CN110705700A - Drift prediction method of soil temperature sensor - Google Patents

Drift prediction method of soil temperature sensor Download PDF

Info

Publication number
CN110705700A
CN110705700A CN201910998346.8A CN201910998346A CN110705700A CN 110705700 A CN110705700 A CN 110705700A CN 201910998346 A CN201910998346 A CN 201910998346A CN 110705700 A CN110705700 A CN 110705700A
Authority
CN
China
Prior art keywords
soil temperature
neural network
temperature sensor
data
layer
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN201910998346.8A
Other languages
Chinese (zh)
Inventor
杨波
余博文
聂玲
龚元昊
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Chongqing University of Science and Technology
Original Assignee
Chongqing University of Science and Technology
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Chongqing University of Science and Technology filed Critical Chongqing University of Science and Technology
Priority to CN201910998346.8A priority Critical patent/CN110705700A/en
Publication of CN110705700A publication Critical patent/CN110705700A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/044—Recurrent networks, e.g. Hopfield networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/047—Probabilistic or stochastic networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Computing Systems (AREA)
  • Software Systems (AREA)
  • Molecular Biology (AREA)
  • Computational Linguistics (AREA)
  • Biophysics (AREA)
  • Biomedical Technology (AREA)
  • Mathematical Physics (AREA)
  • General Health & Medical Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Probability & Statistics with Applications (AREA)
  • Testing Or Calibration Of Command Recording Devices (AREA)

Abstract

The invention discloses a drift prediction method of a soil temperature sensor. The soil temperature has strong randomness and is influenced by various factors, so that the soil temperature sensor can generate drift, and the soil temperature prediction accuracy is high. According to the phenomenon, the prediction of future soil temperature values by using a recurrent neural network-long short-term memory (RecurrentNeralNet-Long short-term-TermMemory) model is provided aiming at the characteristics of sensor drift and the requirements on the accuracy and the real-time performance of soil temperature prediction. The method uses the RNN-LSTM deep neural network to predict the soil temperature basically consistent with the actual soil temperature, shows that the prediction method of the RNN-LSTM model has the characteristics of very high prediction precision and easy realization, and has ideal prediction effect.

