Data collection frequency optimization method, system, device and storage medium

By constructing a frequency-detection accuracy mapping relationship, the data acquisition frequency of intelligent devices is optimized, solving the hardware pressure problem caused by excessively high data acquisition frequency, and reducing the computational load and operating cost of edge devices without affecting detection accuracy.

CN113986700BActive Publication Date: 2024-11-29HITACHI BUILDING TECH GUANGZHOU CO LTD
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
CN202111175573.4
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-10-09
Publication Date
2024-11-29
Estimated Expiration
2041-10-09

AI Technical Summary

Technical Problem

In existing technologies, the excessively high data collection frequency of smart devices puts a heavy burden on hardware modules, increases data transmission and storage costs, and affects device performance.

Method used

By obtaining the upper limit frequency of smart devices, a frequency-detection accuracy mapping relationship is constructed to optimize the data acquisition frequency in order to reduce the computational load and operating cost of edge devices. The method includes building a pre-trained model and training it at different frequencies to determine the optimal acquisition frequency.

Benefits of technology

Without affecting detection accuracy, reduce the data acquisition frequency and decrease the computational load and operating costs of edge devices.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN113986700B_ABST
    Figure CN113986700B_ABST
Patent Text Reader

Abstract

The present invention discloses a method, system, device and storage medium for optimizing data collection frequency, and relates to the field of monitoring technology. Among them, the method for optimizing data collection frequency includes the following steps: obtaining the upper limit frequency of data collection of intelligent devices; obtaining a plurality of first test collection frequencies of gradient descent including the upper limit frequency according to the upper limit frequency; collecting sample data at the first test collection frequency, and inputting the sample data into the detection model to determine the first detection result accuracy of the detection model; determining the frequency-detection accuracy mapping relationship according to a plurality of the first test collection frequencies and a corresponding plurality of the first detection result accuracies; determining the application collection frequency of the intelligent device application according to the frequency-detection accuracy mapping relationship. The present application can reduce the data collection frequency without affecting the detection accuracy requirements, so as to reduce the computing amount and operating cost of the edge device.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of monitoring technology, and in particular to a method, system, device and storage medium for optimizing data acquisition frequency. Background Art

[0002] With the improvement of science and technology, many companies will adopt smart devices to work and perform related calculations of smart device data at the edge of the enterprise to detect and judge on-site smart devices.

[0003] The edge device loads the pre-trained model, and the on-site smart device collects data in real time and uploads it to the edge device. The edge device analyzes and calculates the collected real-time data to obtain the detection results of the smart device. In the detection process of smart devices, in order to ensure the accuracy of the detection results of smart devices, it is necessary to increase the data collection frequency as much as possible to obtain more data for detection. However, if the data collection frequency is too high, it will put a lot of pressure on the I / O serial port of the smart device, the edge computing device processor, memory and other hardware modules, resulting in high data transmission and storage costs, and will also affect the working performance of the device. Summary of the invention

[0004] The present invention aims to solve at least one of the technical problems existing in the prior art. To this end, the present invention proposes a method, system, device and storage medium for optimizing data collection frequency, which can reduce the data collection frequency without affecting the detection accuracy requirement, so as to reduce the computing amount and operating cost of edge devices.

[0005] On the one hand, an embodiment of the present invention provides a method for optimizing data collection frequency, comprising the following steps:

[0006] Get the upper limit frequency of data collection by smart devices;

[0007] According to the upper limit frequency, a plurality of first trial acquisition frequencies of gradient descent including the upper limit frequency are obtained;

[0008] Collecting sample data at the first test collection frequency, and inputting the sample data into a detection model to determine the accuracy of a first detection result of the detection model;

[0009] Determining a frequency-detection accuracy mapping relationship according to a number of the first test collection frequencies and a corresponding number of the first detection result accuracies;

[0010] The application acquisition frequency of the smart device application is determined according to the frequency-detection accuracy mapping relationship.

[0011] According to some embodiments of the present invention, the detection model is obtained by the following steps:

[0012] Build and initialize the pre-trained model;

[0013] The sample data collected at the upper limit frequency is input into the pre-training model for training to obtain the detection model.

