period of time. Therefore, the embodiment of the present application implicitly detects anomalies by learning the performance pattern of normal data from a large amount of historical data.

[0047] Furthermore, each normal data judgment starts from the beginning, and the data is divided into M (M represents a positive integer, which is obtained through experiments according to the specific situation of each sensor data) data points as the length, M+1 data points are used as historical data for training, and the M+1th data is the predicted value of the neural network.

[0048] Next, we build a neural network for abnormal data detection of sensors. Since vibration signals, pressure sensor signals, temperature sensor signals, etc. are all discrete digital signals. Discrete digital signals contain two types of information: time domain and frequency domain. These two types of information are also the basis for fault diagnosis and predictive maintenance. Based on the above considerations, the abnormal data detection model is designed as a dual-input single-output model. Specifically, Figure 4 As shown in the figure, the model is designed in the form of dual input. One branch is responsible for extracting features of time domain information through a convolutional neural network, and the other branch is responsible for analyzing the frequency domain information of the input data. The frequency domain information and time domain information of the two branches are then combined through the adaptive weights and methods of the neural network, and the prediction results are finally output through the full convolution layer.

[0049] In the neural network training process, the mean square error loss is used as the loss function. In an exemplary embodiment, the loss function is as follows: Among them, x i and iare the predicted value and the true value of the model, respectively, and N is the number of samples. The mean square error loss uses squared error to impose a greater penalty on the error, which helps the model better predict the current data value based on historical data. At the same time, it also provides a clear gradient direction during the optimization process, which can help the model converge to the local minimum faster.

[0050] Furthermore, the training data set, validation data set, and test data set are divided according to a certain ratio (for example, 6:2:2), and the model with the smallest loss between the predicted value and the actual value on the validation set is retained during the training process. If the loss value does not decrease within 50 epochs during training, it means that the model has converged and the training is stopped. At this point, the training of the abnormal data detection model is completed.

[0051] In the embodiment of the present application, since there are too few useful data samples in the industrial scene, that is, too few abnormal samples, if the model is trained based on abnormal samples, there will be a problem of inaccurate model prediction. Therefore, the present application considers from another perspective that since normal samples are sufficient, if the performance pattern of normal samples can be learned, abnormal situations can be distinguished. Therefore, the embodiment of the present application predicts the current data value through a large amount of historical normal data, and judges the abnormal situation of sensor data through the predicted value and the actual value.

[0052] In some embodiments of the present application, after the model training is completed, the trained model is tested through a test set to determine the value of N in step S101.

[0053] Figure 5 A schematic diagram of a method for determining data length according to another embodiment of the present application is shown. Figure 5 It can be seen that the method for determining the data length of this embodiment includes the following steps S201-S203:

[0054] Step S201: Select sensor data of different lengths as a test set.

[0055] Among them, each test set includes multiple groups of sensor data with the same length, and the sensor data is normal data;

[0056] Step S202: Determine the mean absolute error of each test set based on the abnormal data detection model.

[0057] Step S203: taking the sensor data length corresponding to the test set with the smallest mean absolute error as the N value.

[0058] Specifically, a positive integer M is randomly selected to start dividing the length of each group of sensor data in the test set. After the sensor data in the test set is preprocessed, it is input into the trained model to obtain the predicted value, and the mean absolute error between the predicted value and the true value of each group of sensor data is calculated.

[0059] In one implementation, the mean absolute error is determined based on the following formula:

[0060]

[0061] Among them, M b represents the mean absolute error, N represents the number of sensor data groups in the test set, and V Pi and V ri Represent the predicted value and true value of each set of sensor data respectively.

[0062] Furthermore, the experimental results are recorded, and the M value is randomly increased or decreased according to the experimental results, and the above process is repeated. b If the value decreases, continue to increase the M value in the next experiment until the M value is increased to M. b When the value no longer changes or decreases, the M value is determined to be the optimal value. Similarly, if the M value decreases, M b If the value decreases, continue to reduce the value of M in the next experiment until the value of M is reduced. b When the value no longer changes or decreases, it is determined that the M value is the optimal value.

