CN103743867A - Kalman filtering formaldehyde detection method based on neural network - Google Patents

Abstract

The invention discloses a Kalman filtering formaldehyde detection method based on a neural network. The method comprises the following steps: (1) initializing a detection environment and determining environmental parameters; (2) simulating data in the detection process to obtain training data of the neural network; (3) establishing the neural network with a two input-one output structure, and by adopting a BP (Back-Propagation) neural network and adding a momentum learning rule, training the neural network; (4) carrying out first detection evaluation; (5) judging whether detection is stopped or not; and (6) if detection is stopped, carrying out dormancy for waiting, and if not, evaluating formaldehyde content by Kalman filtering. According to the method disclosed by the invention, the relationship between detectable quantity and state transfer quantity is established by an offline neural network training method, so that the problem that when conventional Kalman filtering is used to detect formaldehyde, a state equation, particularly state transfer quantity, is hard to determine is solved, and the speed, precision and reliability of formaldehyde detection are greatly improved.

Description

Kalman filtering Analysis Methods for Formaldehyde based on neural network

Technical field

The present invention relates to a kind of Analysis Methods for Formaldehyde, relate in particular to a kind of Kalman filtering Analysis Methods for Formaldehyde based on neural network, belong to detection technique field.

Background technology

Formaldehyde is a kind of colourless, gas of having intense stimulus smell, there is severe toxicity, it mainly applies in timber industry and textile industry, a kind of important industrial chemicals and organic solvent, be widely used in now in the material of house decoration, this makes in new decorating house formaldehyde pollution problem very serious.Prior art mainly contains for the detection method of formaldehyde: spectrophotometric method, electrochemical assay, vapor-phase chromatography, liquid phase chromatography, sensor method etc., although these methods possibility accuracy of detection are higher, but owing to lacking for the analysis and the filtering that detect error, thereby it is larger to detect error, accuracy is not high, to the real content of Formaldehyde in Environment, can not effectively detect.

Kalman filter method is a kind of numerical filter method being proposed by R.E.Kalman nineteen sixty, the representative instance of its processing is limited from one group, comprise noise, prediction in the observation data of object state is estimated to the virtual condition of object, therefore rationally use Kalman filtering can solve well the error filtration problem in formaldehyde examination.For formaldehyde examination problem, the state equation of content of formaldehyde can be expressed as:

X(t+1)=Φ(t)X(t)

Wherein X (t) is t formaldehyde actual content value constantly, and Φ (t) is t state transitions amount constantly, and it represents the t moment and the t+1 relation between formaldehyde actual content constantly.

The observation of formaldehyde examination (detection) equation can be expressed as:

Z(t)=X(t)+v(t)

Wherein Z (t) is t detected value constantly, and X (t) is t formaldehyde actual content value constantly, and v (t) is t detection error constantly, and by law of great number, obeying average is zero, the Gauss normal distribution that variance is definite value.For formaldehyde examination problem, the parameter Φ in state equation (t) and detection error v (t) are determined by actual detection case, be difficult to obtain clear and definite expression formula, so the Kalman filtering of prior art are difficult to directly be used among formaldehyde examination.

Summary of the invention

The object of the present invention is to provide a kind of Kalman filtering Analysis Methods for Formaldehyde based on neural network, by neural network, solve state equation difficult parameters that Kalman filtering runs in the formaldehyde examination problem problem to determine, and used in formaldehyde examination error filtration, to improve the accuracy of existing Analysis Methods for Formaldehyde.

Object of the present invention is achieved by the following technical programs:

A Kalman filtering Analysis Methods for Formaldehyde based on neural network, comprises the following steps:

1) initialization testing environment, determines environmental parameter: use sensor repeatedly to sample the content of formaldehyde in environment, draw one group of sampled data; The maximum detected value of sampled data is designated as to Zmax, and minimum detection value is designated as Zmin, asks this group data mean square deviation and divided by the 2 approximating variances G that obtain detecting error;

2) establishing detection error is G, order zero constantly state value X (0) is respectively Zmin and Zmax, to different state transitions amount Φ, use formula X (t+1)=Φ (t) X (t) and Z (t)=X (t)+v (t) analog detection process, the obedience average that v (t) draws up for Numerical-Mode is zero, variance is the detection error of G, draws one group of detection data Z that different state transitions amount Φ is corresponding;

