WO2020140189A1 - 一种基于多传感器的农田数据融合处理方法 - Google Patents

一种基于多传感器的农田数据融合处理方法 Download PDF

Info

Publication number
WO2020140189A1
WO2020140189A1 PCT/CN2019/000263 CN2019000263W WO2020140189A1 WO 2020140189 A1 WO2020140189 A1 WO 2020140189A1 CN 2019000263 W CN2019000263 W CN 2019000263W WO 2020140189 A1 WO2020140189 A1 WO 2020140189A1
Authority
WO
WIPO (PCT)
Prior art keywords
average value
sensor
value
actual measurement
data fusion
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.)
Ceased
Application number
PCT/CN2019/000263
Other languages
English (en)
French (fr)
Inventor
赵景波
刘信潮
邱腾飞
朱敬旭辉
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.)
Qingdao University of Technology
Original Assignee
Qingdao University of 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 Qingdao University of Technology filed Critical Qingdao University of Technology
Publication of WO2020140189A1 publication Critical patent/WO2020140189A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING OR CALCULATING; COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00Pattern recognition

Definitions

  • the invention relates to the technical field of data processing, in particular to a multi-sensor based farmland data fusion processing method.
  • Data fusion processing technology is a new type of multi-disciplinary technology, which mainly deals with a large amount of data in a comprehensive and unified manner.
  • the technology can process a variety of data, has strong flexibility, and can use multiple discipline methods Analyze the data to produce many different results.
  • Multi-sensor data fusion is the process of processing data collected by sensors using data fusion technology.
  • multi-sensor data fusion is gradually maturing.
  • People summarize the working principle of multi-sensor data fusion technology as follows: use multiple sensors to measure the same measured parameter at different times, and then use the mathematical model or algorithm to synthesize the measured parameters under certain rules Processing, and then can be more scientific and intuitive interpretation of the measured parameters, to facilitate the decision-making analysis of the next step.
  • the image of multi-sensor data fusion technology can be compared to the process of the human body's perception of external information.
  • the human body can perceive external information through "sensors" such as hands, eyes, and nose, and then process these information through the brain to obtain an accurate conclusion.
  • the specific idea is that when the sensor node collects farmland environmental data, it will be transmitted to the nearest LoRa gateway, and then the microcontroller in the LoRa gateway will immediately fuse the data. At this time, the merged raw data has not been processed. Therefore, the influence caused by interference factors can be effectively eliminated, and a very accurate and effective fusion result can be obtained.
  • the object of the present invention is to provide a multi-sensor based farmland data fusion processing method, so as to realize the fusion of data collected by multiple sensors, and further improve the accuracy of the data fusion result.
  • the present invention provides a multi-sensor-based farmland data fusion processing method, which includes:
  • Step S1 Obtain actual measurement values collected by multiple sensors
  • Step S2 Select the maximum actual measurement value and the minimum actual measurement value from the actual measurement values collected by multiple sensors;
  • Step S3 Determine a first average value according to the maximum actual measurement value and the minimum actual measurement value
  • Step S4 Compare and classify the plurality of actual measurement values with the first average value, classify each actual measurement value greater than the first average value as a positive set, and reduce it to less than the first average value Each actual measurement value of is classified as a negative set;
  • Step S5 Determine a second average value based on each actual measurement value in the positive set, and determine a third average value based on each actual measurement value in the negative set;
  • Step S6 Determine a fourth average value according to the second average value and the third average value
  • Step S7 perform judgment and comparison according to the first average value and the fourth average value, and output the first average value
  • Step S8 Determine the data fusion value corresponding to the sensors of this type based on the first average value.
  • the determining the data fusion value corresponding to the sensor based on the first average value includes:
  • Step S81 Determine the variance of each sensor based on the first average value
  • Step S82 Determine the optimal weighting factor of each sensor according to the variance of each sensor
  • Step S83 Determine the data fusion value corresponding to the sensor based on the actual measurement value of each sensor and the optimal weighting factor.
  • the method further includes:
  • Step S9 Determine the total mean square error based on the variance of each sensor and the optimal weighting factor.
  • the judging and comparing according to the first average value and the fourth average value and outputting the first average value include:
  • step S4 determine whether the first average value is equal to the fourth average value; if the first average value is equal to the fourth average value, output the first average value; if the first average value is not equal to For the fourth average value, assign the fourth average value to the first average value, and return to "step S4".