[0063] Table 1 below shows how the average prediction error of the rotary atomizer bearing temperature sensor data changes with the M value. When the M value is small, the average error is large because the data length is too short and the amount of information is insufficient. When the M value is close to 100, the average error is basically reduced to the minimum. When the M value exceeds 100, the historical data used is too long, and there is redundant information that affects the algorithm performance, and the average error begins to increase gradually. Therefore, the M value of 100 is the best value.

[0064] Table 1 Average prediction error of rotary atomizer bearing temperature sensor data

[0065] Serial number M-value Average Error 1 20 0.51 2 30 0.41 3 50 0.35 4 70 0.38 5 90 0.22 6 100 0.10 7 120 0.13 8 110 0.12 9 105 0.12 10 103 1.11 11 101 0.10 12 99 0.11

[0066] It is worth noting that Table 1 is only for illustrative purposes, and the specific N value needs to be determined based on actual tests.

[0067] In the embodiment of the present application, when the sensor data length of the test set is N, the average error is the smallest, that is, when the sensor data length is N, the prediction value of the model is the most accurate.

[0068] In the training process of the abnormal data detection model, 30,000 historical data were used for training and 2,000 data were used for testing. The results show that the maximum error between the predicted value and the actual value output by the abnormal data detection model does not exceed 1.5, and the average error is 0.1. Figure 6As shown, red represents predicted data and blue represents target data, which means that the abnormal data detection model can accurately predict the current data value through historical data.

[0069] In some embodiments of the present application, since the abnormal data detection model is a dual-input single-output model, the sensor data needs to be preprocessed after being acquired.

[0070] In some embodiments, sensor data of length N is preprocessed to generate time domain data and frequency domain data, and the time domain data and frequency domain data are input into a trained abnormal data detection model to obtain a predicted value of the sensor data at the current moment.

[0071] In one embodiment, the preprocessing process is to first filter the sensor data to remove the noise signal, and the filtered time domain waveform is as follows: Figure 7 As shown, the sensor data is then converted from the time domain to the frequency domain. The converted frequency domain waveform is as follows Figure 8 shown.

[0072] Specifically, since the data measured by the sensor always has a lot of uncertainty, such as a lot of noise, and most of this noise conforms to the Gaussian distribution, the Kalman filter algorithm can be used for optimal estimation. The idea of ​​Kalman filtering is to use the state equation of the system to predict the current value and use the observed value measured by the sensor to correct the predicted value. Kalman filtering includes two processes, namely prediction and update.

[0073] Prediction phase: Predict the current state based on the best estimate of the previous state and evaluate the uncertainty of the prediction. The prediction equation is as follows: in, represents the predicted state at time k based on the best estimate at time k-1. A and B are the state transfer matrix and control input matrix, respectively, which describe the dynamic behavior of the system. u k represents the observation noise. The corresponding covariance prediction is as follows: in, represents the prediction error covariance, Q represents the process noise covariance, P k-1 represents the estimated covariance matrix at time k-1.

[0074] Update phase: Use the current observations to correct the predicted values ​​to get a more accurate estimate. The process is as follows: First, calculate the Kalman gain. The calculation process is as follows: Among them, k k represents the Kalman gain, H represents the observation matrix, and R represents the observation noise covariance. Then the estimate is updated according to the observation value as follows: in, represents the estimated state at time k, z krepresents the observation vector at time k. Finally, the error covariance is updated as follows: P k =(Ik k H)P k - , where P k Represents the estimated error covariance, I represents the identity matrix, whose elements on the main diagonal are all 1, and the elements in the other positions are all 0.

[0075] Fourier transform is used to convert the sensor signal from time domain to frequency domain. The conversion process is as follows:

[0076]

[0077] Among them, x(k) represents the kth frequency component, x(n) is the nth sampling point of the original signal, N represents the total number of sample points, and i is an imaginary unit.