[0014] According to some embodiments of the present invention, inputting the sample data collected at the upper limit frequency into the pre-training model for training to obtain the detection model comprises the following steps:

[0015] Inputting the sample data into the pre-trained model to obtain a second detection result of the pre-trained model;

[0016] Determine the accuracy of the second detection result of the pre-trained model according to the second detection result and the label in the sample data;

[0017] The parameters of the pre-trained model are updated according to the accuracy of the second detection result until the accuracy of the second detection result is greater than the first preset accuracy, and the detection model is obtained according to the current parameters.

[0018] According to some embodiments of the present invention, the sample data is obtained by:

[0019] Get the label type entered by the user;

[0020] When the tag type input by the user is a fault tag, collecting first multiple input data based on the fault state of the smart device, and obtaining the sample data according to the tag type and the first multiple input data;

[0021] When the tag type input by the user is a normal tag, multiple second input data items are collected based on the normal working state of the smart device, and the sample data is obtained according to the tag type and the multiple second input data items.

[0022] According to some embodiments of the present invention, determining the frequency-detection accuracy mapping relationship according to a plurality of the first test acquisition frequencies and a corresponding plurality of the first detection result accuracies comprises the following steps:

[0023] Using a plurality of first test acquisition frequencies of gradient descent as abscissas and a first detection result accuracy corresponding to the first test acquisition frequencies as a ordinates to determine a plurality of discrete coordinate points;

[0024] The frequency-detection accuracy mapping relationship is constructed according to a plurality of the discrete coordinate points.

[0025] According to some embodiments of the present invention, determining the application acquisition frequency of the smart device application according to the frequency-detection accuracy mapping relationship includes the following steps:

[0026] Determine the change rate between all two adjacent discrete coordinate points in the frequency-detection accuracy mapping relationship;

[0027] Select two adjacent discrete coordinate points with the largest change rate, and compare the ordinates of the selected two adjacent discrete coordinate points, take the larger ordinate value as the third test result accuracy and use the abscissa value corresponding to the larger ordinate as the second test acquisition frequency;

[0028] When the third detection result accuracy is greater than the second preset accuracy, the second test acquisition frequency is used as the application acquisition frequency.

[0029] According to some embodiments of the present invention, determining the application acquisition frequency of the smart device application according to the frequency-detection accuracy mapping relationship further includes the following steps:

[0030] When the third detection result accuracy is less than the second preset accuracy, the third test acquisition frequency corresponding to the second preset accuracy is determined according to the frequency-detection accuracy mapping relationship, and the third test acquisition frequency is used as the application acquisition frequency.

[0031] On the other hand, an embodiment of the present invention further provides a system for optimizing data acquisition frequency, including:

[0032] The first module is used to obtain the upper limit frequency of data collected by the smart device;

[0033] The second module is used to obtain a plurality of first test acquisition frequencies of gradient descent including the upper limit frequency according to the upper limit frequency;

[0034] A third module is used to collect sample data at the first test collection frequency, and input the sample data into the detection model to determine the accuracy of the first detection result of the detection model;

[0035] A fourth module is used to determine a frequency-detection accuracy mapping relationship according to a number of the first test collection frequencies and a corresponding number of the first detection result accuracies;

[0036] The fifth module is used to determine the application collection frequency of the smart device application according to the frequency-detection accuracy mapping relationship.

[0037] On the other hand, an embodiment of the present invention further provides a device for optimizing data acquisition frequency, comprising:

[0038] at least one processor;

[0039] at least one memory for storing at least one program;

[0040] When the at least one program is executed by the at least one processor, the at least one processor implements the method for optimizing the data acquisition frequency as described above.

[0041] On the other hand, an embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to enable a computer to execute the data acquisition frequency optimization method as described above.

[0042] The above technical solution of the present invention has at least one of the following advantages or beneficial effects: obtaining the upper limit frequency of the data collected by the intelligent device, and obtaining several first test collection frequencies of gradient descent including the upper limit frequency based on the upper limit frequency. Then collect sample data at different first test collection frequencies, and input the sample data into the detection model to determine the accuracy of the first detection result of the detection model. Determine the frequency-detection accuracy mapping relationship according to several first test collection frequencies and corresponding several first detection result accuracies. Determine the application collection frequency of the intelligent device application according to the frequency-detection accuracy mapping relationship. By obtaining sample data at multiple different first test sampling frequencies, the accuracy of the first detection result of the detection model at different first test sampling frequencies is determined, and then the frequency-detection accuracy mapping relationship is determined. The final collection frequency is optimized based on the frequency-detection accuracy mapping relationship, so that the data collection frequency can be reduced without affecting the detection accuracy requirements, so as to reduce the computing amount and operating cost of the edge device. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 is a flow chart of a method for optimizing data acquisition frequency provided by an embodiment of the present invention;

[0044] Figure 2 is a schematic diagram of a data acquisition frequency optimization system provided by an embodiment of the present invention;

[0045] Figure 3 It is a schematic diagram of a data acquisition frequency optimization device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0046] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar components or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0047] In the description of the present invention, it should be understood that descriptions involving orientation, such as up, down, left, right, etc., the orientations or positional relationships indicated are based on the orientations or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as a limitation on the present invention.