  • the total mean square error is determined according to the variance and the weighting factor of each sensor, and the specific formula is:
  • ⁇ 2 is the total mean square error
  • ⁇ i 2 is the variance of the i-th sensor
  • W i is the optimal weighting factor of the i-th sensor
  • n is the total number of sensors.
  • the variance of each sensor is determined based on the first average value, and the specific formula is:
  • Y k is the first average value
  • X i is the actual measurement value collected by the i-th sensor.
  • W i is the optimal weighting factor of the i-th sensor
  • I the variance of the i-th sensor
  • n the total number of sensors.
  • the second average value is determined, and the specific formula is:
  • N i is the weighting factor of the i-th sensor
  • N i 1
  • Y k is the first average value
  • E(H k ) is the second average value
  • X i is the actual measurement value collected by the i-th sensor.
  • the specific formula is:
  • N i is the weighting factor of the i-th sensor
  • N i 1
  • Y k is the first average value
  • E(H -k ) is the third average value
  • X i is the actual measurement value collected by the i-th sensor .
  • the present invention discloses the following technical effects:
  • the invention discloses a multi-sensor based farmland data fusion processing method, which comprises: determining a first average value according to a maximum actual measurement value and a minimum actual measurement value; comparing a plurality of the actual measurement values with the first average value Classify to obtain a positive set and a negative set; determine a second average value based on each actual measurement value in the positive set, and determine a third average value based on each actual measurement value in the negative set; according to the The second average value and the third average value determine a fourth average value; perform judgment and comparison according to the first average value and the fourth average value, and output the first average value; based on the first average value Determine the data fusion value corresponding to this type of sensor to achieve the fusion of the data collected by multiple sensors, and further improve the accuracy of the data fusion result.
  • FIG. 1 is a flowchart of a multi-sensor farmland data fusion processing method according to an embodiment of the present invention
  • FIG. 2 is a schematic diagram of comparison between actual measured values and data fusion values of air temperatures according to an embodiment of the present invention
  • FIG. 3 is a schematic diagram of comparison between actual measured values and data fusion values of air humidity according to an embodiment of the present invention
  • FIG. 4 is a schematic diagram of comparison between actual measured values and data fusion values based on soil temperatures in an embodiment of the present invention
  • FIG. 5 is a schematic diagram of comparison between actual measured values and data fusion values of soil moisture according to an embodiment of the present invention.
  • FIG. 6 is a comparison diagram of actual measured values based on light intensity and data fusion values according to an embodiment of the present invention.
  • the purpose of the present invention is to provide a multi-sensor based farmland data fusion processing method, so as to realize the fusion of data collected by multiple sensors, and further improve the accuracy of the data fusion result.
  • FIG. 1 is a flowchart of a multi-sensor farmland data fusion processing method according to an embodiment of the present invention. As shown in FIG. 1, the present invention discloses a multi-sensor farmland data fusion processing method. The method includes:
  • Step S1 Obtain actual measurement values X i collected by multiple sensors, where i is a positive integer greater than or equal to 1;
  • Step S2 Select the maximum actual measurement value X max and the minimum actual measurement value X min from the actual measurement values X i collected by multiple sensors;
  • Step S3 Determine a first average value Y k according to the maximum actual measurement value X max and the minimum actual measurement value X min , where k is the number of calculations, starting from 1, k belongs to a positive integer;
  • Step S4 Compare and classify the plurality of actual measured values X i with the first average value Y k, and classify each actual measured value greater than the first average value Y k as a positive set H k , Classify each actual measurement value less than the first average value Y k as a negative set H -k ;
  • Step S5 determining a second average value E based on the set of n values of the actual measurement of H k [H k], is determined based on an average of the negative third set of actual measurement values in each of the H -k E[H -k ];
  • Step S6 Determine a fourth average value Y k+1 according to the second average value E[H k ] and the third average value E[H -k ];
  • Step S7 Perform judgment and comparison based on the first average value Y k and the fourth average value Y k+1 , and output the first average value Y k ;
  • Step S8 Determine the data fusion value corresponding to the sensors of this type based on the first average value.
  • Step S3 Determine the first average value Y k according to the maximum actual measurement value X max and the minimum actual measurement value X min , the specific formula is:
  • X max is the maximum actual measurement value
  • X min is the minimum actual measurement value
  • k is the number of calculations, starting from 1, k belongs to a positive integer