Description

Drift prediction method of soil temperature sensor
Technical Field
The invention relates to the technical field of sensor temperature drift prediction, in particular to a drift prediction method of a soil temperature sensor.
Background
At the present stage, with the great trend of artificial intelligence development in China, the soil temperature sensors in industries such as industry and agriculture influence the efficiency and benefit of enterprise development in the aspects of energy conservation, environmental protection, time saving and labor saving. In order to solve the problems of remote and severe environments and the convenience, high efficiency and economy of technical means, enterprises try to establish various new deep learning models and find the change rule of data and the internal quantity relationship, and the temperature drift problem of the soil temperature sensor is as follows:
(1) as the self parts of the sensor are used for a long time, the aging of the self components changes the performance parameters, the relation between input and output causes larger errors, the final result is influenced, the efficiency is reduced, the precision is improved, and the real-time problem is solved. There is also a certain external risk under certain extreme environmental conditions.
(2) In some particularly harsh environments, sensor maintenance or replacement is uneconomical and time consuming. Has direct connection to enterprise production.
Disclosure of Invention
In order to solve the problems, the invention provides a drift prediction method of a soil temperature sensor with small error, high efficiency and high precision, which comprises the following steps
Acquiring historical data of a soil temperature sensor;
carrying out normalization processing on the historical data;
dividing historical data into a training data set and a testing data set according to a proportion;
constructing an RNN-LSTM neural network by using a python platform;
training an RNN-LSTM neural network using the test data set;
and predicting the temperature drift of the soil temperature sensor by using the trained neural network.
Further, in the above-mentioned case,
the history data is processedThe normalization step uses the following formula for normalization, xk＝(x-xmean)/xvarWherein x ismeanIs the mean of the data sequence, xvarIs the variance of the data, xkIs normalized data.
Further, in the above-mentioned case,
the RNN-LSTM neural network includes,
input layer x ═ x1,···,xt-1,xtSectional, h) and a hidden layer h (h)1,···,ht-1,htV, output layer o ═ o (o)1,···,ot-1,ot,···)；
Wherein o ist＝g(V*ht)，
ht＝f(U*xt+W*ht-1)，
The input value of time t is xtT represents the time parameter of the time sequence, the second layer is a hidden layer, and the hidden layer state at the time point t is htWhere f is a nonlinear activation function, the last layer is the output layer, the output layer o at time tt；
The forgetting gate model is expressed by the following formula,
ft＝σ(Wf·[ht-1,xt]+bf)，
wherein Wf、bfRespectively representing the weight and bias of a forgetting gate;
the input gate and candidate gate models are represented using the following formulas,
it＝σ(Wi·[ht-1,xt]+bi)，
Figure BDA0002240472890000031
wherein Wi、WCRepresenting the respective weight, bi、bCRepresents the corresponding bias;
the model function of the memory cell is expressed by the following formula,
Figure BDA0002240472890000032
wherein C istIndicating the value of the status cell;
the output gate model is represented using the following formula:
ot＝σ(Wo[ht-1,xt]+bo)，
the final time-series output is expressed using the following equation:
ht＝ot*tanh(Ct)，
wherein Wo、boRepresenting the weight and bias of the output gate, respectively.
Further, in the above-mentioned case,
the LSTM neural network forgets to remember the gate according to reading ht-1And xtOutputs a value of 0 to 1 and feeds back the value to each in-cell state Ct-1The number in (1) indicates "completely retained" when the number is 1, and indicates "completely discarded" when the number is 0.
Further, in the above-mentioned case,
the neural network model uses a softmax activation function, a classification cross entropy loss function and an Adam optimization function to update model parameters, and the expression is as follows:
softmax activation function expression:
Figure BDA0002240472890000033
cross entropy loss function expression:
Figure BDA0002240472890000034
wherein y isiRepresenting the true classification result, aiRepresenting the ith output value of softmax.
Further, in the above-mentioned case,
the input data of the neural network input layer is normalized temperature, and the output data of the neural network is normalized temperature.
The invention has the beneficial effects that:
the technical scheme of the invention has the characteristics of very high prediction precision and easy realization, and the prediction effect is more ideal.
Drawings
FIG. 1 is a flow chart of the present invention.
Detailed Description
The invention discloses a drift prediction method of a soil temperature sensor. The soil temperature has strong randomness and is influenced by various factors, so that the soil temperature sensor can generate drift, and the soil temperature prediction accuracy is high. According to the phenomenon, the prediction of future soil temperature values by using a recurrent neural network-long short-term memory (RecurrentNeralNet-Long short-term-TermMemory) model is provided aiming at the characteristics of sensor drift and the requirements on the accuracy and the real-time performance of soil temperature prediction. The method comprises the steps of taking certain soil temperature historical data as input, preprocessing the historical data, establishing a drift prediction model of a soil temperature sensor, and realizing soil temperature prediction in one step in advance.
The technical scheme adopted by the invention is as follows: a drift prediction method of a soil temperature sensor is provided, the soil temperature sensor prediction method based on an RNN-LSTM neural network is provided, the neural network is established through a python platform, a training data set is input, and a network autonomously learns a drift prediction network model for predicting the temperature sensor.
As shown in fig. 1, the present invention provides a drift prediction method of a soil temperature sensor, the method comprising the steps of:
historical data of the soil temperature sensor is acquired.
And carrying out standard processing on the historical data.
The historical data is divided into a training data set and a testing data set according to a certain proportion.
The RNN-LSTM neural network is constructed using the python platform.
The RNN-LSTM neural network is trained with the data set.
And predicting the temperature drift of the soil temperature sensor by using the trained neural network.
In an embodiment of the present invention, the normalization process of the historical data includes;
the decimal between [0,1] is represented by the following formula;
xk＝(x-xmean)/xvar
xmeanis the mean of the data sequence, xvarIs the variance of the data, xkIs normalized data.
In one embodiment of the present invention, the neural network includes an input layer x ═ x (x)1,···,xt-1,xtSectional, h) and a hidden layer h (h)1,···,ht-1,htV, output layer o ═ o (o)1,···,ot-1,ot,···)。
ot＝g(V*ht)
ht＝f(U*xt+W*ht-1)
Wherein the first layer is the input layer between the input layer and the hidden layer (denoted by U), the hidden layer and the output layer (denoted by V), the hidden layer and the hidden layer (denoted by W), and the input value at time t is xtT represents a time parameter of the time series; the second layer is a hidden layer with a hidden layer state h at time ttWherein f is a non-linear activation function; the last layer is an output layer, an output layer o at time tt。
The concrete relational expression of the forgetting door model is as follows:
ft＝σ(Wf·[ht-1,xt]+bf)
wherein Wf、bfRepresenting the weight and bias of the forgetting gate, respectively.
The input gate and candidate gate models have specific relations:
it＝σ(Wi·[ht-1,xt]+bi)
Figure BDA0002240472890000061
wherein Wi、WCRepresenting the respective weight, bi、bCRepresenting the corresponding bias.
Model function of memory cell:
wherein C istThe value of the status cell is indicated.
The output door model has the following specific relational expression:
ot＝σ(Wo[ht-1,xt]+bo)
final output in time series:
ht＝ot*tanh(Ct)
wherein Wo、boRepresenting the weight and bias of the output gate, respectively.
In one embodiment of the present invention, the LSTM neural network forgetting to remember the gate is based on reading ht-1And xtTo output a value of between 0 and 1 to each of the at-cell states Ct-1The numbers in (1). If the number is 1, it means "completely retained", and if the number is 0, it means "completely discarded".
In an embodiment of the present invention, the selection of the hyper-parameter includes a learning rate, a training batch, and a neuron number.
The learning rate is between 0 and 1, the training batch is 2800, and the number of neurons is 32.
In one embodiment of the invention, the model uses a softmax activation function, a classification cross entropy loss function (logarithmic loss function), and an Adam optimization function to update model parameters, and the expression is as follows:
softmax activation function expression:
Figure BDA0002240472890000071
cross entropy loss function expression:
Figure BDA0002240472890000072
wherein y isiRepresenting the true classification result. a isiRepresenting the ith output value of softmax.
In an embodiment of the present invention, the input data of the input layer of the neural network is a normalized temperature, and the output data of the neural network is a normalized temperature.
Although the present invention has been described in detail by way of the general description and the specific examples, the present invention solves the problem of the influence of the drift of the soil temperature sensor on the measurement, makes the data acquisition more convenient, solves the problem of the influence of the drift of the soil temperature sensor on the data, greatly improves the efficiency of the data acquisition, reduces the cost, and improves the data acquisition precision, but on the basis of the present invention, some modifications or improvements can be made, which will be obvious to those skilled in the art. Therefore, it is intended that the present invention cover such modifications and variations as fall within the true spirit of the invention.

