An intelligent sensor-based device operation data acquisition and processing system

Abstract

The present application discloses a device operation data acquisition and processing system based on smart sensors, which relates to the field of industrial Internet technology and solves the technical problem that the existing technology cannot consider the mutual influence between each smart sensor when collecting data, resulting in insufficient accuracy of the collected data; the present application analyzes the mutual influence between a number of smart sensors through an artificial intelligence model; then, after collecting a number of device operation data of a target device through a number of set smart sensors, the several device operation data are sequentially used as basic data, and an impact prediction model is extracted based on the basic data; the basic data is corrected based on the impact prediction model; the simulation rules set in the present application simulate the operation of each smart sensor in the device, and a data training set is constructed to train an impact prediction model, which can identify the mutual influence between the smart sensors, eliminate the mutual influence after correcting the basic data, and improve the accuracy of data acquisition.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of industrial internet, and particularly relates to a device running data acquisition and processing system based on intelligent sensors. BACKGROUND

[0002] In the industrial internet, a large amount of industrial data generated in the running process of a device is collected and extracted by intelligent sensors in time. The industrial data is used to evaluate the running state of the device, and the state of the device can also be predicted according to the collected historical data. Therefore, the device running data is very important in the industrial internet.

[0003] At present, in the device running data acquisition scheme based on intelligent sensors, the following problems exist: 1) the sampling frequencies of different intelligent sensors are different, so that the data collected by the intelligent sensors cannot be aligned in time. If the data is not time-aligned, the evaluation accuracy of the device state will be affected; 2) the intelligent sensors are easily affected by the environment and other sensors when collecting the device running data. If the data is not corrected, the real running state of the device cannot be analyzed.

[0004] The application provides a device running data acquisition and processing system based on intelligent sensors to solve the above technical problems. SUMMARY

[0005] The application aims to at least solve one of the technical problems in the prior art. To this end, the application provides a device running data acquisition and processing system based on intelligent sensors, which is used to solve the technical problem that the mutual influence between the intelligent sensors cannot be considered when the data is collected in the prior art, resulting in insufficient accuracy of the collected data.

[0006] To achieve the above-mentioned purpose, the first aspect of the application provides a device running data acquisition and processing system based on intelligent sensors, which comprises a cloud platform and an edge acquisition layer connected to the cloud platform.

[0007] The edge acquisition layer is used to collect a plurality of device running data of a target device through a plurality of intelligent sensors arranged.

[0008] The plurality of device running data are sequentially taken as basic data, and an influence prediction model is extracted based on the basic data. The basic data is corrected based on the influence prediction model to obtain target data.

[0009] The cloud platform is used to analyze the mutual influence between the plurality of intelligent sensors through an artificial intelligence model to obtain an influence prediction model.

[0010] The method is used for receiving a plurality of target data and interpolating the plurality of target data to realize alignment of the plurality of target data.

[0011] Preferably, the mutual influence between the plurality of intelligent sensors is analyzed by the artificial intelligence model, including:

[0012] The simulation training data is extracted, wherein the simulation training data is obtained by setting modes and operation modes of the plurality of intelligent sensors in the simulation device.

[0013] The artificial intelligence model is trained by the simulation training data, and the trained artificial intelligence model is marked as an influence prediction model; wherein the artificial prediction model includes a BP neural network model or an RBF neural network model.

[0014] Preferably, the simulation training data is obtained by:

[0015] A plurality of intelligent sensors of the device are determined, and the intelligent sensors are sequentially taken as target sensors; and other intelligent sensors except the target sensors are marked as non-target sensors.

[0016] Detection data of the target sensor during operation of the device is detected and marked as target data one; and detection data of the target sensor during normal operation of the non-target sensors according to simulation rules is simulated and marked as target data two.

[0017] A change ratio of the target data two and the target data one and a corresponding simulation rule are generated to form a simulation training data; wherein the simulation rule is used to indicate whether the non-target sensor is operated.