3) set up the neural network of two input one export structures, two are input as in every group of analog detection data adjacent two detected values constantly, are output as corresponding Φ value, employing BP neural network, additional momentum learning rules, neural network training;

4) start testing process: set initial formaldehyde examination value Z (0) for (Zmax+Zmin)/2, the formaldehyde examination value that now sensor is obtained is designated as to Z (1), by Z (0), Z (1) input neural network, draw Φ (1) value that neural network prediction is estimated, determine this state equation constantly, iteration Kalman Filter Estimation equation, draws formaldehyde estimated value now output display result;

5) judge whether to stop to detect, if do not stopped, going to step 6); If stop detecting, enter dormant state;

6) remember that the formaldehyde examination amount that last sensor is obtained is Z (t-1), the formaldehyde examination amount that current time sensor is obtained is Z (t), by Z (t-1) and Z (t) input neural network, draw the value of the state transitions amount Φ (t) of current time, determine the state equation of current time, remember that last formaldehyde estimated value is

iteration Kalman Filter Estimation equation, draws the formaldehyde estimated value of current time

output display structure;

7) judge whether to want initialization context, if so, go to step 1), if not, go to step 6).

Object of the present invention can also further realize by following technical measures:

The aforementioned Kalman filtering Analysis Methods for Formaldehyde based on neural network, wherein kalman filter method is as follows:

1) pre-estimation:

wherein

for estimated value,

for estimating evaluation, Φ (t) is state transitions amount;

2) calculate pre-estimation covariance matrix:

wherein P (t) is estimate covariance,

for pre-estimation covariance, Φ (t) is state transitions amount, and Φ ' is (t) transposition of Φ (t);

3) calculate kalman gain matrix:

wherein K (t) is kalman gain, and G is for detecting the approximating variances of error;

4) upgrade and estimate:

$\hat{X}(t+1)=\tilde{X}(t+1)+K(t+1)[Z(t+1)-\tilde{X}(t+1)X];$

5) upgrade and estimate covariance matrix:

$P(t+1)=[1-K(t+1)]\tilde{P}(t+1)[1-K(t+1)]+K(t+1)\·G\·K(t+1);$

6) often obtain one-time detection value, iteration above-mentioned steps once.

The aforementioned Kalman filtering Analysis Methods for Formaldehyde based on neural network, wherein additional momentum learning rules are as follows:

$\mathrm{\ω}(t+1)=\mathrm{\ω}\left(t\right)-(1-a)\mathrm{\η}\frac{\∂E_T}{\∂\mathrm{\ω}\left(t\right)}+\mathrm{\α\Δ\ω}\left(t\right)$

Δ ω (t)=ω (t)-ω (t-1) wherein, ω (t) is the weight of each node of neural network, the training error that ET is neural network, η is weight, α is factor of momentum, gets 0.95.

Compared with prior art, the invention has the beneficial effects as follows: the present invention is by the method for off-line training neural network, set up the relation between detection limit and state transitions amount, and then solved legacy card Kalman Filtering when formaldehyde examination, state equation particularly state transitions amount is difficult to definite problem, this makes Kalman filtering can be useful among formaldehyde examination, compared to existing technology, the present invention has used substantially negligible calculated amount, has improved greatly speed, precision and the reliability of formaldehyde examination.

Accompanying drawing explanation

Fig. 1 is process flow diagram of the present invention;

Fig. 2 is building-block of logic of the present invention;

Fig. 3 is the building-block of logic of kalman filter method.

Embodiment

Below in conjunction with the drawings and specific embodiments, the invention will be further described.

The state equation of formaldehyde examination problem and observation (detection) equation can be written as following form:

X(t+1)=Φ(t)X(t)

Z(t)=X(t)+v(t)

Wherein, X (t) is t formaldehyde actual content value constantly, Φ (t) is t state transitions amount constantly, it represents the t moment and the t+1 relation between formaldehyde actual content constantly, X (t) is t formaldehyde actual content value constantly, and it is the Gauss normal distribution that zero variance is definite value that v (t) obeys average for t detection error constantly.