  • Step S4 Compare and classify the plurality of actual measured values X i with the first average value Y k, and classify each actual measured value greater than the first average value Y k as a positive set H k , Classify each of the actual measured values less than the first average value Y k into a negative set H -k , the specific formula is:
  • Step S5 determining a second average value E based on the set of n values of the actual measurement of H k [H k], is determined based on an average of the negative third set of actual measurement values in each of the H -k E[H -k ],
  • N i is the weighting factor of the i-th sensor
  • N i 1
  • Y k is the first average value
  • E(H k ) is the second average value
  • X i is the actual measurement value collected by the i-th sensor.
  • N i is the weighting factor of the i-th sensor
  • N i 1
  • Y k is the first average value
  • E(H -k ) is the third average value
  • X i is the actual measurement value collected by the i-th sensor .
  • Step S6 Determine a fourth average value Y k+1 according to the second average value E[H k ] and the third average value E[H -k ], the specific formula is:
  • Step S7 Perform judgment and comparison based on the first average value Y k and the fourth average value Y k+1 , and output the first average value Y k , specifically including:
  • Step S8 Determine the data fusion value corresponding to the sensor based on the first average value, which specifically includes:
  • Step S81 Determine the variance of each sensor based on the first average value.
  • the specific formula is:
  • Y k is the first average value
  • X i is the actual measurement value collected by the i-th sensor.
  • Step S82 Determine the optimal weighting factor W i of each sensor according to the variance of each sensor, the specific formula is:
  • W i is the optimal weighting factor of the i-th sensor
  • I the variance of the i-th sensor
  • n the total number of sensors.
  • Step S83 Determine the data fusion value corresponding to the sensor based on the actual measurement value X i of each sensor and the optimal weighting factor W i .
  • Such data fusion is a value corresponding to the sensor, W i is a weighting factor of the optimum sensor i, n is the total number of sensors, X i is the i-th actual measured values acquired by the sensor.
  • Step S9 According to the variance of each sensor And the optimal weighting factor W i determine the total mean square error, the specific formula is:
  • ⁇ 2 is the total mean square error
  • ⁇ i 2 is the variance of the i-th sensor
  • W i is the optimal weighting factor of the i-th sensor
  • n is the total number of sensors.
  • the present invention made an experimental comparison of the information monitoring data in the farmland.
  • the experiment location was in an agricultural shed in Boshan District, Zibo City.
  • an average sensor node was deployed at every distance in the farmland.
  • Five sensor nodes are evenly deployed in the farmland, and each group of nodes can collect the air temperature, air humidity, soil temperature, soil humidity, and light intensity of the farmland, starting at 9:00 in the morning, every 30 minutes, and collecting until 14 : 00, collect 10 sampling data in total.
  • the difference between the maximum value and the minimum value reaches 10 units, it indicates that the data type of the sensor collection node is disturbed by noise, and the group of data is ignored.
  • the actual measured values collected are shown in Table 1.
  • FIG. 2 is a schematic diagram comparing actual measured values and data fusion values based on air temperatures in an embodiment of the present invention
  • FIG. 3 is a schematic diagram comparing actual measured values and data fusion values based on air humidity in an embodiment of the present invention
  • FIG. 4 is an implementation of the present invention Example based on the comparison of actual measured values of each soil temperature and data fusion values
  • FIG. 5 is a schematic comparison of the actual measured values of each soil moisture based on the data fusion values
  • FIG. 6 is a comparison of light intensity based on the embodiments of the invention
  • the round mark curve represents the actual measured value curve detected by the first sensor
  • the square mark curve represents the actual measured value curve detected by the second sensor
  • the diamond shaped curve represents the actual measurement value detected by the second sensor.
  • Value curve the triangle mark curve represents the actual measured value curve detected by the second sensor
  • the five-pointed star curve represents the actual measured value curve detected by the second sensor
  • the thick curve represents the data fusion curve, as can be seen from Figures 2-6.
  • the invention deploys an average of 5 sensor nodes in the farmland, respectively collects air temperature, air humidity, soil temperature, soil humidity and light intensity, and performs fusion processing on the collected data and compares it with the initial value.
  • Each category of data The fusion value is between the actual test values detected by multiple sensors in this category, and the change is stable. Therefore, the comparison results show that the improved adaptive fusion algorithm can achieve better fusion results and can meet the needs of digital farmland information monitoring.
  • the present invention effectively classifies and iterates the actual measured values collected by each sensor, effectively eliminates invalid values, and obtains a more accurate variance value, and thus more accurately determines the optimal weighting factor, thereby obtaining the optimal fusion result (i.e. Data fusion value).