[0018] Preferably, the simulation rule is obtained by:

[0019] The operation state of the non-target sensor is marked to obtain a marking group; wherein the operation state includes operation and non-operation.

[0020] A marking is extracted from the marking group of each non-target sensor each time, and the extracted marking is integrated into a simulation rule; and a plurality of simulation rules are obtained after a plurality of times of extraction.

[0021] Preferably, the base data is corrected based on the influence prediction model, including:

[0022] The influence prediction model of the corresponding intelligent sensor of the base data and the operation state of other intelligent sensors during collection of the base data are extracted.

[0023] The operation state of the other intelligent sensors is preprocessed and input into the influence prediction model to obtain a change ratio; and the base data is corrected based on the change ratio.

[0024] Preferably, the interpolation of the plurality of target data includes:

[0025] extracting a plurality of target data, determining a target moment requiring interpolation processing according to the plurality of target data;

[0026] obtaining the target data of the target moment through interpolation processing.

[0027] Preferably, the target moment requiring interpolation processing is determined according to the plurality of target data, comprising:

[0028] taking the acquisition moment corresponding to any target data as a judgment moment, calculating the number of target data corresponding to the judgment moment;

[0029] judging whether the number of target data is consistent with the number of intelligent sensors; if yes, the judgment moment does not require interpolation processing; if no, taking the judgment moment as the target moment.

[0030] Preferably, the target moment requiring interpolation processing is determined according to the plurality of target data, comprising:

[0031] selecting the intelligent sensor with the worst stability based on the target data, and marking it as a reference sensor; taking the acquisition moment of the target data corresponding to the reference sensor as a judgment moment;

[0032] calculating the number of target data corresponding to the judgment moment; judging whether the number of target data is consistent with the number of intelligent sensors; if yes, the judgment moment does not require interpolation processing; if no, taking the judgment moment as the target moment.

[0033] Preferably, the intelligent sensor with the worse stability is selected based on the target data, comprising:

[0034] constructing a data fluctuation curve based on the target data of the plurality of intelligent sensors;

[0035] selecting the intelligent sensor with the worst stability as the reference sensor based on the data fluctuation curve.

[0036] Preferably, after the target moment is determined, the intelligent sensor corresponding to the missing data of the target moment is identified; the target data of the target moment is obtained through interpolation processing based on the target data collected by the intelligent sensor.

[0037] Compared with the prior art, the application has the beneficial effects that:

[0038] 1.The application analyzes the mutual influence between a plurality of intelligent sensors through an artificial intelligence model to obtain an influence prediction model; after a plurality of device operation data of a target device are collected by a plurality of intelligent sensors set, the plurality of device operation data are sequentially taken as basic data, and the influence prediction model is extracted based on the basic data; the basic data is corrected based on the influence prediction model to obtain target data; the simulation rule set in the application simulates the operation of each intelligent sensor in the device, and a data training set is constructed to train the influence prediction model, which can identify the mutual influence between the intelligent sensors, eliminate the mutual influence after the correction of the basic data, and improve the accuracy of data collection.

[0039] 2.In the application, the cloud platform receives a plurality of target data, and performs interpolation processing on the plurality of target data to align the plurality of target data; specifically, the data stability of the target data collected by each intelligent sensor is first judged, and the worst one is selected as the reference sensor; if the data stability is poor, the interpolation accuracy is low, so the accuracy of the interpolation result can be ensured through the scheme of the embodiment, and the collection time of each target data in the cloud platform after interpolation includes the target data of all other intelligent sensors, the alignment in time is realized, and the data processing time in subsequent device evaluation is saved. BRIEF DESCRIPTION OF DRAWINGS

[0040] In order to more clearly illustrate the technical solutions in the embodiments of the application or the prior art, the drawings needed to be used in the embodiments or the prior art description will be briefly introduced. Obviously, the drawings in the following description only represent some embodiments of the application, and other drawings can be obtained by those skilled in the art without creative labor.