[0078] In an embodiment of the present application, the noise in the sensor data is filtered out by a Kalman filtering algorithm, and the sensor data after the noise is filtered out is converted from the time domain to the frequency domain to obtain accurate and comprehensive sensor data, so that the input data of the abnormal data detection model is more accurate and comprehensive, thereby outputting more accurate prediction data.

[0079] In some embodiments of the present application, after the predicted value of the sensor data at the current moment is obtained through the abnormal data detection model, the abnormality degree value is determined by the predicted value and the actual value.

[0080] Fig. 9 A flow chart of a method for determining an abnormality degree value according to an embodiment of the present application is shown. Fig. 9 It can be seen that the method for determining the abnormality level value in this embodiment includes the following steps S301-S302:

[0081] Step S301: Determine the normal probability value of the current sensor data based on the absolute difference at the current moment, the mean absolute difference of the test set, and the standard deviation of the test set.

[0082] In some embodiments, the mean absolute difference of the test set and the standard deviation of the test set are obtained and saved during model testing.

[0083] In some embodiments, the normal probability value represents the probability that the current sensor data is normal data.

[0084] In some embodiments, the z value is calculated based on the absolute difference at the current moment, the mean of the absolute difference of the test set, and the standard deviation of the test set. A normal probability value is obtained based on the z value and the normal distribution table.

[0085] In one implementation, the z value is calculated based on the following formula:

[0086]

[0087] Among them, x represents the absolute difference at the current moment, u represents the mean of the absolute difference of the test set, and σ represents the standard deviation of the test set.

[0088] Step S302: Based on the normal probability value, determine the abnormality level value of the current sensor data.

[0089] In some embodiments, the abnormality level value is determined based on the following formula:

[0090] P sa =1-P c

[0091] Among them, P sa Indicates the abnormality value, P c Represents the normal probability value.

[0092] In some embodiments of the present application, since the data used by the training model of the present application is normal data,

[0093] In some embodiments of the present application, before calculating the normal probability value of the sensor data, the absolute difference between the predicted value and the actual value of the current sensor data can be compared with a preset threshold to preliminarily determine whether the current sensor data is abnormal. If it is determined to be abnormal, steps S301-S302 are executed. If there is no abnormality, the predicted value of the sensor data at the next moment is predicted.

[0094] Fig.10 A flow chart of a method for determining an abnormality degree value according to another embodiment of the present application is shown. Fig.10 It can be seen that the method for determining the abnormality level value in this embodiment includes the following steps S401 to S403:

[0095] Step S401: Determine the absolute difference based on the predicted value and the actual value of the sensor data at the current moment.

[0096] Step S402: If the absolute difference is greater than a preset threshold, it is determined that the sensor data at the current moment is abnormal.

[0097] Step S403: Determine the normal probability value of the current sensor data based on the absolute difference at the current moment, the mean absolute difference of the test set, and the standard deviation of the test set.

[0098] Step S404: Based on the normal probability value, determine the abnormality level value of the current sensor data.

[0099] In the embodiment of the present application, by judging the absolute difference and threshold between the predicted value and the actual value, it is preliminarily determined whether the sensor data is abnormal. If there is no abnormality, the sensor data at the next moment is predicted directly; if there is an abnormality, the abnormality degree value is judged, which saves unnecessary resources and improves detection efficiency.

[0100] In some embodiments of the present application, different abnormal data values ​​correspond to different abnormality handling strategies.

[0101] In some embodiments, the abnormal data values ​​include no abnormality, general abnormality, moderate abnormality, and severe abnormality.

[0102] In one embodiment, when the abnormal data value is no abnormality, the abnormality handling strategy is normal operation. Specifically, when the abnormal data value is no abnormality, it means that the rotary atomizer is in a normal operating state, and no special measures need to be taken at this time, and the current operating state can be maintained.

