Precision Data Acquisition System Based on Multi-parameter Sensors

Abstract

The present invention relates to the field of data acquisition and processing, and specifically to a precision data acquisition system based on a multi-parameter sensor, comprising a sensor module, a data acquisition module, a data processing module, a data storage module, a data analysis module, a user interface module and a central control module; the present invention adopts an intelligent perception method based on continuous multi-parameter data to clean the data, performs object detection on the multi-parameter data through an intelligent super-perception model, removes redundant low-precision data, and simultaneously determines the optimal model parameters by fusing multiple parameters, improves the accuracy of redundant data recognition, and improves the data acquisition rate of the multi-parameter sensor by removing low-precision fragment frames, thereby reducing the system calculation cost; the present invention adopts an automatically generated context model to supplement the information of high-precision data and make up for the problem of missing fragment frames, so that the high-precision data set has more integrity, improves the data quality, and retains more effective precision data.

Description

Technical Field

[0001] The invention belongs to the field of data acquisition and processing, and in particular relates to a precision data acquisition system based on a multi-parameter sensor. Background Art

[0002] Multi-parameter data acquisition refers to the process of collecting multiple different parameters or multiple types of data at the same time. This data acquisition can be achieved by using multiple sensors, each of which is responsible for measuring a specific parameter. Through multi-parameter data acquisition, more comprehensive and integrated information about the system, environment or process can be obtained. However, when multi-parameter sensors are collecting data, a large amount of redundant data and low-precision data will be generated, resulting in low precision of the collected multi-parameter data. The main method for collecting multi-parameter sensor data is simple multi-parameter acquisition, and the method for precise data processing is mainly simple data filtering and empirical data calibration. The existing methods are difficult to quickly analyze and accurately process redundant data and high-dimensional data of multi-parameter data. Summary of the invention

[0003] In view of the problem that the prior art is difficult to quickly analyze and accurately process redundant data and high-dimensional data of multi-parameter data, the present invention adopts an intelligent perception method based on continuous multi-parameter data to clean the data, and uses an intelligent super-perception model to detect objects in multi-parameter data, and efficiently identifies and removes redundant low-precision data. At the same time, by fusing multiple parameters, the optimal recognition parameters are determined, the accuracy of redundant data recognition is improved, and the data acquisition rate of multi-parameter sensors is increased by removing low-precision fragment frames, thereby reducing the system computing cost. The present invention adopts an automatically generated context model to supplement the information of high-precision data, that is, frame supplementation, to make up for the low data continuity and feature missing problems caused by removing low-precision fragment frames, so that the high-precision data set is more complete, the data quality is further improved, and more effective and precise data is retained.

[0004] The multi-parameter sensor-based precision data acquisition system provided by the present invention comprises a sensor module, a data acquisition module, a data processing module, a data storage module, a data analysis module, a user interface module and a central control module;

[0005] The sensor module uses a sensor to collect multi-parameter data and transmits the multi-parameter data to a data acquisition module;

[0006] The data acquisition module collects and caches the multi-parameter data and transmits the multi-parameter data to the data processing module;

[0007] The data processing module cleans, calibrates and preprocesses the multi-parameter data to obtain processed data, and transmits the processed data to the data storage module and the data analysis module;

[0008] The data storage module stores and manages the processed data;

[0009] The data analysis module analyzes and studies the processed data and extracts data features;

[0010] The user interface module provides a user interaction interface, where users can perform operations and access data;

[0011] The central control module manages and controls the operation and configuration of the entire data acquisition system.