[0041] Figure 1 The method steps of the device operation data collection and processing system in the first embodiment of the application are shown in the figure;

[0042] Figure 2 The system principle diagram of the device operation data collection and processing system in the first embodiment of the application is shown in the figure;

[0043] Figure 3 The method steps of the data stability judgment in the second embodiment of the application are shown in the figure. DETAILED DESCRIPTION

[0044] The technical solutions of the application will be described in detail below with reference to the embodiments. Obviously, the described embodiments are only some of the embodiments of the application, not all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the application.

[0045] The intelligent sensor is a sensor with information processing function, integrates a sensitive element, a microprocessor and a communication interface together, and can realize data collection, processing and transmission. The intelligent sensor not only has the basic function of the traditional sensor, but also has intelligent characteristics such as self-calibration, self-diagnosis and self-adaptation, can improve the measurement accuracy and reliability, and adapt to complex industrial environment.

[0046] Due to the above advantages, the intelligent sensor has been widely used in the industrial internet. Due to the intelligent characteristics, the intelligent sensor is less affected by the environment. However, in the industrial internet, there are many devices, and the deployment of the intelligent sensor is relatively complex, and the influence of the device on the intelligent sensor cannot be calibrated by pre-embedding a calibration model, which will cause the intelligent sensor to be easily affected by other devices, even other intelligent sensors during operation, so that the collection result of the intelligent sensor is inaccurate, and the evaluation and prediction of the device state are affected.

[0047] In order to calibrate the influence of other devices on the intelligent sensor, the technical scheme is designed to solve the above technical problems. Embodiments

[0048] Please refer to Figures 1-2 The first aspect embodiment of the present application provides a device running data acquisition and processing system based on an intelligent sensor, comprising a cloud platform and an edge acquisition layer connected thereto;

[0049] The edge acquisition layer is used for collecting a plurality of device running data of a target device through a plurality of intelligent sensors arranged;

[0050] In turn, a plurality of device running data are used as basic data, an influence prediction model is extracted based on the basic data; the basic data is corrected based on the influence prediction model to obtain target data;

[0051] The cloud platform is used for analyzing the mutual influence between a plurality of intelligent sensors through an artificial intelligence model to obtain an influence prediction model; and

[0052] For receiving a plurality of target data, and performing interpolation processing on the plurality of target data according to the influence prediction model, to realize alignment of the plurality of target data.

[0053] The device operation data acquisition and processing system in the embodiment mainly includes a cloud platform and a data acquisition layer. The data acquisition layer is mainly responsible for data acquisition and standardized processing. Data processing through the data acquisition layer can reduce the pressure of the cloud platform and improve the smoothness of the entire system. The main work of the data acquisition layer is to collect device operation related data through the set intelligent sensors, including but not limited to temperature, vibration, oil level, displacement, etc. After standardized processing of the collected data, it is sent to the cloud platform. Standardized processing includes data format unification, transmission protocol unification, etc.

[0054] One of the tasks of the cloud platform is to train an impact prediction model and send the impact prediction model to the data acquisition layer for further correction of the collected basic data. The second task is to time-align the received target data to provide qualified data for subsequent device state evaluation and prediction, thereby improving efficiency.

[0055] The intelligent sensor itself has certain data calibration capability. For example, for changes in temperature and humidity of the device, the data can be calibrated through the built-in calibration model to eliminate the influence of environmental changes on the data acquisition result. However, it is inevitable to deploy other intelligent sensors to arrange some data transmission lines. These data transmission lines will produce signal interference around when transmitting signals, and some intelligent sensors will also produce signal interference around when running. These interferences cannot be calibrated by built-in calibration models when the intelligent sensor is shipped.