Claims (6)

1. A method for predicting drift of a soil temperature sensor, the method comprising the steps of:
acquiring historical data of a soil temperature sensor;
carrying out normalization processing on the historical data;
dividing historical data into a training data set and a testing data set according to a proportion;
constructing an RNN-LSTM neural network by using a python platform;
training an RNN-LSTM neural network using the test data set;
and predicting the temperature drift of the soil temperature sensor by using the trained neural network.
2. The drift prediction method of a soil temperature sensor according to claim 1, wherein said step of normalizing said historical data uses the following formula for normalization,
xk＝(x-xmean)/xvarwherein x ismeanIs the mean of the data sequence, xvarIs the variance of the data, xkIs normalized data.
3. The method of claim 2, wherein the RNN-LSTM neural network comprises,
input layer x ═ x1,…,xt-1,xt…), hidden layer h ═ h (h)1,…,ht-1,ht…), output layer o ═ o (o)1,…,ot-1,ot…); wherein o ist＝g(V*ht)，
ht＝f(U*xt+W*ht-1)，
The input value of time t is xtT represents the time parameter of the time sequence, the second layer is a hidden layer, and the hidden layer state at the time point t is htWhere f is a nonlinear activation function, the last layer is the output layer, the output layer o at time tt；
The forgetting gate model is expressed by the following formula,
ft＝σ(Wf·[ht-1,xt]+bf)，
wherein Wf、bfRespectively representing the weight and bias of a forgetting gate;
the input gate and candidate gate models are represented using the following formulas,
it＝σ(Wi·[ht-1,xt]+bi)，
wherein Wi、WCRepresenting the respective weight, bi、bCRepresents the corresponding bias;
the model function of the memory cell is expressed by the following formula,
Figure FDA0002240472880000021
wherein C istIndicating the value of the status cell;
the output gate model is represented using the following formula:
ot＝σ(Wo[ht-1,xt]+bo)，
the final time-series output is expressed using the following equation:
ht＝ot*tanh(Ct)，
wherein Wo、boRepresenting the weight and bias of the output gate, respectively.
4. The method of claim 3, wherein the LSTM neural network forgetting to gate is based on reading ht-1And xtOutputs a value of 0 to 1 and feeds back the value to each in-cell state Ct-1The number in (1) indicates "completely retained" when the number is 1, and indicates "completely discarded" when the number is 0.
5. The drift prediction method for the soil temperature sensor as claimed in claim 4, wherein the neural network model uses a softmax activation function, a classification cross entropy loss function, and an Adam optimization function to update model parameters, and the expression is as follows:
softmax activation function expression:
Figure FDA0002240472880000031
cross entropy loss function expression:
Figure FDA0002240472880000032
wherein y isiRepresenting the true classification result, aiRepresenting the ith output value of softmax.
6. The method of claim 5, wherein the input data to the neural network input layer is normalized temperature and the neural network output data is normalized temperature.
CN201910998346.8A 2019-10-21 2019-10-21 Drift prediction method of soil temperature sensor Pending CN110705700A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910998346.8A CN110705700A (en) 2019-10-21 2019-10-21 Drift prediction method of soil temperature sensor

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910998346.8A CN110705700A (en) 2019-10-21 2019-10-21 Drift prediction method of soil temperature sensor