[0012] Furthermore, in the data processing module, the multi-parameter data is cleaned, calibrated and preprocessed to obtain processed data, which specifically includes the following steps:

[0013] Step S1: Data classification, classifying the multi-parameter data to obtain low-precision multi-parameters and high-precision multi-parameters, converting the low-precision multi-parameters using a low-precision ADC module to obtain low-precision digital data, and converting the high-precision multi-parameters using a high-precision ADC module to obtain high-precision digital data;

[0014] Step S2: intelligent data cleaning, using an intelligent perception method based on continuous multi-parameter data to clean the digital data to obtain cleaned data;

[0015] Further, step S2 specifically includes the following steps:

[0016] Step S21: fragment model processing, using the fragment model to predict low-precision digital data to obtain a prediction score;

[0017] Further, step S21 specifically includes the following steps:

[0018] Step S211: forward sampling, randomly adopting a different sampling frequencies to perform forward sampling on the low-precision digital data, and flattening the a kinds of forward sampling results into one-dimensional vectors and then superimposing them to obtain a forward vector;

[0019] Step S212: negative sampling, randomly using a different sampling frequencies to perform negative sampling on the low-precision digital data, and flattening the a positive sampling results into a one-dimensional vector and then superimposing them to obtain a negative vector;

[0020] Step S213: Data encoding, using an encoding function to map the positive vector and the negative vector to a high-dimensional space to obtain a positive supervector and a negative supervector. The formula used is as follows:

[0021] ;

[0022] In the formula, The vector representation of the positive vector and the negative vector is ={F1, F2, ..., Fn}, {F1, F2, ..., Fn} represents each parameter in the multi-parameter, is an n×D matrix, where All elements in are sampled from an iid Gaussian distribution. is an element sampled from an iid uniform distribution, These are the calculated positive and negative supervectors;

[0023] Step S214: encoding training, performing multiple data encoding operations on the positive supervector and the negative supervector to obtain the final positive supervector and the negative supervector, normalizing them, and using an activation function to perform prediction to obtain a prediction score;

[0024] Step S22: Perception model processing, using an intelligent super perception model to clean low-precision digital data to obtain super-sensing high-precision data;

[0025] Further, step S22 specifically includes the following steps:

[0026] Step S221: setting hyperparameters, setting T score, T detection value and initial stride, where the initial stride is 0;

[0027] Step S222: Data clipping, performing sliding window clipping on the low-precision digital data to obtain m data segments;

[0028] Step S223: fragment processing, repeating step S21, performing fragment model processing on m data fragments, and obtaining m prediction scores;

[0029] Step S224: score judgment, compare the m predicted scores with the T score, if the predicted score is greater than the T score, then the initial stride is increased by 1, and the T score is tested on the m predicted scores in turn to obtain the final stride number;

[0030] Step S225: Data cleaning, comparing the final stride number with the T detection value. If the final stride number is greater than the T detection value, it means that the low-precision digital data contains interference information, and the low-precision digital data is removed. If the final stride number is less than the T detection value, it means that the low-precision digital data is actually high-precision data, and the low-precision digital data is converted into super-sensing high-precision data, that is, cleaned data.

[0031] Step S3: parameter optimization, optimizing the parameters in the intelligent perception method based on continuous multi-parameter data to obtain the final cleaned data;

[0032] Step S4: data classification processing, defining the final cleaned data and high-precision digital data as high-precision data, and using the method of automatically generating a context model to complete the information of the high-precision data classification, i.e., fill in the frame, to obtain processed data;

[0033] Further, step S4 specifically includes the following steps:

[0034] Step S41: data classification, dividing the processed data according to whether it is IoT data, and obtaining IoT data and non-IoT data;

[0035] Step S42: IOT data processing, repeating step S213, performing data encoding on the IOT data to obtain a hyper-dimensional IOT vector;

[0036] Step S43: non-IOT data processing, using a cluster analysis method to divide the non-IOT data into different data tuples;

[0037] Step S44: context model processing, using the context model to perform information completion operations on the hyperdimensional IOT vector and all data tuples respectively to obtain processed data.