[0056] For example, the main interference received by the intelligent sensor during operation is electromagnetic interference and signal crosstalk. Electromagnetic interference mainly refers to the electromagnetic radiation generated by the internal electronic components of some intelligent sensors when they are working. For example, high-frequency signal cables will generate an electromagnetic field when they are running, which will interfere with nearby intelligent sensors, especially magnetic sensors, capacitive sensors, and inductive sensors that are more sensitive to electromagnetic fields. Signal crosstalk mainly occurs when multiple intelligent sensors are installed adjacent or opposite to each other. In this case, same frequency interference may occur, such as ultrasonic sensors.

[0057] In addition, the intelligent sensors on the device mainly provide data for the device user, so the setting position and setting method of the intelligent sensors are not fixed, mainly considering the convenience of the device user. If there are many intelligent sensors and they are close to each other, they may interfere with each other, thereby affecting the accuracy of the data. Moreover, due to the flexibility of the setting position and setting method, it is not possible to calibrate through the pre-set calibration model.

[0058] The embodiment is to solve the above problems. After determining the device position and setting mode, intelligent sensors are set for each device. The operation of the intelligent sensors is simulated by simulation rules to obtain sufficient simulation training data. The artificial intelligence model is trained by the simulation training data. The trained artificial intelligence model contains the mapping relationship of different operation states of other intelligent sensors to the measurement results of the target sensor.

[0059] In actual operation, after the intelligent sensor collects the basic data, whether other intelligent sensors have an impact on the basic data collection can be obtained through the mapping relationship corresponding to the intelligent sensor. If there is an impact, the change rate can be combined with the basic data to complete data correction and obtain target data. The target data is the data that eliminates the influence of other intelligent sensors.

[0060] It should be noted that the intelligent sensor is not only affected by other intelligent sensors, but also affected by other devices. That is, the operation state of other devices can be introduced into the simulation rules to identify the data error of the intelligent sensor under the operation state of other intelligent sensors or other devices, so as to obtain more accurate target data.

[0061] One of the focuses of the technical solution provided by the embodiment is how to build an impact prediction model. Next, the construction method of the impact prediction model will be described in detail as follows:

[0062] Before building the impact prediction model, simulation is performed to obtain simulation training data. The specific scheme can be referred to as follows: a plurality of intelligent sensors are set in the device, and the intelligent sensors are sequentially taken as target sensors; other intelligent sensors except the target sensor are marked as non-target sensors; the detection data of the target sensor when the device is running is detected and marked as target data one; the detection data of the target sensor when the non-target sensor is simulated to operate normally is simulated according to the simulation rules and marked as target data two; the change rate of the target data two and the target data one, and the corresponding simulation rules generate a simulation training data.

[0063] First, a plurality of intelligent sensors set in the device need to be determined, including the setting position, signal transmission mode, etc. Before the device formally works, simulation is performed to obtain simulation training data. Specifically, one of the intelligent sensors is taken as a target sensor, and the others are taken as non-target sensors.

[0064] The operation state of the non-target sensor is marked, so that each non-target sensor corresponds to a marker group. In the marker group, the marker corresponding to the operation state is that the operation state and the non-operation state each correspond to a marker.

[0065] A mark is extracted from each non-target sensor, and all the extracted marks of the non-target sensors are spliced to form an array, which can be used as a simulation rule. A plurality of simulation rules corresponding to each target sensor can be obtained by permutation and combination. Each simulation rule can be understood as the working state of each non-target sensor when the target sensor is working. In an extreme case, all the non-target sensors can not work or can all work. It should be noted that the working states of the target sensor and the non-target sensors in the embodiment are independent of each other, that is, whether any non-target sensor works or not will not affect the normal working of the target sensor.

[0066] Through the above scheme, the intelligent sensors are sequentially determined as target sensors, and the simulation rules corresponding to the target sensors are determined by permutation and combination. Each target sensor corresponds to a plurality of simulation rules. By simulating the actual running condition of the intelligent sensors in the equipment through the simulation rules, the basic data can be collected.