[0048] In the description of the present invention, if there is a description of first, second, etc., it is only for the purpose of distinguishing technical features, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features or implicitly indicating the sequence of the indicated technical features.

[0049] The embodiment of the present invention provides a method for optimizing data acquisition frequency, referring to Figure 1 The method for optimizing the data acquisition frequency of the embodiment of the present invention includes but is not limited to step S110, step S120, step S130, step S140 and step S150.

[0050] Step S110, obtaining the upper limit frequency of data collection by the smart device.

[0051] In some embodiments, the upper limit frequency of data collection by the smart device can be set according to the data transmission load of the smart device and the maximum computing power of the edge device. For example, the edge device is connected to the smart device on site, and the user inputs the upper limit frequency to the edge device through an interactive device such as a keyboard or a touch screen. The edge device can send the upper limit frequency to the smart device, thereby controlling the smart device to collect data according to the upper limit frequency.

[0052] Step S120, obtaining a plurality of first trial acquisition frequencies of gradient descent including the upper limit frequency according to the upper limit frequency.

[0053] In some embodiments, the upper limit frequency is the maximum frequency at which data can be collected based on the actual conditions of the smart device and the edge device. Taking the upper limit frequency as the maximum value, the first test collection frequency of several gradient descents can be determined. Specifically, the first test collection frequency can be determined by the following formula:

[0054] a n =a1+(n-1)×d;

[0055] Among them, a n represents the nth first test acquisition frequency, a1 represents the upper limit frequency, d is the decreasing parameter set by the user, and d<0, n={1,2,3,4,......,k}, k can be set by the number of first test acquisition frequencies required by the user.

[0056] Step S130, collecting sample data at a first test collection frequency, and inputting the sample data into a detection model to determine the accuracy of a first detection result of the detection model.

[0057] Step S140: determining a frequency-detection accuracy mapping relationship according to a plurality of first test collection frequencies and a plurality of corresponding first detection result accuracies.

[0058] Step S150: determining the application collection frequency of the smart device application according to the frequency-detection accuracy mapping relationship.

[0059] In some embodiments, the edge device first controls the smart device to collect sample data at a first first test collection frequency and inputs the sample data into the detection model to obtain the first first detection result accuracy corresponding to the detection model, and then controls the smart device to collect sample data at a second first test collection frequency and inputs the sample data into the detection model to obtain the second first detection result accuracy corresponding to the detection model, and so on, each first test collection frequency corresponds to a first detection result accuracy. The frequency-detection accuracy mapping relationship is determined based on a number of first test collection frequencies and their corresponding first detection result accuracies, and then the appropriate application collection frequency is analyzed and calculated based on the frequency-detection accuracy mapping relationship, and the application collection frequency is sent to the smart device. The smart device collects data according to the application collection frequency and uploads the collected data to the detection model for detection, which can reduce the data collection frequency without affecting the detection accuracy requirements, so as to reduce the computing power and operating costs of the edge device.

[0060] Another embodiment of the present invention further provides a method for obtaining the detection model in step S130 , and the method for obtaining the detection model includes but is not limited to step S210 and step S220 .

[0061] Step S210, construct and initialize a pre-trained model.

[0062] Step S220, inputting the sample data collected at the upper limit frequency into the pre-training model for training to obtain a detection model.

[0063] Furthermore, step S220 also includes but is not limited to step S310, step S320 and step S330.

[0064] Step S310, inputting the sample data into the pre-trained model to obtain a second detection result of the pre-trained model.

[0065] Step S320, determining the accuracy of the second detection result of the pre-trained model according to the second detection result and the label in the sample data.

[0066] Step S330, update the parameters of the pre-trained model according to the accuracy of the second detection result until the accuracy of the second detection result is greater than the first preset accuracy, and then obtain the detection model according to the current parameters.