[0103] In one embodiment, when the abnormal data value is a general abnormality, the abnormality handling strategy is to issue a first warning message to remind the staff to check the equipment. Specifically, when the abnormal data value is a general abnormality, it may be due to some parameters slightly deviating from the normal range or sensor data fluctuating greatly. At this time, the system will issue a warning and recommend preliminary inspection and debugging, such as checking whether the sensor is slightly contaminated or loose, adjusting the equipment parameters, etc., to ensure that the equipment can continue to operate stably.

[0104] In one embodiment, when the abnormal data value is a moderate abnormality, the abnormality handling strategy is to issue a second warning message, and the second warning message is used to prompt the staff to take emergency measures for the equipment. Specifically, when the system detects a moderate abnormality, it usually means that the equipment has a more obvious fault or potential risk. The system will immediately issue an alarm and recommend emergency measures, such as stopping the machine for inspection, replacing faulty parts, adjusting process parameters, etc., to avoid further damage to the equipment or production interruption.

[0105] In one implementation, when the abnormal data value is a severe abnormality, the abnormality handling strategy is to issue a third warning message, which is used to prompt the staff to repair the equipment. Specifically, if the system detects a severe abnormality, it means that the equipment has suffered a serious failure or is about to face the risk of losing control. At this time, the system will immediately start the emergency shutdown procedure and recommend that professionals be immediately organized to conduct a comprehensive inspection and repair of the equipment to ensure that the equipment can resume normal operation as soon as possible, while preventing the accident from expanding and causing more serious consequences.

[0106] In an embodiment of the present application, the abnormality level of sensor data is determined to determine the operating status of the equipment, predict the potential risks of the equipment, and execute the corresponding abnormality handling strategy based on the abnormality level to avoid irreparable damage to the equipment and effectively extend the service life of the equipment.

[0107] In summary, this application aims at the challenges of harsh environment, numerous variables, and complex and changeable operating state of rotary atomizer during waste incineration, especially when sensor data is susceptible to dust pollution, high temperature and mechanical vibration interference, and traditional monitoring methods are difficult to accurately identify abnormal conditions. A new sensor data detection method is proposed, which integrates a multi-dimensional sensor array to fully cover the working area of ​​the rotary atomizer, ensure the comprehensiveness and accuracy of data collection, and transmit the sensor data to the server in real time through wired network technology. After the server receives the data, it effectively removes false positive abnormal signals caused by normal fluctuations of the environment and equipment through pre-processing techniques such as filtering and spectrum analysis. Further, a generative machine learning model is used to deeply analyze the processed data, learn and establish an abnormal data detection model under normal working conditions. By comparing real-time data with model predictions, the system can instantly identify abnormal data and abnormal degree values ​​that deviate from the normal mode, accurately locate sensor failures or abnormal operation of the rotary atomizer, and thus warn of potential equipment damage or process out-of-control risks in advance, thereby helping enterprises to discover and solve problems in a timely manner and effectively extend the service life of the equipment. In addition, the solution also has fast reasoning speed and high prediction accuracy, providing a solid foundation for enterprises to maximize the use of equipment, and helping enterprises to formulate more scientific and reasonable equipment maintenance plans based on real-time data, further ensuring the stability and sustainability of the production process.

[0108] In some embodiments of the present application, a device for detecting abnormal data is provided, and the device for detecting abnormal data corresponds one-to-one to the method for detecting abnormal data in the above embodiments. Fig.11 As shown, the abnormal data detection device includes: an acquisition module 501 and a processing module 502. The detailed description of each functional module is as follows:

[0109] An acquisition module is used to acquire sensor data of length N, where N is a positive integer; wherein when the length of the test set data is N, the average error between the predicted value and the actual value of the test set is the smallest;

[0110] The processing module is used to input the sensor data of length N into the abnormal data detection model to obtain the predicted value of the sensor data at the current moment; based on the predicted value and actual value of the sensor data at the current moment, determine the abnormality degree value of the sensor data at the current moment; based on the abnormality degree value, adopt a corresponding abnormality handling strategy.