[0067] After setting the simulation rules for each intelligent sensor, each intelligent sensor is still sequentially taken as a target sensor. Then the target sensor and the corresponding equipment are controlled to run, and the running environment corresponding to the target sensor is simulated according to the simulation rules in sequence, and the detection data of the target sensor is collected in real time. In this way, the detection data corresponding to the target sensor under one working scene of each non-target sensor can be collected each time, and the corresponding simulation rule and detection data are integrated into one simulation training data. In this way, a plurality of simulation training data corresponding to the target sensor can be obtained, and a plurality of simulation training data corresponding to each intelligent sensor can be obtained.

[0068] It should be noted that although a plurality of intelligent sensors can be set on the equipment according to needs, the number thereof will not be particularly large, and the number thereof is between several and ten. Moreover, the running state of the intelligent sensor can be controlled through the Internet of Things, the generation of the simulation rule is not difficult, and the actual simulation process is relatively simple, which will not bring a large amount of data processing. Moreover, in order to provide sufficient simulation training data, the running environment of the target sensor can also be simulated by combining the simulation rule with the running state of the equipment, such as the preheating stage, the normal running stage or the running life.

[0069] After the simulation training data corresponding to a plurality of intelligent sensors is determined through the above scheme, all the simulation training data can be integrated for training an artificial intelligence model. The specific training process can be referred to as follows:

[0070] The simulation training data corresponding to each intelligent sensor is obtained, and a digital mark corresponding to the intelligent sensor is inserted in each simulation training data, which is used to identify the target sensor to which the simulation training data belongs.

[0071] Then the running state of the non-target sensor corresponding to the digital mark and the simulation rule in each piece of simulation training data is preprocessed as the model input data; the change rate of the detection data of the target sensor corresponding to the digital mark is preprocessed as the model output data. The above-mentioned preprocessing is for input into the artificial intelligence model for training. It should be noted that the change rate here is calculated by the detection data of the target sensor when the non-target sensor is running under the simulation rule and the reference data, which can be calculated as: (detection data-reference data) / reference data; the reference data is the detection data collected by the target sensor in real time when other non-target sensors are not running.

[0072] According to the existing model training method, the artificial intelligence model constructed is trained by the model input data and the model output data, and the influence prediction model can be obtained. In the training process, the simulation training data is divided into training set, test set and validation set in proportion, and the proportion can be 5:3:2.

[0073] After the influence prediction model is obtained, it can be distributed to the data acquisition layer. After receiving the basic data collected by each intelligent sensor in real time, the data acquisition layer corrects the basic data by using the influence prediction model. The specific correction process can be referred to as follows:

[0074] After the data acquisition layer obtains the basic data, it determines the corresponding equipment of the basic data, and then extracts the influence prediction model corresponding to the equipment. The intelligent sensor that collects the basic data is regarded as the target sensor, the digital mark of the target sensor is identified, and whether other intelligent sensors work at the collection time of the basic data is determined according to the collection time of the basic data, that is, the running state of other intelligent sensors is determined. The digital mark and the running state of other intelligent sensors are processed as the model input data, and are input into the influence prediction model to obtain the change rate. Through the change rate, the correction of the basic data can be realized.

[0075] It should be noted that the intelligent sensor is not always working, but works according to the set sampling frequency, so when the intelligent sensor does not collect data, it can be considered that it will not affect other intelligent sensors, and its running state can be determined as not running. Of course, if a certain intelligent sensor does not run, its internal electronic components may still affect other intelligent sensors because it is in standby state, so the standby state of the intelligent sensor is also marked as the running state, and the training process of the influence prediction model is also processed according to this principle.

[0076] After the base data is corrected by affecting the prediction model, the part of the base data affected by other intelligent sensors can be removed to obtain the data of the device itself, i.e., target data. The target data is used for subsequent evaluation of the running state of the device, which can ensure the accuracy of the evaluation result.

