CN102914623A - Fusing method of temperature compensation of humidity sensor - Google Patents

Abstract

The invention relates to a fusing method of temperature compensation of a humidity sensor. A characteristic function y=f(x) of the humidity sensor is given, wherein y represents the output of the humidity sensor, x represents the input of the humidity sensor, a temperature parameter is provided additionally, when t0<t1<t2<t3, according to the characteristics that a characteristic curve of the humidity sensor is linear in the middle and is nonlinear at two ends under the influence of the temperature, through a linear least square method, a linear equation f2(x, t) influenced by the temperature within a range of [t1, t2] is obtained; and when t0<t<t1 or t2<t<t3, error compensation caused by the temperature influence is carried out on f1(x, t) and f3 (x, t) through an RBF (Radial Basis Function) neutral network. The fusing method has the beneficial effect that an actual value which is closer to an output value of the humidity sensor is finally obtained.

Description

A kind of fusion method of humidity sensor temperature compensation

Technical field

The invention belongs to the soft compensation technique of sensor field, be specifically related to the application of method for self-adaption amalgamation in the humidity sensor temperature compensation that combine with least square of a kind of RBF neural network.

Background technology

Air humidity is an important parameter in the meteorological observation, and meteorological department has dropped into the automatic weather station of service operation at present, and the warm and humid integrated transducer of the HMP45D that mostly adopts Finland Vaisala company to produce is measured air humidity.The HMP45D Temperature Humidity Sensor is the thin polymer film capacitive transducer of new generation with HUMICAP technology, in the process of using, calculate the relative humidity size by linear relationship, yet in the practical business process, the measurement of humidity under high/low temperature condition occurs, can not satisfy the problem of measuring accuracy and reliability.Therefore, the non-linear effects that compensation temperature produces becomes the problem that sensor research staff and worker in meteorology are concerned about the most.The non-linear effects that humidity sensor is produced in order to eliminate temperature, the researcher is mainly round hardware compensating and two kinds of methods of software compensation both at home and abroad.The hardware compensating method causes whole measuring system poor reliability and precision low owing to be subject to the impact of the factor such as electronic devices and components drift in the circuit, can't accomplish omnidistance compensation, uses in the actual use procedure to be restricted.Software compensation method has the multivariate regression analytic approach, yet when application multivariate regression analytic approach is carried out Data Fusion, need to find the solution large-scale matrix equation, even equation is ill-condition equation sometimes, can't find the solution, and fitting precision is lower, often can not satisfies application request; Utilize the non-linear mapping capability of BP neural network also can realize the temperature compensation of humidity sensor, but there are some problems in the method in application process, be absorbed in easily local extremum, make failure to train, and when data volume was larger, pace of learning was slower, and fitting precision is lower.

Summary of the invention

The object of the invention is to overcome the deficiency of above prior art, and the fusion method of the less humidity sensor temperature compensation of a kind of error is provided, and specifically has following technical scheme to realize:

The fusion method of described humidity sensor temperature compensation, the characteristic function of establishing humidity sensor are y=f (x), x ∈ [δ ₀, δ _h], wherein y is the output of sensor, and x is the input of sensor, and input reference signal is δ _h-δ ₀, it is characterized in that set up temperature parameters t, linear in the middle of the family curve under the temperature effect presents according to humidity sensor, the nonlinear characteristics in two ends make t0＜t1＜t2＜t3, obtain formula (1),

$y=f(x,t)=\left\{\begin{array}{c}{f}_1(x,t),t\&Element;[{\mathrm{<figure-callout\; id="0"\; label="t"\; filenames="121019155826.png"\; state="\{\{state\}\}">t</figure-callout>}}_0,t_1]\\ f_2(x,t),t\&Element;[{\mathrm{<figure-callout\; id="1"\; label="t"\; filenames="121019155826.png"\; state="\{\{state\}\}">t</figure-callout>}}_1,t_2]\\ f_3(x,t),t\&Element;[{\mathrm{<figure-callout\; id="2"\; label="t"\; filenames="121019155826.png"\; state="\{\{state\}\}">t</figure-callout>}}_2,t_3]\end{array}\right.---\left(1\right)$