[0111] In some embodiments of the present application, in the above device, the value of N is determined based on the following method:

[0112] Selecting sensor data of different lengths as test sets, each test set includes multiple groups of sensor data of the same length, and the sensor data are normal data;

[0113] Determining a mean absolute error for each test set based on the abnormal data detection model;

[0114] The sensor data length corresponding to the test set with the smallest mean absolute error is taken as the N value.

[0115] In some embodiments of the present application, in the above-mentioned device, the processing module 502 is specifically used to preprocess the sensor data of length N to generate time domain data and frequency domain data; and input the time domain data and frequency domain data into the abnormal data detection model.

[0116] In some embodiments of the present application, in the above-mentioned device, the processing module 502 is specifically used to determine the normal probability value of the current sensor data based on the absolute difference at the current moment, the mean absolute difference of the test set and the standard deviation of the test set; based on the normal probability value, determine the abnormality degree value of the current sensor data.

[0117] In some embodiments of the present application, in the above-mentioned device, the processing module 502 is also used to determine the absolute difference based on the predicted value and the actual value of the sensor data at the current moment; if the absolute difference is greater than a preset threshold, it is determined that the sensor data at the current moment is abnormal.

[0118] In some embodiments of the present application, in the above-mentioned device, the processing module 502 is specifically used for: if the abnormality level value is no abnormality, the abnormality handling strategy is normal operation; if the abnormality level value is a general abnormality, the abnormality handling strategy is to issue a first warning message, and the first warning message is used to prompt the staff to check the equipment; if the abnormality level value is a moderate abnormality, the abnormality handling strategy is to issue a second warning message, and the second warning message is used to prompt the staff to take emergency measures for the equipment; if the abnormality level value is a serious abnormality, the abnormality handling strategy is to start the emergency shutdown procedure and issue a third warning message, and the third warning message is used to prompt the staff to repair the equipment.

[0119] In some embodiments of the present application, in the above-mentioned device, the acquisition module 501 is specifically used to acquire sensor data with a length of N based on an array sensor.

[0120] It should be noted that any of the above-mentioned abnormal data detection devices can implement the above-mentioned abnormal data detection method in a one-to-one correspondence, which will not be repeated here.

[0121] Fig.12 FIG. 1 is a schematic diagram showing the structure of an electronic device provided in an embodiment of the present application. Fig.12 As shown, at the hardware level, the electronic device includes a processor, and optionally also includes an internal bus, a network interface, and a memory. Among them, the memory may include a memory, such as a high-speed random access memory (RAM), and may also include a non-volatile memory (non-volatile memory), such as at least one disk storage, etc. Of course, the electronic device may also include hardware required for other services.

[0122] The processor, network interface and memory can be interconnected through an internal bus, which can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Fig.12 Only one bidirectional arrow is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0123] The memory is used to store programs. Specifically, the program may include program code, and the program code includes computer operation instructions. The memory may include internal memory and non-volatile memory, and provides instructions and data to the processor.

[0124] The processor reads the corresponding computer program from the non-volatile memory into the memory and then runs it, forming an abnormal data detection device at the logical level. The processor executes the program stored in the memory and is specifically used to execute the above method.

[0125] The processor may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by an integrated logic circuit of hardware in the processor or an instruction in the form of software. The above processor may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in the embodiments of the present application can be implemented or executed. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The steps of the method disclosed in the embodiments of the present application can be directly embodied as a hardware decoding processor to be executed, or a combination of hardware and software modules in the decoding processor to be executed. The software module may be located in a mature storage medium in the field such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory, and the processor reads the information in the memory and completes the steps of the above method in combination with its hardware.

[0126] The electronic device can execute the abnormal data detection method provided in multiple embodiments of the present application, and realize the abnormal data detection device in Fig.11 The functions of the illustrated embodiment will not be described in detail in the embodiments of the present application.