Publications (1)

Publication Number Publication Date
CN110705700A true CN110705700A (en) 2020-01-17

Family

ID=69201835

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910998346.8A Pending CN110705700A (en) 2019-10-21 2019-10-21 Drift prediction method of soil temperature sensor

Country Status (1)

Country Link
CN (1) CN110705700A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111829693A (en) * 2020-07-30 2020-10-27 成都运达科技股份有限公司 Bearing temperature detection method and system based on LSTM model of time and space dimensions

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102238900A (en) * 2008-10-03 2011-11-09 拜尔健康护理有限责任公司 System and method for predicting ambient temperature in a fluid analyte meter
CN103063880A (en) * 2012-12-21 2013-04-24 浙江大学 Method of temperature drift estimating and compensating in scanning probe microscopy
CN108803576A (en) * 2018-07-24 2018-11-13 广东工业大学 A kind of fault early warning method and relevant apparatus of temperature control system
CN109685676A (en) * 2018-10-11 2019-04-26 上海交通大学 Intelligent liquid manure management method and system
CN110210174A (en) * 2019-06-21 2019-09-06 西南交通大学 A kind of EMU axle box bearing temperature predicting method based on multilayer LSTM

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102238900A (en) * 2008-10-03 2011-11-09 拜尔健康护理有限责任公司 System and method for predicting ambient temperature in a fluid analyte meter
CN103063880A (en) * 2012-12-21 2013-04-24 浙江大学 Method of temperature drift estimating and compensating in scanning probe microscopy
CN108803576A (en) * 2018-07-24 2018-11-13 广东工业大学 A kind of fault early warning method and relevant apparatus of temperature control system
CN109685676A (en) * 2018-10-11 2019-04-26 上海交通大学 Intelligent liquid manure management method and system
CN110210174A (en) * 2019-06-21 2019-09-06 西南交通大学 A kind of EMU axle box bearing temperature predicting method based on multilayer LSTM

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
付宇: "航空发动机进气道静电传感器理论与实验研究", 《中国优秀博硕士学位论文全文数据库（博士）工程科技II辑》 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111829693A (en) * 2020-07-30 2020-10-27 成都运达科技股份有限公司 Bearing temperature detection method and system based on LSTM model of time and space dimensions
CN111829693B (en) * 2020-07-30 2021-10-22 成都运达科技股份有限公司 Bearing temperature detection method and system based on LSTM model of time and space dimensions

Similar Documents

Publication Publication Date Title
CN110414788B (en) Electric energy quality prediction method based on similar days and improved LSTM
CN108764539B (en) Upstream and downstream water level prediction method for cascade power station
CN111260136A (en) Building short-term load prediction method based on ARIMA-LSTM combined model
CN110210993B (en) Urban short-term gas load prediction method based on cyclic neural network model
CN111310965A (en) Aircraft track prediction method based on LSTM network
CN111027772A (en) Multi-factor short-term load prediction method based on PCA-DBILSTM
WO2021093362A1 (en) Precision compensation method for discharge caustic alkali concentration measurement device during evaporation process
CN111638707B (en) Intermittent process fault monitoring method based on SOM clustering and MPCA
CN111241755A (en) Power load prediction method
CN114282443B (en) Residual service life prediction method based on MLP-LSTM supervised joint model
CN112182954B (en) LSTM-based fluid simulation data prediction model
CN113537469B (en) Urban water demand prediction method based on LSTM network and Attention mechanism
CN114218872A (en) Method for predicting remaining service life based on DBN-LSTM semi-supervised joint model
CN112330027A (en) Search engine index-based power load prediction method
CN113722997A (en) New well dynamic yield prediction method based on static oil and gas field data
CN113139605A (en) Power load prediction method based on principal component analysis and LSTM neural network
CN115860177A (en) Photovoltaic power generation power prediction method based on combined machine learning model and application thereof
CN114862032B (en) XGBoost-LSTM-based power grid load prediction method and device
CN112508286A (en) Short-term load prediction method based on Kmeans-BilSTM-DMD model
CN114595861A (en) MSTL (modeling, transformation, simulation and maintenance) and LSTM (least Square TM) model-based medium-and-long-term power load prediction method
CN116703644A (en) Attention-RNN-based short-term power load prediction method
CN110705700A (en) Drift prediction method of soil temperature sensor
CN108710966B (en) Photovoltaic power generation power prediction method based on multi-cluster ESN neural network
CN114239796A (en) Power system state estimation method based on extended Kalman filtering
CN117113086A (en) Energy storage unit load prediction method, system, electronic equipment and medium

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
RJ01 Rejection of invention patent application after publication
RJ01 Rejection of invention patent application after publication

Application publication date: 20200117