[0077] After the target data of each intelligent sensor is corrected, the data acquisition layer sends the target data to the cloud platform. Due to the different sampling frequencies of each intelligent sensor, the target data received by the cloud platform cannot be aligned in time. If multi-source data fusion processing is required for the evaluation of the device, the detection data of all intelligent sensors cannot be obtained at the same time, and additional processing is required for the evaluation of the device, which will waste a lot of time and cannot obtain the evaluation state of the device in time.

[0078] An optional target data can be selected, and the collection time corresponding to the target data is used as the judgment time. In order to meet the needs of subsequent device evaluation, the judgment time needs to correspond to the target data of other intelligent sensors, and preferably all target data of intelligent sensors at the same judgment time.

[0079] The number of target data corresponding to the judgment time can be identified. If the number of target data is equal to the number of intelligent sensors on the device, the judgment time does not need to be processed. However, if the number of target data is less than the number of intelligent sensors, it means that the target data of some intelligent sensors is missing at the judgment time, and therefore the target data of these intelligent sensors needs to be interpolated.

[0080] The collection time that needs to be interpolated is used as the target time. The target data missing at the target time can be determined according to the target data collected by the cloud platform, and the intelligent sensors are marked as interpolation sensors. The target data of the target time can be calculated by constructing a data change curve according to the target data of the interpolation sensors and bringing the target time into the data change curve. The calculated target data is associated with the target time to complete the interpolation of the interpolation sensors. Next, other interpolation sensors or other target times are processed. After interpolation, the collection time corresponding to any target data in the cloud platform has target data of other intelligent sensors, and in subsequent device evaluation and analysis, multi-source data at the same time can be extracted, saving the time for temporary processing.

[0081] Embodiment two: The difference between embodiment one and embodiment two is that the scheme for determining the target time is different.

[0082] In order to ensure that the target data can be aligned in time, interpolation processing is used in the present example. Interpolation processing is affected by data fluctuations. The greater the data fluctuations, the worse the interpolation effect. Therefore, when performing interpolation processing, data with smaller fluctuations is given priority.

[0083] The data stability of each intelligent sensor is analyzed according to the target data received by the cloud platform. When analyzing the target stability, the target data collected during device abnormalities should be excluded in order to affect the interpolation effect. Data stability mainly determines whether the target data of each intelligent sensor is easily affected by other intelligent sensors or the environment. If it is not easily affected, it is determined that the stability of the intelligent sensor is good.

[0084] Data stability determination can be based on data fluctuation curves. For example, a data fluctuation curve is constructed based on the target data of several intelligent sensors. The least stable intelligent sensor is used as the reference sensor based on the data fluctuation curve.

[0085] Of course, the determination of data stability can also be based on data analysis methods, such as descriptive statistical analysis, regression analysis, variance analysis, and machine learning algorithms to complete stability analysis.

[0086] Please refer to Figure 3 In some other preferred embodiments, the influence prediction model obtained in embodiment one can also be used for stability evaluation:

[0087] The cloud platform calculates the change rate of each target data corresponding to the change rate curve by using the influence prediction model. The change rate curve of each intelligent sensor is fitted to evaluate the data stability by using a data analysis method.

[0088] Since the change rate only considers the mutual influence between intelligent sensors (including other preset interference sources) when calculating, the evaluation of data stability based on the change rate can screen out which intelligent sensor is less affected by other intelligent sensors and which is more affected. It can avoid the influence of non-intelligent sensors on the target data and reduce the accuracy of data stability evaluation.

[0089] The above embodiments are only used to illustrate the technical method of the present application and are not limiting. Although the present application has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical method of the present application can be modified or replaced equivalently without departing from the spirit and scope of the technical method of the present application.