[0038] The beneficial results achieved by the present invention using the above scheme are as follows:

[0039] (1) In view of the problem that it is difficult to quickly analyze and accurately process redundant data and high-dimensional data of multi-parameter data in the prior art, the present invention adopts an intelligent perception method based on continuous multi-parameter data to clean the data, and uses an intelligent super-perception model to detect objects in multi-parameter data, and efficiently identify and remove redundant low-precision data. At the same time, by fusing multiple parameters, the optimal recognition parameters are determined, the accuracy of redundant data recognition is improved, and by removing low-precision fragment frames, the data acquisition rate of multi-parameter sensors is improved, and the system computing cost is reduced;

[0040] (2) The present invention adopts an automatically generated context model to supplement the high-precision data, that is, frame filling processing, to make up for the low data continuity and feature missing problems caused by removing low-precision fragment frames, so that the high-precision data set is more complete, further improves the data quality, and retains more effective and precise data. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 A module diagram of a precision data acquisition system based on a multi-parameter sensor provided by the present invention;

[0042] Figure 2 Schematic diagram of the process of processing multi-parameter data by the data processing module.

[0043] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention. DETAILED DESCRIPTION

[0044] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0045] Example 1, see Figure 1 The precision data acquisition system based on the multi-parameter sensor provided by the present invention includes a sensor module, a data acquisition module, a data processing module, a data storage module, a data analysis module, a user interface module and a central control module;

[0046] The sensor module uses a sensor to collect multi-parameter data and transmits the multi-parameter data to a data acquisition module;

[0047] The data acquisition module collects and caches the multi-parameter data and transmits the multi-parameter data to the data processing module;

[0048] The data processing module cleans, calibrates and preprocesses the multi-parameter data to obtain processed data, and transmits the processed data to the data storage module and the data analysis module;

[0049] The data storage module stores and manages the processed data;

[0050] The data analysis module analyzes and studies the processed data and extracts data features;

[0051] The user interface module provides a user interaction interface, where users can perform operations and access data;

[0052] The central control module manages and controls the operation and configuration of the entire data acquisition system.

[0053] Example 2, see Figure 2 Based on the above embodiment, this embodiment cleans, calibrates and preprocesses the multi-parameter data in the data processing module to obtain processed data, which specifically includes the following steps:

[0054] Step S1: Data classification, classifying the multi-parameter data to obtain low-precision multi-parameters and high-precision multi-parameters, converting the low-precision multi-parameters using a low-precision ADC module to obtain low-precision digital data, and converting the high-precision multi-parameters using a high-precision ADC module to obtain high-precision digital data;

[0055] Step S2: intelligent data cleaning, using an intelligent perception method based on continuous multi-parameter data to clean the digital data to obtain cleaned data;

[0056] Step S3: parameter optimization, optimizing the parameters in the intelligent perception method based on continuous multi-parameter data to obtain the final cleaned data;

[0057] Step S4: Data classification processing, defining the final cleaned data and high-precision digital data as high-precision data, and using the method of automatically generating a context model to complete the information of the high-precision data classification, i.e., perform frame completion processing, to obtain processed data.

[0058] By performing the above operations, in view of the problem that the existing technology is difficult to quickly analyze and accurately process redundant data and high-dimensional data of multi-parameter data, the present invention adopts an intelligent perception method based on continuous multi-parameter data to clean the data, and uses an intelligent super-perception model to detect objects in multi-parameter data, and efficiently identify and remove redundant low-precision data. At the same time, by fusing multiple parameters, the optimal recognition parameters are determined to improve the accuracy of redundant data recognition, and by removing low-precision fragment frames, the data acquisition rate of multi-parameter sensors is improved, thereby reducing the system computing cost.

[0059] Embodiment 3: This embodiment is based on the above embodiment, and step S2 specifically includes the following steps:

[0060] Step S21: fragment model processing, using the fragment model to predict the low-precision digital data to obtain a prediction score, specifically including the following steps:

[0061] Step S211: forward sampling, randomly adopting a different sampling frequencies to perform forward sampling on the low-precision digital data, and flattening the a kinds of forward sampling results into one-dimensional vectors and then superimposing them to obtain a forward vector;

[0062] Step S212: negative sampling, randomly using a different sampling frequencies to perform negative sampling on the low-precision digital data, and flattening the a positive sampling results into a one-dimensional vector and then superimposing them to obtain a negative vector;

[0063] Step S213: Data encoding, using an encoding function to map the positive vector and the negative vector to a high-dimensional space to obtain a positive supervector and a negative supervector. The formula used is as follows:

[0064] ;

[0065] In the formula, The vector representation of the positive vector and the negative vector is ={F1, F2, ..., Fn}, {F1, F2, ..., Fn} represents each parameter in the multi-parameter, including temperature, gas concentration, location and timestamp, etc. is an n×D matrix, where All elements in are sampled from an iid Gaussian distribution. is an element sampled from an iid uniform distribution, These are the calculated positive and negative supervectors;

[0066] Step S214: encoding training, performing multiple data encoding operations on the positive supervector and the negative supervector to obtain the final positive supervector and the negative supervector, normalizing them, and using an activation function to perform prediction to obtain a prediction score;

[0067] Step S22: Perception model processing, using an intelligent super perception model to clean low-precision digital data to obtain super-sensing high-precision data, specifically including the following steps:

[0068] Step S221: setting hyper parameters, setting T score, T detection value and initial stride, where the initial stride is 0;

[0069] Step S222: Data clipping, performing sliding window clipping on the low-precision digital data to obtain m data segments;

[0070] Step S223: fragment processing, repeating step S21, performing fragment model processing on m data fragments, and obtaining m prediction scores;

[0071] Step S224: score judgment, compare the m predicted scores with the T score, if the predicted score is greater than the T score, then the initial stride is increased by 1, and the T score is tested on the m predicted scores in turn to obtain the final stride number;

[0072] Step S225: Data cleaning, compare the final stride number with the T detection value. If the final stride number is greater than the T detection value, it means that the low-precision digital data contains interference information, and the low-precision digital data is removed. If the final stride number is less than the T detection value, it means that the low-precision digital data is actually high-precision data, and the low-precision digital data is converted into super-sensing high-precision data, that is, cleaned data.

[0073] Embodiment 4: This embodiment is based on the above embodiment, and step S4 specifically includes the following steps:

[0074] Step S41: data classification, dividing the processed data according to whether it is IoT data, and obtaining IoT data and non-IoT data;

[0075] Step S42: IOT data processing, repeating step S213, performing data encoding on the IOT data to obtain a hyper-dimensional IOT vector;

[0076] Step S43: non-IOT data processing, using a cluster analysis method to divide the non-IOT data into different data tuples;

[0077] Step S44: context model processing, using the context model to perform information completion operations on the hyperdimensional IOT vector and all data tuples respectively to obtain processed data.

[0078] By performing the above operations, the present invention uses an automatically generated context model to supplement the high-precision data with information, that is, frame filling processing, to make up for the low data continuity and feature missing problems caused by removing low-precision fragment frames, so that the high-precision data set is more complete, further improves the data quality, and retains more effective and precise data.

[0079] Embodiment 5: This embodiment is based on the above embodiment, and the context model specifically includes the following details:

[0080] Model selection and training: Select an RNN-based context prediction model, use existing parameter data and context information for model training, and learn the relationship between parameters and the impact of context.

[0081] Context construction: Based on the missing parameter values, the corresponding context information is constructed to ensure that the context information can provide sufficient relevance and basis for accurate prediction of the parameters.

[0082] Parameter supplementation: input the constructed context information into the trained context prediction model to predict the value of the missing parameters;

[0083] Supplementary value selection: Based on the results predicted by the model, the most appropriate supplementary value is selected as the estimate of the missing parameter.

[0084] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article 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, article or device.

[0085] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

[0086] The present invention and its embodiments are described above, and such description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. In short, if ordinary technicians in the field are inspired by it, without departing from the purpose of the invention, they can design a structure and embodiment similar to the technical solution without creativity, which should belong to the protection scope of the present invention.