[0127] An embodiment of the present application also proposes a computer-readable storage medium, which stores one or more programs, and the one or more programs include instructions. When the instructions are executed by an electronic device including multiple application programs, the electronic device can execute the abnormal data detection method provided by multiple embodiments of the present application.

[0128] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0129] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0130] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0131] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0132] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0133] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0134] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0135] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity or device including the element.

[0136] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0137] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. A method for detecting abnormal data, characterized in that: The method comprises: Obtain sensor data of length N, where N is a positive integer; when the length of the test set data is N, the average error between the predicted value and the actual value of the test set is the smallest; Inputting the sensor data of length N into the abnormal data detection model to obtain the predicted value of the sensor data at the current moment; Determine the abnormality level of the sensor data at the current moment based on the predicted value and the actual value of the sensor data at the current moment; Based on the abnormality level value, adopt the corresponding abnormality handling strategy.

2. The method according to claim 1, characterized in that The value of N is determined based on: Selecting sensor data of different lengths as test sets, each test set includes multiple groups of sensor data of the same length, and the sensor data are normal data; Determining a mean absolute error for each test set based on the abnormal data detection model; The sensor data length corresponding to the test set with the smallest mean absolute error is taken as the N value.

3. The method according to claim 1 or 2, characterized in that: The step of inputting the sensor data having a length of N into an abnormal data detection model comprises: Preprocessing the sensor data of length N to generate time domain data and frequency domain data; The time domain data and the frequency domain data are input into the abnormal data detection model.

4. The method according to claim 1, characterized in that Determining the abnormality level of sensor data at the current moment includes: Determine the normal probability value of the current sensor data based on the absolute difference at the current moment, the mean absolute difference of the test set, and the standard deviation of the test set; Based on the normal probability value, an abnormality degree value of the current sensor data is determined.

5. The method according to claim 4, characterized in that Before determining the normal probability value of the current sensor data, the method further includes: Determine the absolute difference based on the predicted value and the actual value of the sensor data at the current moment; If the absolute difference is greater than a preset threshold, it is determined that the sensor data at the current moment is abnormal.

6. The method according to claim 1, characterized in that The adopting of a corresponding exception handling strategy based on the abnormality degree value includes at least one of the following: If the abnormality level value is no abnormality, the abnormality handling strategy is normal operation; If the abnormality level is a general abnormality, the abnormality handling strategy is to issue a first warning message, and the first warning message is used to prompt the staff to check the equipment; If the abnormality level is a moderate abnormality, the abnormality handling strategy is to issue a second warning message, and the second warning message is used to prompt the staff to take emergency measures for the equipment; If the abnormality level value is a serious abnormality, the abnormality handling strategy is to start the emergency shutdown procedure and issue a third warning message, and the third warning message is used to prompt the staff to repair the equipment.

7. The method according to claim 1, characterized in that The step of acquiring sensor data having a length of N includes: Based on the array sensor, sensor data with a length of N is obtained.

8. A device for detecting abnormal data, characterized in that: The device comprises: An acquisition module is used to acquire sensor data of length N, where N is a positive integer; wherein when the length of the test set data is N, the average error between the predicted value and the actual value of the test set is the smallest; The processing module is used to input the sensor data of length N into the abnormal data detection model to obtain the predicted value of the sensor data at the current moment; based on the predicted value and actual value of the sensor data at the current moment, determine the abnormality degree value of the sensor data at the current moment; based on the abnormality degree value, adopt a corresponding abnormality handling strategy.

9. An electronic device, comprising: processor; as well as A memory arranged to store computer executable instructions, wherein when the executable instructions are executed, the processor executes the steps of the abnormal data detection method according to any one of claims 1 to 7.

10. A computer-readable storage medium storing one or more programs, which, when executed by an electronic device including a plurality of application programs, enables the electronic device to perform the steps of the abnormal data detection method as described in any one of claims 1 to 7.

Images