[0067] In some embodiments, after the pre-training model is constructed and initialized, the edge device controls the smart device to collect sample data at the upper limit frequency, and trains the pre-training model to obtain a detection model. Collecting sample data at the upper limit frequency to train the pre-training model can collect as much sample data as possible in the same time, thereby improving the training efficiency of the model.

[0068] In some embodiments, after the sample data is input into the initialized pre-trained model, the detection result output by the pre-trained model can be obtained, and the accuracy of the pre-trained model detection can be evaluated according to the detection result and the label in the sample data, that is, the second detection result accuracy, so as to update the parameters of the model. For the pre-trained model, the accuracy of the model detection result can be measured by the loss function, which is defined on a single sample data and used to measure the detection error of a sample data. Specifically, the loss value of the sample data is determined by the label of the single sample data and the detection result of the model on the sample data. In actual training, a lot of sample data will be input in one training process, so the cost function is generally used to measure the overall error of multiple sample data. The cost function is defined on a certain amount of sample data, and is used to calculate the average value of the prediction error of all sample data, which can better measure the detection effect of the model. For general machine learning models, based on the aforementioned cost function, plus the regularization term that measures the complexity of the model, it can be used as the objective function of the training, and the loss value of the entire sample data can be calculated based on the objective function. There are many types of commonly used loss functions, such as 0-1 loss function, square loss function, absolute loss function, logarithmic loss function, cross entropy loss function, etc., which can all be used as loss functions of machine learning models, which will not be elaborated one by one here. In the embodiment of the present application, any loss function can be selected to determine the loss value of the training. Based on the loss value of the training, the parameters of the model are updated using the back propagation algorithm, and a trained pre-trained model, i.e., a detection model, can be obtained by iterating several rounds. Specifically, the number of iterations can be pre-set, or the training is considered to be completed when the detection model meets the accuracy requirements, i.e., when the first preset accuracy is reached.

[0069] Another embodiment of the present invention further provides a method for obtaining sample data in step S130 , and the method for obtaining sample data includes but is not limited to step S410 , step S420 , and step S430 .

[0070] Step S410: obtaining the tag type input by the user.

[0071] Step S420: When the tag type input by the user is a fault tag, first multiple input data are collected based on the fault state of the smart device, and sample data is obtained according to the tag type and the first multiple input data.

[0072] Step S430: When the tag type input by the user is a normal tag, multiple second input data items are collected based on the normal working state of the smart device, and sample data is obtained according to the tag type and the multiple second input data items.

[0073] In some embodiments, a user can input a fault label into a smart device and then manually create a fault condition for the smart device, so that the smart device collects multiple input data under the fault condition and uploads them to an edge device, and generates sample data together with the fault label and the first multiple input data.

[0074] In some embodiments, the user can input a normal label into the smart device and then start the smart device normally, so that the smart device collects multiple input data under normal working conditions and uploads them to the edge device, and generates sample data together with the normal label and the second multiple input data.

[0075] In some embodiments, the first multiple input data or the second multiple data may include input current, voltage, action duration, etc. when the smart device is running. For example, if the smart device is an elevator, it is necessary to collect data related to the elevator door motor, and the first multiple input data or the second multiple input data may include current, voltage, door opening and closing duration, action distance, torque, etc.

[0076] According to some specific embodiments of the present invention, step S140 includes but is not limited to step S510 and step S520.

[0077] Step S510, using a plurality of first test acquisition frequencies of gradient descent as abscissas and using the first detection result accuracy corresponding to the first test acquisition frequencies as a ordinates to determine a plurality of discrete coordinate points.

[0078] Step S520: construct a frequency-detection accuracy mapping relationship according to a plurality of discrete coordinate points.

[0079] In some embodiments, several first test acquisition frequencies with gradient descent are used as horizontal coordinates, and the first detection result accuracy corresponding to the first test acquisition frequency is used as vertical coordinates to determine several discrete coordinate points, and then the several discrete coordinate points are connected to construct a frequency-detection accuracy relationship line graph.

[0080] In some implementations, several first test acquisition frequencies with gradient descent are used as the horizontal coordinates, and the first detection result accuracy corresponding to the first test acquisition frequency is used as the vertical coordinate to determine several discrete coordinate points, and then the discrete coordinate points are fitted by curve fitting methods such as interpolation and least squares method, so as to obtain a function expression of frequency-detection accuracy.