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Artificial Intelligence (AREA)
  • Testing Or Calibration Of Command Recording Devices (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

一种基于多传感器的农田数据融合处理方法,包括:根据最大实际测量值和最小实际测量值确定第一平均值(S3);将多个实际测量值与所述第一平均值进行比较分类,获得正集合和负集合(S4);基于所述正集合中的各所述实际测量值确定第二平均值,基于所述负集合中的各所述实际测量值确定第三平均值(S5);根据所述第二平均值和所述第三平均值确定第四平均值(S6);根据所述第一平均值和所述第四平均值进行判断比较,输出所述第一平均值(S7);基于所述第一平均值确定该类传感器对应的数据融合值(S8),以实现对多个传感器采集的数据进行融合,进一步提高数据融合结果的准确性。

Description

一种基于多传感器的农田数据融合处理方法
本申请要求于2019年1月2日提交中国专利局、申请号为201910001608.9、发明名称为“一种基于多传感器的农田数据融合处理方法”的中国专利申请的优先权,其全部内容通过引用结合在本申请中。
技术领域
本发明涉及数据处理技术领域,特别是涉及一种基于多传感器的农田数据融合处理方法。
背景技术
数据融合处理技术是一种新型的多学科结合的技术,主要对大量的数据进行综合统一的处理,该技术可以处理各种各样的数据,具有很强的灵活性,可以利用多种学科方法对数据进行分析从而产生多种不同的结果。
多传感器数据融合顾名思义就是利用数据融合技术对传感器采集的信息进行处理的过程,随着科学技术的发展和信息技术的崛起,多传感器数据融合正逐渐走向成熟。人们将多传感器数据融合技术的工作原理总结为:在不同的时间下利用多个传感器对同一个被测参数进行测量,然后将测量到的参数利用数学模型或算法等手段在一定规则下进行综合处理,进而能够对被测参数进行更加科学直观的解释,便于下一步采取的决策分析。可以将多传感器数据融合技术形象的比喻成人体感知外界信息的过程,人体可以通过手、眼、鼻等“传感器”来感知外界信息,然后通过大脑来统一处理这些信息进而得到一个准确的结论。
由于农田面积较大,在同一片农田内需要布置多个传感器节点来均匀的采集环境信息,若要对多亩农田同时进行采集就会产生大量的数据。另外,由于大部分农田所处的地理环境都较为偏僻,其自然环境也较为复杂,因此很容易对传感器的测量精度造成干扰,如果在同时传输大量信息时,产生的能耗也会对传感器的采集精度造成巨大影响。为了达到数字农田信息采集系统对于采集精度的要求,采用了处理信息无损和高性能著称的数据级融合方式对农田的环境数据进行融合。具体的思路是当传感器节点采集到农田环境数据后,会传出给就近的LoRa网关,随后LoRa网关内的微控制器立即对这些数据进行融合,此时融合的这些原始数据还未经过任 何处理,因此可以有效的排除干扰因素造成的影响,得出非常准确有效的融合结果。
发明内容
基于此,本发明的目的是提供一种基于多传感器的农田数据融合处理方法,以实现对多个传感器采集的数据进行融合,进一步提高数据融合结果的准确性。
为实现上述目的,本发明提供了一种基于多传感器的农田数据融合处理方法,所述方法包括:
步骤S1:获取多个传感器采集的实际测量值;
步骤S2:从多个传感器采集的实际测量值中选取最大实际测量值和最小实际测量值;
步骤S3:根据所述最大实际测量值和所述最小实际测量值确定第一平均值;
步骤S4:将多个所述实际测量值与所述第一平均值进行比较分类,将大于所述第一平均值的各所述实际测量值分类为正集合,将小于所述第一平均值的各所述实际测量值分类为负集合;
步骤S5:基于所述正集合中的各所述实际测量值确定第二平均值,基于所述负集合中的各所述实际测量值确定第三平均值;
步骤S6:根据所述第二平均值和所述第三平均值确定第四平均值;
步骤S7:根据所述第一平均值和所述第四平均值进行判断比较,输出所述第一平均值;
步骤S8:基于所述第一平均值确定该类传感器对应的数据融合值。
可选的,所述基于所述第一平均值确定该类传感器对应的数据融合值,包括:
步骤S81:基于所述第一平均值确定各传感器的方差;
步骤S82:根据各传感器的方差确定各传感器的最优加权因子;
步骤S83:基于各传感器的实际测量值和最优加权因子确定该类传感器对应的数据融合值。
可选的,所述所述方法还包括:
步骤S9:根据各传感器的所述方差和所述最优加权因子确定总均方 误差。
可选的,所述根据所述第一平均值和所述第四平均值进行判断比较,输出所述第一平均值,包括:
判断所述第一平均值与所述第四平均值的差值是否小于或等于设定阈值;如果所述差值小于或等于设定阈值,则输出所述第一平均值;如果所述差值大于设定阈值,则将所述第四平均值赋值给所述第一均值,并返回“步骤S4”;
或者判断所述第一平均值是否等于所述第四平均值;如果所述第一平均值等于所述第四平均值,则输出所述第一平均值;如果所述第一平均值不等于所述第四平均值,则将所述第四平均值赋值给所述第一均值,并返回“步骤S4”。
可选的,所述根据各传感器的所述方差和所述加权因子确定总均方误差,具体公式为:
Figure PCTCN2019000263-appb-000001
其中,σ 2为总均方误差,σ i 2为第i个传感器的所述方差,W i为第i个传感器的最优加权因子,n为传感器的总个数。
可选的,所述基于所述第一平均值确定各传感器的方差,具体公式为:
Figure PCTCN2019000263-appb-000002
其中,
Figure PCTCN2019000263-appb-000003
为第i个传感器的方差,Y k为第一平均值,X i为第i个传感器采集的实际测量值。
可选的,确定各传感器的加权因子,具体公式为:
Figure PCTCN2019000263-appb-000004
其中,W i为第i个传感器的最优加权因子,
Figure PCTCN2019000263-appb-000005
为第i个传感器的方差,n为传感器的总个数。
可选的,确定第二平均值,具体公式为:
Figure PCTCN2019000263-appb-000006
其中,N i为第i个传感器的加权因子,N i=1,Y k为第一平均值,E(H k)为第二平均值,X i为第i个传感器采集的实际测量值。
可选的,确定第三平均值,具体公式为:
Figure PCTCN2019000263-appb-000007
其中,N i为第i个传感器的加权因子,N i=1,Y k为第一平均值,E(H -k)为第三平均值,X i为第i个传感器采集的实际测量值。
根据本发明提供的具体实施例,本发明公开了以下技术效果:
本发明公开一种基于多传感器的农田数据融合处理方法,包括:根据最大实际测量值和最小实际测量值确定第一平均值;将多个所述实际测量值与所述第一平均值进行比较分类,获得正集合和负集合;基于所述正集合中的各所述实际测量值确定第二平均值,基于所述负集合中的各所述实际测量值确定第三平均值;根据所述第二平均值和所述第三平均值确定第四平均值;根据所述第一平均值和所述第四平均值进行判断比较,输出所述第一平均值;基于所述第一平均值确定该类传感器对应的数据融合值,以实现对多个传感器采集的数据进行融合,进一步提高数据融合结果的准确性。
说明书附图
为了更清楚地说明本发明实施例或现有技术中的技术方案,下面将对实施例中所需要使用的附图作简单地介绍,显而易见地,下面描述中的附图仅仅是本发明的一些实施例,对于本领域普通技术人员来讲,在不付出创造性劳动性的前提下,还可以根据这些附图获得其他的附图。
图1为本发明实施例基于多传感器的农田数据融合处理方法流程图;
图2为本发明实施例基于各空气温度的实际测量值与数据融合值对比示意图;
图3为本发明实施例基于各空气湿度的实际测量值与数据融合值对比示意图;
图4为本发明实施例基于各土壤温度的实际测量值与数据融合值对比示意图;
图5为本发明实施例基于各土壤湿度的实际测量值与数据融合值对比示意图;
图6为本发明实施例基于各光照强度的实际测量值与数据融合值对 比示意图。
具体实施方式
下面将结合本发明实施例中的附图,对本发明实施例中的技术方案进行清楚、完整地描述,显然,所描述的实施例仅仅是本发明一部分实施例,而不是全部的实施例。基于本发明中的实施例,本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例,都属于本发明保护的范围。
本发明的目的是提供一种基于多传感器的农田数据融合处理方法,以实现对多个传感器采集的数据进行融合,进一步提高数据融合结果的准确性。
为使本发明的上述目的、特征和优点能够更加明显易懂,下面结合附图和具体实施方式对本发明作进一步详细的说明。
图1为本发明实施例基于多传感器的农田数据融合处理方法流程图,如图1所示,本发明公开一种多传感器的农田数据融合处理方法,所述方法包括:
步骤S1:获取多个传感器采集的实际测量值X i,其中i为大于等于1的正整数;
步骤S2:从多个传感器采集的实际测量值X i中选取最大实际测量值X max和最小实际测量值X min
步骤S3:根据所述最大实际测量值X max和所述最小实际测量值X min确定第一平均值Y k,其中,k为进行计算的次数,从1开始,k属于正整数;
步骤S4:将多个所述实际测量值X i与所述第一平均值Y k进行比较分类,将大于所述第一平均值Y k的各所述实际测量值分类为正集合H k,将小于所述第一平均值Y k的各所述实际测量值分类为负集合H -k
步骤S5:基于所述正集合H k中的各所述实际测量值确定第二平均值E[H k],基于所述负集合H -k中的各所述实际测量值确定第三平均值E[H -k];
步骤S6:根据所述第二平均值E[H k]和所述第三平均值E[H -k]确定第四平均值Y k+1
步骤S7:根据所述第一平均值Y k和所述第四平均值Y k+1进行判断比 较,输出所述第一平均值Y k
步骤S8:基于所述第一平均值确定该类传感器对应的数据融合值。
下面对各个步骤进行详细论述:
步骤S3:根据所述最大实际测量值X max和所述最小实际测量值X min确定第一平均值Y k,具体公式为:
Figure PCTCN2019000263-appb-000008
其中,X max为最大实际测量值,X min为最小实际测量值,k为进行计算的次数,从1开始,k属于正整数;
步骤S4:将多个所述实际测量值X i与所述第一平均值Y k进行比较分类,将大于所述第一平均值Y k的各所述实际测量值分类为正集合H k,将小于所述第一平均值Y k的各所述实际测量值分类为负集合H -k,具体公式为:
Figure PCTCN2019000263-appb-000009
步骤S5:基于所述正集合H k中的各所述实际测量值确定第二平均值E[H k],基于所述负集合H -k中的各所述实际测量值确定第三平均值E[H -k],
确定第二平均值的具体公式为:
Figure PCTCN2019000263-appb-000010