As Fig. 1 carves, showing, is the Kalman filtering Analysis Methods for Formaldehyde process flow diagram based on neural network of the present invention, comprises the following steps:

1) sampling initialization detected parameters, is used sensor to carry out real time sample, and the content of formaldehyde in environment is repeatedly sampled, and draws one group of sampled data; To these group data, its maximum detected value is designated as Zmax, and minimum detection value is designated as Zmin, ask this group data mean square deviation and divided by 2 for detecting the approximating variances G of error;

2) simulative neural network training data, use Computer Numerical Simulation, if detection error is G, Φ gets not the value that t changes in time, it presses 101 values of the every 0.01 one step value of 0.5-1.5, for 101 Φ values, order zero constantly state value X (0) is Zmin, use formula X (t+1)=Φ (t) X (t) and Z (t)=X (t)+v (t) analog detection process, the obedience average that v (t) draws up for Numerical-Mode is zero, variance is the detection error of G, draw the detection data of one group of Z that different Φ is corresponding, same order zero constantly state value X (0) is Zmax, analog detection process draws the detection data of one group of Z that different Φ is corresponding respectively again,

3) neural network training, set up the neural network of two input one export structures, as shown in Figure 2, two are input as in every group of analog detection data adjacent two detected values constantly, be Z (t+1) and Z (t), be output as corresponding Φ value, adopt BP neural network, additional momentum learning rules, neural network training;

Additional momentum learning rules are on traditional BP learning method basis, give renewal momentum when weighting regulates, and can recall like this locally optimal solution of training, concrete update rule as shown in the formula:

$\mathrm{\ω}(t+1)=\mathrm{\ω}\left(t\right)-(1-a)\mathrm{\η}\frac{\∂E_T}{\∂\mathrm{\ω}\left(t\right)}+\mathrm{\α\Δ\ω}\left(t\right)$

Δ ω (t)=ω (t)-ω (t-1) wherein, the weight that ω (t) is each node of neural network is carved the training error that ET is neural network, and η is weight, and α is factor of momentum, gets 0.95.

4) start testing process, set initial formaldehyde examination amount Z (0) for (Zmax+Zmin)/2, the formaldehyde examination amount that now sensor is obtained is designated as to Z (1), by Z (0), Z (1) input neural network, draw Φ (1) value that neural network prediction is estimated, determine this state equation constantly, iteration Kalman Filter Estimation equation, draws formaldehyde estimated value now

output to detecting instrument and show;

As shown in Figure 3, Kalman Filter Estimation is by the form that is expressed as Equation Iterative of traditional least mean-square estimate, carries out filtering estimation, because its real-time, robustness and needed storage space are less, therefore very applicable formaldehyde examination problem, its steps flow chart is:

1, pre-estimation:

$\tilde{X}(t+1)=\mathrm{\Φ}\left(t\right)\hat{X}\left(t\right)$

2, calculate pre-estimation covariance matrix:

$\tilde{P}(t+1)=\mathrm{\Φ}\left(t\right)P\left(t\right){\mathrm{\Φ}}^{\′}\left(t\right)$

3, calculate kalman gain matrix:

$K(t+1)=\tilde{P}(t+1){[\tilde{P}(t+1)+G]}^{-1}$

4, upgrade and estimate:

$\hat{X}(t+1)=\tilde{X}(t+1)+K(t+1)[Z(t+1)-\tilde{X}(t+1)X];$

5, upgrade and estimate covariance matrix:

$P(t+1)=[1-K(t+1)]\tilde{P}(t+1)[1-K(t+1)]+K(t+1)\·G\·K(t+1);$

6, often obtain one-time detection value, iteration above-mentioned steps once;

Wherein,

for estimated value,

for estimating evaluation, P (t) is estimate covariance,

for pre-estimation covariance, Φ (t) is state transitions amount, and Φ ' is (t) transposition of Φ (t), in this problem, equate with Φ (t), K (t) is kalman gain, and G is step 1) in the approximating variances that obtains, Z (t) is for detecting data.

5) operation to detecting instrument according to user, judges whether to stop to detect, if do not stopped, going to step 6); If stop detecting, enter dormancy waiting status;

6) remember that the formaldehyde examination amount that last sensor is obtained is Z (t-1), the formaldehyde examination amount that now sensor is obtained is Z (t), by Z (t-1) and Z (t) input neural network, draw the value of state transitions amount Φ (t) now, determine this state equation constantly, remember that last formaldehyde estimated value is

iteration Kalman Filter Estimation equation, draws formaldehyde estimated value now

output to instrument and show;

7) judge that whether detecting instrument initialization button is pressed, and if so, goes to step 1), if not, go to step 6).