[0081] According to some specific embodiments of the present invention, step S150 includes but is not limited to step S610, step S620, step S630 and step S640.

[0082] Step S610, determining the change rate between all two adjacent discrete coordinate points in the frequency-detection accuracy mapping relationship.

[0083] Step S620, select two adjacent discrete coordinate points with the largest change rate, and compare the vertical coordinates of the selected two adjacent discrete coordinate points, take the value of the larger vertical coordinate as the third detection result accuracy and use the horizontal coordinate value corresponding to the larger vertical coordinate as the second test acquisition frequency.

[0084] Step S630: When the third detection result accuracy is greater than the second preset accuracy, the second test acquisition frequency is used as the application acquisition frequency.

[0085] Step S640: When the third detection result accuracy is less than the second preset accuracy, a third test acquisition frequency corresponding to the second preset accuracy is determined according to a frequency-detection accuracy mapping relationship, and the third test acquisition frequency is used as the application acquisition frequency.

[0086] In some embodiments, when the frequency-detection accuracy mapping relationship is a frequency-detection accuracy relationship line graph, determine the rate of change between all two adjacent discrete coordinate points in the frequency-detection accuracy mapping relationship, that is, the slope of the line segment between the two adjacent discrete coordinate points. Select the two adjacent discrete coordinate points with the largest rate of change, that is, the two adjacent discrete coordinate points where the accuracy of the detection result decreases the fastest, compare the ordinates of the selected two adjacent discrete coordinate points, take the value of the larger ordinate as the third detection result accuracy and use the value of the abscissa corresponding to the larger ordinate as the second test acquisition frequency. When the third detection result accuracy is greater than the user's second preset accuracy, the second test acquisition frequency is used as the application acquisition frequency. In this embodiment, the critical point where the accuracy of the detection result decreases the fastest is found according to the frequency-detection accuracy line graph, and the acquisition frequency corresponding to the position is applied to the smart device, so that when the smart device is detected, the acquisition frequency is appropriately reduced while the accuracy of the detection result is guaranteed as much as possible, so as to reduce the computing amount and operating cost of the edge device. Furthermore, when the accuracy of the third detection result is less than the second preset accuracy, the third test collection frequency corresponding to the second preset accuracy is estimated according to the frequency-detection accuracy mapping line chart, and the third test collection frequency is used as the application collection frequency, so that the smart device can collect data according to the collection frequency corresponding to the acceptable second preset accuracy set by the user.

[0087] According to some specific embodiments of the present invention, when the frequency-detection accuracy mapping relationship is a function expression of frequency-detection accuracy, the derivative expression of the function expression is calculated, and the critical point where the detection accuracy drops the fastest is determined based on the derivative expression, and the ordinate of the critical point is used as the third detection result accuracy, and the abscissa of the critical point is used as the second test acquisition frequency. When the third detection result accuracy is greater than the user's second preset accuracy, the second test acquisition frequency is used as the application acquisition frequency. When the third detection result accuracy is less than the second preset accuracy, the third test acquisition frequency corresponding to the second preset accuracy is calculated based on the function expression of frequency-detection accuracy, and the third test acquisition frequency is used as the application acquisition frequency.

[0088] The embodiment of the present invention also provides a data acquisition frequency optimization system, referring to Figure 2 , the data acquisition frequency optimization system includes a first module, a second module, a third module, a fourth module and a fifth module.

[0089] The first module is used to obtain the upper limit frequency of data collected by the smart device.

[0090] The second module is used to obtain a plurality of first test acquisition frequencies of gradient descent including the upper limit frequency according to the upper limit frequency.

[0091] The third module is used to collect sample data at a first test collection frequency, and input the sample data into the detection model to determine the accuracy of the first detection result of the detection model.

[0092] The fourth module is used to determine a frequency-detection accuracy mapping relationship according to a plurality of first test acquisition frequencies and a plurality of corresponding first detection result accuracies.

[0093] The fifth module is used to determine the application collection frequency of the smart device application according to the frequency-detection accuracy mapping relationship.

[0094] It can be understood that the contents of the above-mentioned data acquisition frequency optimization method embodiment are applicable to the present system embodiment, the functions specifically implemented by the present system embodiment are the same as those in the data acquisition frequency optimization method embodiment, and the beneficial effects achieved are also the same as those achieved by the above-mentioned data acquisition frequency optimization method embodiment.