Wherein, the characteristic function curve of humidity sensor is to be the function of continuous monotone variation in the input signal range ability; As t1＜t＜t2, obtain f by linear least square ₂(x, t) is at the linear equation of [t1, t2] interior temperature influence; As t0＜t＜t1 or t2＜t＜t3, by the RBF neural network to f ₁(x, t) and f ₃(x, t) carries out the error compensation brought by temperature effect.

The further design of the fusion method of described humidity sensor temperature compensation is, as t1＜t＜t2, add that in the described linear equation of finally trying to achieve parametric coefficients ξ compensation temperature produces minor impact, directly try to achieve corresponding input signal x with straight-line equation according to the output after the A/D conversion, simplify the processing mode of output signal.

The further design of the fusion method of described humidity sensor temperature compensation is that described error compensation of being undertaken by the RBF neural network comprises the steps:

1) netinit is chosen n training sample at random as cluster centre c _i(i=1,2 ..., n);

2) grouping of arest neighbors rule is pressed in the training sample set of input, then readjust cluster centre, the c that obtains at last _iCenter for network function;

3) find the solution the variance of network function when being Gaussian function

I=1,2 ..., n, wherein c _MaxBe the ultimate range between the selected center;

4) weights between calculating hidden layer and the output layer

(p=1,2 ..., P; I=1,2 ..., n);

5). by formula

Then calculation of performance indicators J carries out the weights adjustment, wherein, and y _kDesirable output,

Be reality output, k=1,2 ... L, L, M are empirical values;

6). judge the performance index J≤ζ that whether satisfies condition, ζ is empirical value, if satisfy, finishes training, otherwise makes J=0 return step 2, repeats above-mentioned training process, until satisfy condition.

The present invention is according to the family curve of humidity sensor under the impact of temperature, one section presents linearity, two ends present nonlinear characteristics, for the less linearity range of temperature effect, utilize the method for least square to simulate straight-line equation, add corresponding parametric coefficients, the impact of compensation temperature, make the input input be good linear relationship, for the larger characteristic two ends of temperature effect, utilize the RBF neural network model to realize the temperature compensation of humidity sensor, so that input and output also are good linear relationship, finally obtain the output valve of humidity sensor more near actual value.

Description of drawings

Fig. 1 is the humidity sensor temperature compensation principle;

Fig. 2 is that temperature is on humidity sensor measurement result's impact;

Fig. 3 is linear least square fitting a straight line synoptic diagram;

Fig. 4 is the humidity sensor input-output curve after the temperature compensation.

Embodiment

Below in conjunction with accompanying drawing the present invention program is elaborated.

The fusion method of described humidity sensor temperature compensation, the characteristic function of establishing humidity sensor are y=f (x), x ∈ [δ ₀, δ _h], wherein y is the output of sensor, and x is the input of sensor, and input reference signal is δ _h-δ ₀, it is characterized in that set up temperature parameters t, linear in the middle of the family curve under the temperature effect presents according to humidity sensor, the nonlinear characteristics in two ends make t0＜t1＜t2＜t3, obtain formula (1),

$y=f(x,t)=\left\{\begin{array}{c}{f}_1(x,t),t\&Element;[{\mathrm{<figure-callout\; id="0"\; label="t"\; filenames="121019155826.png"\; state="\{\{state\}\}">t</figure-callout>}}_0,t_1]\\ f_2(x,t),t\&Element;[{\mathrm{<figure-callout\; id="1"\; label="t"\; filenames="121019155826.png"\; state="\{\{state\}\}">t</figure-callout>}}_1,t_2]\\ f_3(x,t),t\&Element;[{\mathrm{<figure-callout\; id="2"\; label="t"\; filenames="121019155826.png"\; state="\{\{state\}\}">t</figure-callout>}}_2,t_3]\end{array}\right.---\left(1\right)$