其中,N i为第i个传感器的加权因子,N i=1,Y k为第一平均值,E(H k)为第二平均值,X i为第i个传感器采集的实际测量值。
确定第三平均值的具体公式为:
Figure PCTCN2019000263-appb-000011
其中,N i为第i个传感器的加权因子,N i=1,Y k为第一平均值,E(H -k)为第三平均值,X i为第i个传感器采集的实际测量值。
步骤S6:根据所述第二平均值E[H k]和所述第三平均值E[H -k]确定第四平均值Y k+1,具体公式为:
Figure PCTCN2019000263-appb-000012
步骤S7:根据所述第一平均值Y k和所述第四平均值Y k+1进行判断比较,输出所述第一平均值Y k,具体包括:
步骤S71:判断所述第一平均值Y k与所述第四平均值Y k+1的差值是否小于或等于设定阈值;如果所述差值小于或等于设定阈值,则输出所述第一平均值Y k;如果所述差值大于设定阈值,则令k=k+1(即将所述第四平均值赋值给所述第一均值),并返回“步骤S4”;
或者步骤S72:判断所述第一平均值Y k是否等于所述第四平均值Y k+1;如果所述第一平均值Y k等于所述第四平均值Y k+1,则输出所述第一平均值Y k;如果所述第一平均值Y k不等于所述第四平均值Y k+1,则令k=k+1(即将所述第四平均值赋值给所述第一均值),并返回“步骤S4”。
步骤S8:基于所述第一平均值确定该类传感器对应的数据融合值,具体包括:
步骤S81:基于所述第一平均值确定各传感器的方差,具体公式为:
Figure PCTCN2019000263-appb-000013
其中,
Figure PCTCN2019000263-appb-000014
为第i个传感器的方差,Y k为第一平均值,X i为第i个传感器采集的实际测量值。
步骤S82:根据各传感器的方差确定各传感器的最优加权因子W i,具体公式为:
Figure PCTCN2019000263-appb-000015
其中,W i为第i个传感器的最优加权因子,
Figure PCTCN2019000263-appb-000016
为第i个传感器的方差,n为传感器的总个数。
步骤S83:基于各传感器的实际测量值X i和最优加权因子W i确定该类传感器对应的数据融合值。
Figure PCTCN2019000263-appb-000017
其中,
Figure PCTCN2019000263-appb-000018
为该类传感器对应的数据融合值,W i为第i个传感器的最优加权因子,n为传感器的总个数,X i为第i个传感器采集的实际测量值。
步骤S9:根据各传感器的所述方差
Figure PCTCN2019000263-appb-000019
和所述最优加权因子W i确定总均方误差,具体公式为:
Figure PCTCN2019000263-appb-000020
其中,σ 2为总均方误差,σ i 2为第i个传感器的所述方差,W i为第i个传感器的最优加权因子,n为传感器的总个数。
为了检验该方法的可行性,本发明对农田内的信息监测数据进行了实验对比,实验地点为淄博市博山区某农业大棚内,实验时在农田内每隔一段距离平均部署一个传感器节点,在农田中均匀部署5个传感器节点,每组节点均能够采集农田的空气温度、空气湿度、土壤温度、土壤湿度以及光照强度,从早晨9:00开始,每隔30分钟采集一次,一直采集到14:00,共采集10次采样数据。在每个采集节点采集某一数据的过程中,当最大值和最小值的差值达到10个单位时,表明传感器采集节点的本数据类型受到了噪声的干扰,则忽略该组数据,各传感器采集的实际测量值如表1所示。
表1 传感器采集的实际测量值
Figure PCTCN2019000263-appb-000021
Figure PCTCN2019000263-appb-000022
图2为本发明实施例基于各空气温度的实际测量值与数据融合值对比示意图;图3为本发明实施例基于各空气湿度的实际测量值与数据融合值对比示意图;图4为本发明实施例基于各土壤温度的实际测量值与数据融合值对比示意图;图5为本发明实施例基于各土壤湿度的实际测量值与数据融合值对比示意图;图6为本发明实施例基于各光照强度的实际测量值与数据融合值对比示意图,圆形标志曲线代表第一传感器检测的实际测量值曲线,正方形标志曲线代表第二传感器检测的实际测量值曲线,菱形标志曲线代表第二传感器检测的实际测量值曲线,三角标志曲线代表第二传感器检测的实际测量值曲线,五角星标志曲线代表第二传感器检测的实际测量值曲线,粗曲线代表数据融合的曲线,由图2-图6所可知,本发明在农田中平均部署了5个传感器节点,分别采集空气温度、空气湿度、土壤温度、土壤湿度和光照强度,对采集到的各数据分别进行融合处理后与初始值对比,每一类别的数据融合值位于该类别多个传感器检测的实际测试值之间,且变化稳定,因此表明对比结果表明该改进的自适应融合算法能够达到较好的融合效果,能够满足数字化农田信息监测的需求。
本发明通过对各传感器采集到的实际测量值进行分类迭代,有效的排除了无效值,得到更加准确的方差值,进而更加准确的确定最优加权因子,从而得到最优的融合结果(即数据融合值)。
本说明书中各个实施例采用递进的方式描述,每个实施例重点说明的都是与其他实施例的不同之处,各个实施例之间相同相似部分互相参见即可。
本文中应用了具体个例对本发明的原理及实施方式进行了阐述,以上实施例的说明只是用于帮助理解本发明的方法及其核心思想;同时,对于本领域的一般技术人员,依据本发明的思想,在具体实施方式及应用范围 上均会有改变之处。综上所述,本说明书内容不应理解为对本发明的限制。