The present invention is based on the Kalman filtering Analysis Methods for Formaldehyde of neural network, can be used for the intelligent instrument that environment content of formaldehyde is detected, by formaldehyde quantity detection sensor, carry out data acquisition, pass to the microprocessor of detecting instrument, microprocessor uses method of the present invention, obtains accurate and effective content of formaldehyde.

In addition to the implementation, the present invention can also have other embodiments, and all employings are equal to the technical scheme of replacement or equivalent transformation formation, all drop in the protection domain of requirement of the present invention.

Claims (3)

1. the Kalman filtering Analysis Methods for Formaldehyde based on neural network, is characterized in that, comprises the following steps:

1) initialization testing environment, determines environmental parameter: use sensor repeatedly to sample the content of formaldehyde in environment, draw one group of sampled data; The maximum detected value of sampled data is designated as to Zmax, and minimum detection value is designated as Zmin, asks this group data mean square deviation and divided by the 2 approximating variances G that obtain detecting error;

2) establishing detection error is G, order zero constantly state value X (0) is respectively Zmin and Zmax, to different state transitions amount Φ, use formula X (t+1)=Φ (t) X (t) and Z (t)=X (t)+v (t) analog detection process, the obedience average that v (t) draws up for Numerical-Mode is zero, variance is the detection error of G, draws one group of detection data Z that different state transitions amount Φ is corresponding;

3) set up the neural network of two input one export structures, two are input as in every group of analog detection data adjacent two detected values constantly, are output as corresponding Φ value, employing BP neural network, additional momentum learning rules, neural network training;

4) start testing process: set initial formaldehyde examination value Z (0) for (Zmax+Zmin)/2, the formaldehyde examination value that now sensor is obtained is designated as to Z (1), by Z (0), Z (1) input neural network, draw Φ (1) value that neural network prediction is estimated, determine this state equation constantly, iteration Kalman Filter Estimation equation, draws formaldehyde estimated value now

output display result;

5) judge whether to stop to detect, if do not stopped, going to step 6); If stop detecting, enter dormant state;

6) remember that the formaldehyde examination amount that last sensor is obtained is Z (t-1), the formaldehyde examination amount that current time sensor is obtained is Z (t), by Z (t-1) and Z (t) input neural network, draw the value of the state transitions amount Φ (t) of current time, determine the state equation of current time, remember that last formaldehyde estimated value is

iteration Kalman Filter Estimation equation, draws the formaldehyde estimated value of current time

output display structure;

7) judge whether to want initialization context, if so, go to step 1), if not, go to step 6).

2. the Kalman filtering Analysis Methods for Formaldehyde based on neural network as claimed in claim 1, is characterized in that, comprises the following steps:

1) pre-estimation:

wherein

for estimated value, for estimating evaluation, Φ (t) is state transitions amount;

2) calculate pre-estimation covariance matrix:

wherein P (t) is estimate covariance,

for pre-estimation covariance, Φ (t) is state transitions amount, and Φ ' is (t) transposition of Φ (t);

3) calculate kalman gain matrix:

wherein K (t) is kalman gain, and G is for detecting the approximating variances of error;

4) upgrade and estimate:

$\hat{X}(t+1)=\tilde{X}(t+1)+K(t+1)[Z(t+1)-\tilde{X}(t+1)X];$

5) upgrade and estimate covariance matrix:

$P(t+1)=[1-K(t+1)]\tilde{P}(t+1)[1-K(t+1)]+K(t+1)\·G\·K(t+1);$

6) often obtain one-time detection value, iteration above-mentioned steps once.

3. the Kalman filtering Analysis Methods for Formaldehyde based on neural network as claimed in claim 1, is characterized in that, described additional momentum learning rules are as follows:

$\mathrm{\ω}(t+1)=\mathrm{\ω}\left(t\right)-(1-a)\mathrm{\η}\frac{\∂E_T}{\∂\mathrm{\ω}\left(t\right)}+\mathrm{\α\Δ\ω}\left(t\right)$

Δ ω (t)=ω (t)-ω (t-1) wherein, ω (t) is the weight of each node of neural network, the training error that ET is neural network, η is weight, α is factor of momentum, gets 0.95.

Images