Wherein, the characteristic function curve of humidity sensor is to be the function of continuous monotone variation in the input signal range ability; As t1＜t＜t2, obtain f by linear least square ₂(x, t) is at the linear equation of [t1, t2] interior temperature influence; As t0＜t＜t1 or t2＜t＜t3, by the RBF neural network to f ₁(x, t) and f ₃(x, t) carries out the error compensation brought by temperature effect.

As t1＜t＜t2, add that in the described linear equation of finally trying to achieve parametric coefficients ξ compensation temperature produces minor impact, directly try to achieve corresponding input signal x with straight-line equation according to the output after the A/D conversion, simplify the processing mode of output signal.

Error compensation by the RBF neural network is carried out comprises the steps:

1) netinit is chosen n training sample at random as cluster centre c _i(i=1,2 ..., n);

2) grouping of arest neighbors rule is pressed in the training sample set of input, then readjust cluster centre, the c that obtains at last _iCenter for network function;

3) find the solution the variance of network function when being Gaussian function I=1,2 ..., n, wherein c _MaxBe the ultimate range between the selected center;

4) weights between calculating hidden layer and the output layer

(p=1,2 ..., P; I=1,2 ..., n);

5). by formula $J=\underset{k=1}{\overset{L}{\mathrm{\&Sigma;}}}{\left|\right|y_k-\widehat{y_k}\left|\right|}_2=\frac{1}{2}\underset{l=1}{\overset{L}{\mathrm{\&Sigma;}}}\underset{p=1}{\overset{M}{\mathrm{\&Sigma;}}}{(y_{\mathrm{pk}}-\widehat{y_{\mathrm{pk}}})}^2$ y _kDesirable output,

Be reality output, k=1,2 ... L, L, M are empirical values, then calculation of performance indicators J carries out the weights adjustment, wherein, y _kDesirable output,

Be reality output, k=1,2 ... L, L, M are empirical values;

6). judge the performance index J≤ζ that whether satisfies condition, ζ is empirical value, if satisfy, finishes training, otherwise makes J=0 return step 2, repeats above-mentioned training process, until satisfy condition.

Table 1 is the experimental data of actual measurement gained, and the measurement error value under the condition of different temperatures is that measured value deducts standard value, and the measurement error value in the different temperatures situation is depicted as smooth curve, and then analysis temperature is on the impact of humidity sensor.Resulting under different temperatures and different humidity condition the Curve of the Measuring Error of humidity sensor, as shown in Figure 2.

Table 1

Fig. 3 utilizes the linear least square method that linearity range is preferably carried out fitting a straight line, and the straight-line equation that obtains is that y=1.0293*x-1.6659 adds minor impact error amount 0.8, and the input and output behind over-fitting present good linear relationship.Certainly need to choose different values for different sensor model numbers in the actual application process, then just can calculate actual humidity value in this segment according to the straight-line equation of match, compensation precision is higher.

Fig. 4 utilizes the method for RBF neural network to carrying out temperature Compensation Study between inelastic region, temperature range two ends, the part sample in the experimental data as training sample, the part sample is as test sample book, as the data in the input vector, the relative humidity standard value is as desired output temperature value and relative humidity measurement value.Sample value is imported neural network, and the input node gets 4, and the hidden neuron number gets 10, and hidden neuron adopts Gaussian function, and the output layer neuron adopts purelin Linear function, and the humidity value of humidity sensor in the nonlinear temperature interval carried out temperature compensation.The input and output that can obtain after the neural networks compensate present good linear relationship, can access more the humidity value near actual value.