Claims (9)

  1. 一种基于多传感器的农田数据融合处理方法,其特征在于,所述方法包括:
    步骤S1:获取多个传感器采集的实际测量值;
    步骤S2:从多个传感器采集的实际测量值中选取最大实际测量值和最小实际测量值;
    步骤S3:根据所述最大实际测量值和所述最小实际测量值确定第一平均值;
    步骤S4:将多个所述实际测量值与所述第一平均值进行比较分类,将大于所述第一平均值的各所述实际测量值分类为正集合,将小于所述第一平均值的各所述实际测量值分类为负集合;
    步骤S5:基于所述正集合中的各所述实际测量值确定第二平均值,基于所述负集合中的各所述实际测量值确定第三平均值;
    步骤S6:根据所述第二平均值和所述第三平均值确定第四平均值;
    步骤S7:根据所述第一平均值和所述第四平均值进行判断比较,输出所述第一平均值;
    步骤S8:基于所述第一平均值确定该类传感器对应的数据融合值。
  2. 根据权利要求1所述的基于多传感器的农田数据融合处理方法,其特征在于,所述基于所述第一平均值确定该类传感器对应的数据融合值,包括:
    步骤S81:基于所述第一平均值确定各传感器的方差;
    步骤S82:根据各传感器的方差确定各传感器的最优加权因子;
    步骤S83:基于各传感器的实际测量值和最优加权因子确定该类传感器对应的数据融合值。
  3. 根据权利要求2所述的基于多传感器的农田数据融合处理方法,其特征在于,所述所述方法还包括:
    步骤S9:根据各传感器的所述方差和所述最优加权因子确定总均方误差。
  4. 根据权利要求1所述的基于多传感器的农田数据融合处理方法,其特征在于,所述根据所述第一平均值和所述第四平均值进行判断比较,输出所述第一平均值,包括:
    判断所述第一平均值与所述第四平均值的差值是否小于或等于设定阈值;如果所述差值小于或等于设定阈值,则输出所述第一平均值;如果所述差值大于设定阈值,则将所述第四平均值赋值给所述第一均值,并返回“步骤S4”;
    或者判断所述第一平均值是否等于所述第四平均值;如果所述第一平均值等于所述第四平均值,则输出所述第一平均值;如果所述第一平均值不等于所述第四平均值,则将所述第四平均值赋值给所述第一均值,并返回“步骤S4”。
  5. 根据权利要求3所述的基于多传感器的农田数据融合处理方法,其特征在于,所述根据各传感器的所述方差和所述加权因子确定总均方误差,具体公式为:
    Figure PCTCN2019000263-appb-100001
    其中,σ 2为总均方误差,σ i 2为第i个传感器的所述方差,W i为第i个传感器的最优加权因子,n为传感器的总个数。
  6. 根据权利要求2所述的基于多传感器的农田数据融合处理方法,其特征在于,所述基于所述第一平均值确定各传感器的方差,具体公式为:
    Figure PCTCN2019000263-appb-100002
    其中,
    Figure PCTCN2019000263-appb-100003
    为第i个传感器的方差,Y k为第一平均值,X i为第i个传感器采集的实际测量值。
  7. 根据权利要求2所述的基于多传感器的农田数据融合处理方法,其特征在于,确定各传感器的加权因子,具体公式为:
    Figure PCTCN2019000263-appb-100004
    其中,W i为第i个传感器的最优加权因子,
    Figure PCTCN2019000263-appb-100005
    为第i个传感器的方差,n为传感器的总个数。
  8. 根据权利要求1所述的基于多传感器的农田数据融合处理方法,其特征在于,确定第二平均值,具体公式为:
    Figure PCTCN2019000263-appb-100006
    其中,N i为第i个传感器的加权因子,N i=1,Y k为第一平均值,E(H k) 为第二平均值,X i为第i个传感器采集的实际测量值。
  9. 根据权利要求1所述的基于多传感器的农田数据融合处理方法,其特征在于,确定第三平均值,具体公式为:
    Figure PCTCN2019000263-appb-100007
    其中,N i为第i个传感器的加权因子,N i=1,Y k为第一平均值,E(H -k)为第三平均值,X i为第i个传感器采集的实际测量值。
PCT/CN2019/000263 2019-01-02 2019-12-30 一种基于多传感器的农田数据融合处理方法 Ceased WO2020140189A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910001608.9 2019-01-02
CN201910001608.9A CN109766941B (zh) 2019-01-02 2019-01-02 一种基于多传感器的农田数据融合处理方法

Publications (1)

Publication Number Publication Date
WO2020140189A1 true WO2020140189A1 (zh) 2020-07-09

Family

ID=66453569

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/000263 Ceased WO2020140189A1 (zh) 2019-01-02 2019-12-30 一种基于多传感器的农田数据融合处理方法

Country Status (2)

Country Link
CN (1) CN109766941B (zh)
WO (1) WO2020140189A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113873459A (zh) * 2021-07-16 2021-12-31 合肥工业大学 土壤成分采集的多传感器优化部署方法

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109766941B (zh) * 2019-01-02 2021-09-21 青岛理工大学 一种基于多传感器的农田数据融合处理方法
CN110298409A (zh) * 2019-07-03 2019-10-01 广东电网有限责任公司 面向电力可穿戴设备的多源数据融合方法
CN111766370A (zh) * 2020-07-08 2020-10-13 王善举 一种土壤全氮含量实时检测系统
CN112268719A (zh) * 2020-09-29 2021-01-26 河南科技大学 一种联合收割机割台远程故障诊断方法