[0095] Reference Figure 3 , Figure 3 Schematic diagram of a data acquisition frequency optimization device provided by an embodiment of the present invention. The data acquisition frequency optimization device of the embodiment of the present invention includes one or more control processors and a memory. Figure 3 A control processor and a memory are taken as an example.

[0096] The control processor and the memory can be connected via a bus or other means. Figure 3 The example of connecting through bus is taken in the following.

[0097] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely arranged relative to the control processor, and these remote memories may be connected to the optimization device of the data acquisition frequency via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0098] Those skilled in the art will understand that Figure 3 The device structure shown in the figure does not constitute a limitation on the device for optimizing the data acquisition frequency, and may include more or less components than those shown in the figure, or a combination of certain components, or a different arrangement of components.

[0099] The non-transient software program and instructions required to implement the data acquisition frequency optimization method applied to the data acquisition frequency optimization device in the above-mentioned embodiment are stored in the memory, and when executed by the control processor, the data acquisition frequency optimization method applied to the data acquisition frequency optimization device in the above-mentioned embodiment is executed.

[0100] In addition, an embodiment of the present invention further provides a computer-readable storage medium, which stores computer-executable instructions, and the computer-executable instructions are executed by one or more control processors, so that the one or more control processors can execute the data acquisition frequency optimization method in the above method embodiment.

[0101] It will be appreciated by those skilled in the art that all or some of the steps and systems in the disclosed method above may be implemented as software, firmware, hardware and appropriate combinations thereof. Some physical components or all physical components may be implemented as software executed by a processor, such as a central processing unit, a digital signal processor or a microprocessor, or may be implemented as hardware, or may be implemented as an integrated circuit, such as an application specific integrated circuit. Such software may be distributed on a computer-readable medium, which may include a computer storage medium (or a non-transitory medium) and a communication medium (or a temporary medium). As known to those skilled in the art, the term computer storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules or other data). Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tapes, disk storage or other magnetic storage devices, or any other medium that may be used to store desired information and may be accessed by a computer. Furthermore, it is well known to those skilled in the art that communication media typically embodies computer readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media.

[0102] The embodiments of the present invention are described in detail above with reference to the accompanying drawings, but the present invention is not limited to the above embodiments, and various changes can be made within the knowledge scope of ordinary technicians in the relevant technical field without departing from the purpose of the present invention.

Claims

1. A method for optimizing data acquisition frequency, characterized in that: The following steps are involved: Get the upper limit frequency of data collection by smart devices; According to the upper limit frequency, a plurality of first trial acquisition frequencies of gradient descent including the upper limit frequency are obtained; Collecting sample data at the first test collection frequency, and inputting the sample data into a detection model to determine the accuracy of a first detection result of the detection model; Determining a frequency-detection accuracy mapping relationship according to a number of the first test collection frequencies and a corresponding number of the first detection result accuracies; Determine the application acquisition frequency of the smart device application according to the frequency-detection accuracy mapping relationship; When the frequency-detection accuracy mapping relationship is a frequency-detection accuracy relationship line graph, determining the application acquisition frequency of the smart device application according to the frequency-detection accuracy mapping relationship includes: Determine the change rate between all two adjacent discrete coordinate points in the frequency-detection accuracy mapping relationship; Select two adjacent discrete coordinate points with the largest change rate, compare the ordinates of the selected two adjacent discrete coordinate points, take the value of the larger ordinate as the third test result accuracy, and use the value of the abscissa corresponding to the larger ordinate as the second test acquisition frequency; Or when the frequency-detection accuracy mapping relationship is a function expression of frequency-detection accuracy, determining the application acquisition frequency of the smart device application according to the frequency-detection accuracy mapping relationship includes: Determining the critical point at which the accuracy of the detection result decreases most rapidly according to the frequency-detection accuracy mapping relationship; The corresponding third detection result accuracy and second test acquisition frequency are determined according to the critical point, wherein the ordinate of the critical point is used as the third detection result accuracy, and the abscissa of the critical point is used as the second test acquisition frequency.

2. The data acquisition frequency optimization method according to claim 1, characterized in that: The detection model is obtained by the following steps: Build and initialize the pre-trained model; The sample data collected at the upper limit frequency is input into the pre-training model for training to obtain the detection model.