In order to check the effect of the blending algorithm that the present invention proposes, this algorithm and least square polynomial fit and BP neural net method are compared analysis and research.In the experiment simulation process, we find when temperature range larger samples number is more, single BP neural net method learning training speed is obviously slack-off, it is large that error of fitting also becomes, be absorbed in easily local minimum, affect compensation effect, and in the process of least square polynomial solving, when the highest power item of polynomial expression value was larger, fitting precision was relatively high, but two ends present oscillatory occurences, and along with the change of data volume is large, speed is slack-off gradually, and it is large that error also becomes gradually, and it is undesirable that fitting effect becomes.The method that the present invention proposes is owing to be that match is carried out respectively in the by stages, and compensation effect can reach satisfied precision.

Table 2 has provided the humidity match value after the compensation in employing BP neural network and the p-30 ° of C ~ 0 ° C of least square polynomial method and the 20 ° of C ~ 50 ° C temperature ranges, the effect temperature compensation of contrast this paper method, can draw, present in obvious linearity and the nonlinear situation at temperature characteristics, the model of temperature compensation that the algorithm that utilizes least square and RBF neural network to combine is set up, simple, precision is higher, and network learning and training speed is fast.

Table 2

The self-adaptation blending algorithm that the present invention utilizes the RBF neural network to combine with least square, effectively compensation temperature is on the impact of humidity sensor generation, the outstanding advantages of the method is simple simultaneously, the match training speed is fast, compensation precision is high, can be used for the temperature compensation of business humidity sensor, greatly improve accuracy of measurement and the reliability of sensor.

Claims (3)

1. the fusion method of a humidity sensor temperature compensation, the characteristic function of establishing humidity sensor is y=f (x), x ∈ [δ ₀, δ _h], wherein y is the output of sensor, and x is the input of sensor, and input reference signal is δ _h-δ ₀, it is characterized in that, set up temperature parameters t, according to the family curve of humidity sensor under temperature effect, make t0＜t1＜t2＜t3, obtain formula (1),

$y=f(x,t)=\left\{\begin{array}{c}{f}_1(x,t),t\&Element;[t_0,t_1]\\ f_2(x,t),t\&Element;[t_1,t_2]\\ f_3(x,t),t\&Element;[t_2,t_3]\end{array}\right.---\left(1\right)$

Wherein, the characteristic function curve of humidity sensor is to be the function of continuous monotone variation in the input signal range ability; As t1＜t＜t2, obtain f by linear least square ₂(x, t) is at the linear equation of [t1, t2] interior temperature influence; As t0＜t＜t1 or t2＜t＜t3, by the RBF neural network to f ₁(x, t) and f ₃(x, t) carries out the error compensation brought by temperature effect.

2. the fusion method of humidity sensor temperature compensation according to claim 1, it is characterized in that, as t1＜t＜t2, in the described linear equation of finally trying to achieve, add parametric coefficients ξ compensation temperature, directly try to achieve corresponding input signal x with straight-line equation according to the output after the A/D conversion.

3. the fusion method of humidity sensor temperature compensation according to claim 1 is characterized in that, described error compensation of being undertaken by the RBF neural network comprises the steps:

1). netinit, choose at random n training sample as cluster centre c _i(i=1,2 ..., n);

2). the grouping of arest neighbors rule is pressed in the training sample set of input, then readjust cluster centre, the c that obtains at last _iCenter for network function;

3). find the solution the variance of network function when being Gaussian function

I=1,2 ..., n, wherein c _MaxBe the ultimate range between the selected center;

4). calculate the weights between hidden layer and the output layer (p=1,2 ..., P; I=1,2 ..., n);

5). by formula

Then calculation of performance indicators J carries out the weights adjustment, wherein, and y _kDesirable output, Be reality output, k=1,2 ... L, L, M are empirical values;

6). judge the performance index J≤ζ that whether satisfies condition, ζ is empirical value, if satisfy, finishes training, otherwise makes J=0 return step 2, repeats above-mentioned training process, until satisfy condition.

Images