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20060082490A1 (en) * 2002-03-26 2006-04-20 Hai-Wen Chen Method and system for data fusion using spatial and temporal diversity between sensors
CN103822662A (zh) * 2014-02-10 2014-05-28 北京林业大学 农林用温湿度照度智能传感器
CN109766941A (zh) * 2019-01-02 2019-05-17 青岛理工大学 一种基于多传感器的农田数据融合处理方法

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20060082490A1 (en) * 2002-03-26 2006-04-20 Hai-Wen Chen Method and system for data fusion using spatial and temporal diversity between sensors
CN103822662A (zh) * 2014-02-10 2014-05-28 北京林业大学 农林用温湿度照度智能传感器
CN109766941A (zh) * 2019-01-02 2019-05-17 青岛理工大学 一种基于多传感器的农田数据融合处理方法

Non-Patent Citations (5)

* Cited by examiner, † Cited by third party
Title
MALLI WANG ET AL: "Research on digital farmland information monitoring system based on Internet of Things technology", CHINA MASTER’S THESES FULL-TEXT DATABASE, AGRICULTURE, vol. 40, no. 9, 15 May 2019 (2019-05-15), pages 1 - 7, XP055716198, DOI: 10.13733/j.jcam.issn.2095-5553..2019.09.28 *
SUN, YUWEN: "Research and implementation of field environment monitoring system based on wireless sensor networks", CHINA DOCTORAL DISSERTATIONS FULL-TEXT DATABASE, 31 December 2013 (2013-12-31), XP055716201 *
TANG, YAPENG: "Data processing based on adaptive weighted data fusion algorithm", COMPUTER TECHNOLOGY AND DEVELOPMENT, vol. 25, no. 4, 30 April 2015 (2015-04-30), pages 53 - 56, XP009521787, ISSN: 1673-629X *
WANG, HAO ET AL: "Soil moisture monitoring based on improved adaptive weighted fusion algorithm", JOURNAL OF CHINESE AGRICULTURAL MECHANIZATION, vol. 40, no. 1, 15 January 2019 (2019-01-15), pages 152 - 155, XP009521786, ISSN: 2095-5553 *
ZHANG, ZHIWEI ET AL: "Hierarchical data aggregation algorithm of WSNs based on the adaptive weighted and LZW", CHINESE JOURNAL OF SENSORS AND ACTUATORS, vol. 24, no. 08, 31 August 2011 (2011-08-31), pages 1193 - 1196, XP009521788, ISSN: 1004-1699 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113873459A (zh) * 2021-07-16 2021-12-31 合肥工业大学 土壤成分采集的多传感器优化部署方法
CN113873459B (zh) * 2021-07-16 2024-03-12 合肥工业大学 土壤成分采集的多传感器优化部署方法

Also Published As

Publication number Publication date
CN109766941A (zh) 2019-05-17
CN109766941B (zh) 2021-09-21

Similar Documents

Publication Publication Date Title
WO2020140189A1 (zh) 一种基于多传感器的农田数据融合处理方法
CN115342814B (zh) 一种基于多传感器数据融合的无人船定位方法
CN103888541B (zh) 一种融合拓扑势和谱聚类的社区发现方法及系统
WO2019233189A1 (zh) 一种传感网络异常数据检测方法
CN108090515B (zh) 一种基于数据融合的环境等级评估方法
CN107704966A (zh) 一种基于天气大数据的能源负荷预测系统及方法
CN106650785A (zh) 基于证据分类和冲突衡量的加权证据融合方法
CN106529596B (zh) 一种基于可穿戴设备的室内、室外场景识别方法
CN114861706B (zh) 一种基于质量评估和深度迁移学习的心电身份识别方法
CN114487129A (zh) 基于声发射技术的柔性材料的损伤识别方法
CN106644035B (zh) 一种基于时频变换特性的振动源识别方法及系统
CN110567721B (zh) 一种滚动轴承故障诊断方法及系统
CN112347894A (zh) 基于迁移学习和高斯混合模型分离的单株植被提取方法
CN105698861A (zh) 一种室内环境舒适度评价方法
CN107230197A (zh) 基于卫星云图和rvm的热带气旋客观定强方法
CN102163301A (zh) 基于bp人工神经网络的农作物害虫发生量的预测方法
CN107808375A (zh) 融合多种上下文深度学习模型的水稻病害图像检测方法
CN120804767A (zh) 一种配网线路故障检测数据处理方法、系统、设备及介质
CN116561692B (zh) 一种动态更新的实时量测数据检测方法
CN114862840A (zh) 一种融合专家先验的烤烟部位区分方法
CN116776245B (zh) 一种基于机器学习的三相逆变器设备故障诊断方法
CN115619605B (zh) 一种基于语义信息的传统民居环境质量评估方法
CN105407496B (zh) 一种识别无线传感器网络中错误测量值的方法
CN110604572A (zh) 基于人脑特征图谱的脑活动状态识别方法
CN113947796A (zh) 一种基于身份识别的人体温度趋势检测方法及装置

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19907732

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 22/11/2021)

122 Ep: pct application non-entry in european phase

Ref document number: 19907732

Country of ref document: EP

Kind code of ref document: A1