3. The data acquisition frequency optimization method according to claim 2, characterized in that: The step of inputting the sample data collected at the upper frequency limit into the pre-training model for training to obtain the detection model comprises the following steps: Inputting the sample data into the pre-trained model to obtain a second detection result of the pre-trained model; Determine the accuracy of the second detection result of the pre-trained model according to the second detection result and the label in the sample data; The parameters of the pre-trained model are updated according to the accuracy of the second detection result until the accuracy of the second detection result is greater than the first preset accuracy, and the detection model is obtained according to the current parameters.

4. The data acquisition frequency optimization method according to claim 1, characterized in that: The sample data is obtained in the following way: Get the label type entered by the user; When the tag type input by the user is a fault tag, collecting first multiple input data based on the fault state of the smart device, and obtaining the sample data according to the tag type and the first multiple input data; When the tag type input by the user is a normal tag, multiple second input data items are collected based on the normal working state of the smart device, and the sample data is obtained according to the tag type and the multiple second input data items.

5. The data acquisition frequency optimization method according to claim 1, characterized in that: Determining the frequency-detection accuracy mapping relationship according to a plurality of the first test collection frequencies and a corresponding plurality of the first detection result accuracies comprises the following steps: Using a plurality of first test acquisition frequencies of gradient descent as abscissas and a first detection result accuracy corresponding to the first test acquisition frequencies as a ordinates to determine a plurality of discrete coordinate points; The frequency-detection accuracy mapping relationship is constructed according to a plurality of the discrete coordinate points.

6. The data acquisition frequency optimization method according to claim 5, characterized in that: The data acquisition frequency optimization method further comprises the following steps: When the third detection result accuracy is greater than the second preset accuracy, the second test acquisition frequency is used as the application acquisition frequency; Alternatively, when the third detection result accuracy is less than the second preset accuracy, a third test acquisition frequency corresponding to the second preset accuracy is determined according to the frequency-detection accuracy mapping relationship, and the third test acquisition frequency is used as the application acquisition frequency.

7. A data acquisition frequency optimization system, characterized in that: include: The first module is used to obtain the upper limit frequency of data collected by the smart device; The second module is used to obtain a plurality of first test acquisition frequencies of gradient descent including the upper limit frequency according to the upper limit frequency; A third module is used to collect sample data at the first test collection frequency, and input the sample data into the detection model to determine the accuracy of the first detection result of the detection model; A fourth module is used to determine a frequency-detection accuracy mapping relationship according to a number of the first test collection frequencies and a corresponding number of the first detection result accuracies; The fifth module is used to determine the application acquisition frequency of the smart device application according to the frequency-detection accuracy mapping relationship; wherein, when the frequency-detection accuracy mapping relationship is a frequency-detection accuracy relationship line graph, the fifth module is specifically used to perform the following steps: Determine the change rate between all two adjacent discrete coordinate points in the frequency-detection accuracy mapping relationship; Select two adjacent discrete coordinate points with the largest change rate, compare the ordinates of the selected two adjacent discrete coordinate points, take the value of the larger ordinate as the third test result accuracy, and use the value of the abscissa corresponding to the larger ordinate as the second test acquisition frequency; Or when the frequency-detection accuracy mapping relationship is a function expression of frequency-detection accuracy, the fifth module is specifically used to perform the following steps: Determining the critical point at which the accuracy of the detection result decreases most rapidly according to the frequency-detection accuracy mapping relationship; The corresponding third detection result accuracy and second test acquisition frequency are determined according to the critical point, wherein the ordinate of the critical point is used as the third detection result accuracy, and the abscissa of the critical point is used as the second test acquisition frequency.

8. A device for optimizing data acquisition frequency, characterized in that: include: at least one processor; at least one memory for storing at least one program; When the at least one program is executed by the at least one processor, the at least one processor implements the method for optimizing the data acquisition frequency as described in any one of claims 1 to 6.

9. A computer-readable storage medium storing a program executable by a processor, characterized in that: The program executable by the processor is used to implement the method for optimizing the data acquisition frequency as claimed in any one of claims 1 to 6 when executed by the processor.

Citation Information

Patent Citations

  • Displacement detection sensor sampling frequency and detection accuracy design method

    CN111102950A

  • Data acquisition strategy determination method and device, storage medium and electronic equipment

    CN111796979A

  • Target detection model training method and device, electronic equipment and storage medium

    CN111832614A

  • Data acquisition method, device and system

    CN113129473A

  • Calibrating reliability of multi-label classification neural networks

    US20210287081A1