Claims

1. A device operation data acquisition and processing system based on intelligent sensors, characterized in that: Includes the cloud platform and the edge collection layer connected to it; Edge collection layer: used to collect a number of device operation data of the target device through a number of set smart sensors; wherein the device operation data refers to data related to the operating status of the target device; and sequentially using the plurality of equipment operation data as basic data, extracting an impact prediction model based on the basic data; and correcting the basic data based on the impact prediction model to obtain target data; Cloud platform: used to analyze the mutual influence between the plurality of smart sensors through an artificial intelligence model to obtain the influence prediction model; and Used to receive the plurality of target data, and perform interpolation processing on the plurality of target data to achieve alignment of the plurality of target data; The mutual influence between several smart sensors is analyzed by an artificial intelligence model, including: Extracting simulation training data; wherein the simulation training data is obtained by simulating the setting mode and operation mode of the plurality of said smart sensors in the device; An artificial intelligence model is constructed by training simulated training data, and the trained artificial intelligence model is marked as an impact prediction model; wherein the artificial prediction model includes a BP neural network model or an RBF neural network model; The acquisition of the simulation training data includes: Determine a number of smart sensors provided on the device, and sequentially select the smart sensors as target sensors; and mark other smart sensors except the target sensors as non-target sensors; The detection data of the target sensor when the detection device is running is marked as target data one; the detection data of the target sensor when the non-target sensor is operating normally is simulated according to the simulation rules and marked as target data two; Generate a piece of simulated training data using the change ratio of the target data 2 to the target data 1 and the corresponding simulation rule; wherein the simulation rule is used to indicate whether the non-target sensor is operating; The acquisition of the simulation rules includes: Marking the operating status of the non-target sensor to obtain a marking group; wherein the operating status includes operating and not operating; Extracting one marker from the marker group of each non-target sensor each time, integrating the extracted markers into a simulation rule, and obtaining a plurality of simulation rules after performing multiple extractions; Correcting the basic data based on the impact prediction model includes: Extracting the impact prediction model of the smart sensor corresponding to the basic data and the operating status of other smart sensors when the basic data is collected; The operating states of other intelligent sensors are pre-processed and then input into the impact prediction model to obtain a change ratio; and the basic data is corrected based on the change ratio.

2. The device operation data acquisition and processing system based on intelligent sensors according to claim 1 is characterized in that: The interpolation processing of the plurality of target data includes: Extracting a number of target data, and determining a target time that needs to be interpolated based on the target data; The target data at the target time is obtained through interpolation processing.

3. The device operation data acquisition and processing system based on intelligent sensors according to claim 2 is characterized in that: The step of determining a target time point for interpolation processing based on a plurality of target data includes: Taking the collection time corresponding to any target data as the judgment time, calculating the number of target data corresponding to the judgment time; Determine whether the number of target data is consistent with the number of smart sensors; if yes, no interpolation processing is required at the judgment moment; if not, the judgment moment is used as the target moment.

4. The device operation data acquisition and processing system based on intelligent sensors according to claim 2 is characterized in that: The step of determining a target time point for interpolation processing based on a plurality of target data includes: Based on the target data, the smart sensor with the worst stability is selected and marked as the reference sensor; the time when the target data corresponding to the reference sensor is collected is used as the judgment time; Calculate the target data quantity corresponding to the judgment moment; determine whether the target data quantity is consistent with the number of smart sensors; if so, no interpolation processing is required at the judgment moment; if not, use the judgment moment as the target moment.

5. The device operation data acquisition and processing system based on intelligent sensors according to claim 4 is characterized in that: Selecting a smart sensor with poor stability based on the target data includes: Constructing a data fluctuation curve based on target data of a plurality of the smart sensors; Based on the data fluctuation curve, the smart sensor with the worst stability is used as a reference sensor.

6. The device operation data acquisition and processing system based on intelligent sensors according to claim 5 is characterized in that: After determining the target moment, the smart sensor corresponding to the missing data at the target moment is identified; and interpolation processing is performed based on the target data collected by the smart sensor to obtain the target data at the target moment.

Images