On-line multi-component gas analysis system based on spectral absorption technology and analysis method thereof

An on-line multi-component gas analysis system based on a spectral absorption technology, characterized by comprising a housing, wherein the housing is provided with an air inlet hole and an air outlet hole which are respectively connected with an air inlet pipe and an air outlet pipe, and the housing is further provided with a dust removal and dehumidification device, a micro vacuum pump, an electronic flowmeter, and a plurality of different gas sensors;

• the dust removal and dehumidification device, which is connected with the air inlet hole through a pipeline for performing dust removal and dehumidification treatment on a gas entering from the air inlet hole;
• the micro vacuum pump, one end of which being connected with the dust removal and dehumidification device through a pipeline, and the other end of which being connected with the electronic flowmeter, and the electronic flowmeter being configured to control and acquire the flow rate of the input gas;
• and
• the plurality of different gas sensors, which are connected in series through a pipeline, a beginning side of the series connection of the plurality of different gas sensors is connected with the electronic flowmeter for detecting the gas concentration in the gas passing through the electronic flowmeter, a tail side of the series connection of the plurality of different gas sensors is connected to the air outlet hole through a pipeline for discharging the gas passing through the plurality of different gas sensors in sequence through the air outlet hole;
• the plurality of different gas sensors comprise an oxygen sensor, a methane sensor, a carbon monoxide sensor, a carbon dioxide sensor, a hydrogen sulfide sensor, a sulfur dioxide sensor, a nitrogen dioxide sensor and an ammonia sensor;
• the on-line multi-component gas analysis system based on the spectral absorption technology further comprises a main board, and a central processing module, a data storage module, a data analysis module and a communication module are provided on the main board;
• the central processing module comprises a signal converter and a processor;
• the signal converter is respectively connected with the micro vacuum pump, the electronic flowmeter, and the plurality of different gas sensors for receiving electrical signals transmitted by the micro vacuum pump, the electronic flowmeter, and the plurality of different gas sensors, and converting the received electrical signals into digital signals;
• the processor is connected with the signal converter for receiving the digital signals transmitted by the signal converter, and processing the data signals;
• the data storage module is connected with the processor for receiving data sent by the processor, and storing the received data;
• the data analysis module is connected with the data storage module for acquiring data transmitted by the plurality of different gas sensors in the data storage module, and performing analysis on the data;
• and
• the communication module is configured to perform data transmission with an outbound device;
• an analysis method of the data analysis module comprises the following steps:
• Step a, generating local feature level data with an SVM algorithm;
• taking the concentration data of the plurality of different kinds of gases at each time in the data analysis module as an eigenvector of a data sample, then inputting a typical non-linear eigenvalue vector into a data fusion processing center, utilizing a non-linear mapping function of a support vector machine to perform fusion recognition on various symptom information obtained by a network, and generate the local feature level data;
• Step b, constructing an evidence set of a multi-component gas overrun value according to a data fusion D-S evidence body fusion method, comprising the following steps:
• Step b1, according to the number of times that a multi-component gas overrun occurs of the local feature level data, establishing an identification architecture of a multi-component gas overrun D-S evidence theory as:
• Θ={FAFB} (1)
• wherein:
• θ represents an identification set of the multi-component gas overrun D-S evidence theory;
• FA represents the number of times that a multi-component gas overrun occurs;
• and
• FB represents the number of times that a multi-component gas overrun does not occur;
• the multi-component gas overrun is specifically:
• a concentration of one or more gases measured in a specific test environment detected by the methane sensor, the carbon monoxide sensor, the carbon dioxide sensor, the hydrogen sulfide sensor, the sulfur dioxide sensor, the nitrogen dioxide sensor, and the ammonia sensor exceeds a threshold value;
• Step b2, taking the established support vector machine local feature level data as an input of a D-S multi-source information dissolution evidence body, calculating an overrun uncertainty probability, which is implemented as follows:
• m
• ⁡
• (
• θ
• )
• =
• E
• n
• (
• 2
• )
• E
• n
• =
• ∑
• i
• =
• 1
• N
• (
• p
• ⁡
• (
• F
• i
• )
• -
• y
• ⁡
• (
• E
• i
• )
• )
• 2
• (
• 3
• )
• wherein in the formula (2) and formula (3):
• m(θ) represents an overrun uncertainty probability;
• p(Fi) represents an actual output of the SVM;
• y(Ei) represents a desired output;
• En represents an overrun SVM output error;
• and
• N represents the number of support vectors;
• Step b3, in the multi-component gas overrun data fusion identification process, determining that the multi-component gas overrun data has no intersection, wherein, when mi(C)={m1,m2, . . .,mn}, there is an evidence body composed of n transfer functions, which is synthesized according to the following formula:
• m
• ⁡
• (
• C
• )
• =
• ∑
• ⋂
• F
• j
• =
• C
• ∏
• i
• =
• 1
• n
• m
• i
• (
• F
• j
• )
• 1
• -
• ∑
• ⋂
• F
• j
• =
• φ
• ∏
• i
• =
• 1
• n
• m
• i
• (
• F
• j
• (
• 4
• )
• in the formula (4):
• m(C) represents a source information fusion evidence body synthesized by n transfer functions;
• Fj represents the number of times that the multi-component gas overrun occurs, j=1,2, . . . , t, t is the number of overrun gas types;
• and
• φ represents an intersection of multi-component gas overrun ratios;
• and
• Step c, determining whether the multi-component gas is overrun according to a decision rule for the multi-component gas overrun early warning identification:
• wherein, the decision rule for the multi-component gas overrun early warning identification is:
• m(θ)<0.01 and m(C)>0.5;
• and
• when the above decision rule is satisfied, considering that the gas concentration of the currently detected environment is excessive, and performing early warning.