Claims

1. Precision data acquisition system based on multi-parameter sensors, characterized by: It includes a sensor module, a data acquisition module, a data processing module, a data storage module, a data analysis module, a user interface module and a central control module; The sensor module uses a sensor to collect multi-parameter data and transmits the multi-parameter data to a data acquisition module; The data acquisition module collects and caches the multi-parameter data and transmits the multi-parameter data to the data processing module; The data processing module cleans, calibrates and preprocesses the multi-parameter data to obtain processed data, and transmits the processed data to the data storage module and the data analysis module; The data storage module stores and manages the processed data; The data analysis module analyzes and studies the processed data and extracts data features; The user interface module provides a user interaction interface, where users can perform operations and access data; The central control module manages and controls the operation and configuration of the entire data acquisition system; Among them, in the data processing module, the multi-parameter data is cleaned, calibrated and preprocessed to obtain processed data, which specifically includes the following steps: Step S1: Data classification, classifying the multi-parameter data to obtain low-precision multi-parameters and high-precision multi-parameters, converting the low-precision multi-parameters using a low-precision ADC module to obtain low-precision digital data, and converting the high-precision multi-parameters using a high-precision ADC module to obtain high-precision digital data; Step S2: Intelligent data cleaning, using an intelligent perception method based on continuous multi-parameter data to clean the digital data to obtain cleaned data, specifically including the following steps: Step S21: fragment model processing, using the fragment model to predict the low-precision digital data to obtain a prediction score, specifically including the following steps: Step S211: forward sampling, randomly adopting a different sampling frequencies to perform forward sampling on the low-precision digital data, and flattening the a kinds of forward sampling results into one-dimensional vectors and then superimposing them to obtain a forward vector; Step S212: negative sampling, randomly using a different sampling frequencies to perform negative sampling on the low-precision digital data, and flattening the a positive sampling results into a one-dimensional vector and then superimposing them to obtain a negative vector; Step S213: Data encoding, using an encoding function to map the positive vector and the negative vector to a high-dimensional space to obtain a positive supervector and a negative supervector. The formula used is as follows: ； In the formula, The vector representation of the positive vector and the negative vector is ={F1, F2, ..., Fn}, {F1, F2, ..., Fn} represents each parameter in the multi-parameter, is an n×D matrix, where All elements in are sampled from an iid Gaussian distribution. is an element sampled from an iid uniform distribution, These are the calculated positive and negative supervectors; Step S214: encoding training, performing multiple data encoding operations on the positive supervector and the negative supervector to obtain the final positive supervector and the negative supervector, normalizing them, and using an activation function to perform prediction to obtain a prediction score; Step S22: Perception model processing, using an intelligent super perception model to clean low-precision digital data to obtain super-sensing high-precision data, specifically including the following steps: Step S221: setting hyperparameters, setting T score, T detection value and initial stride, where the initial stride is 0; Step S222: Data clipping, performing sliding window clipping on the low-precision digital data to obtain m data segments; Step S223: fragment processing, repeating step S21, performing fragment model processing on m data fragments, and obtaining m prediction scores; Step S224: score judgment, compare the m predicted scores with the T score, if the predicted score is greater than the T score, then the initial stride is increased by 1, and the T score is tested on the m predicted scores in turn to obtain the final stride number; Step S225: Data cleaning, comparing the final stride number with the T detection value. If the final stride number is greater than the T detection value, it means that the low-precision digital data contains interference information, and the low-precision digital data is removed. If the final stride number is less than the T detection value, it means that the low-precision digital data is actually high-precision data, and the low-precision digital data is converted into super-sensing high-precision data, that is, cleaned data; Step S3: parameter optimization, optimizing the parameters in the intelligent perception method based on continuous multi-parameter data to obtain the final cleaned data; Step S4: Data classification processing, defining the final cleaned data and high-precision digital data as high-precision data, and using the method of automatically generating a context model to complete the high-precision data classification, i.e., fill in the frame, to obtain processed data, specifically including the following steps: Step S41: data classification, dividing the processed data according to whether it is IoT data, and obtaining IoT data and non-IoT data; Step S42: IOT data processing, repeating step S213, performing data encoding on the IOT data to obtain a hyper-dimensional IOT vector; Step S43: non-IOT data processing, using a cluster analysis method to divide the non-IOT data into different data tuples; Step S44: context model processing, using the context model to perform information completion operations on the hyperdimensional IOT vector and all data tuples respectively to obtain processed data.

Images