Interior air quality monitoring and ventilation control method and system for train

Patent metadata

Field	Value
Publication number	US-11999388-B2
Application number	US-202017292281-A
Country	US
Kind code	B2
Filing date	Jul 29, 2020
Priority date	Aug 2, 2019
Publication date	Jun 4, 2024
Grant date	Jun 4, 2024

Abstract

The present invention discloses an interior air quality monitoring and ventilation control method and system for a train. The interior air quality monitoring and ventilation control method for the train comprises: acquiring multiple groups of interior and exterior air quality detection data; acquiring interior and exterior air comprehensive evaluation index Q 0 and Q 1 by using the experimental data; training an exterior fresh air volume control model if Q 0 ≥Q 1 , or else training an interior air purification control model; detecting the interior and exterior air quality detection data; acquiring the interior and exterior air comprehensive evaluation index Q 0 and Q 1 by using the detection data; if Q 0 ≥Q 1 , calling the exterior fresh air volume control model to obtain the required ventilation volume level and controlling a ventilation system with the output result; otherwise, calling the interior air purification control model to obtain the required ventilation volume level and air purification device power level, controlling the ventilation system and the air purification device with the output results. The present invention can apply suitable ventilation control strategies according to different degrees of air quality, to achieve a health guarantee for the interior air quality of the high-speed train under the conditions of energy conservation and environmental protection.

First claim

The invention claimed is:

1. A modeling method for a train interior air quality monitoring and ventilation control model, comprising the following steps:
A) acquiring a plurality of groups of a first interior air quality detection data and a first exterior air quality detection data;
B) acquiring an interior air comprehensive evaluation index (Q0) by using the first interior air quality detection data, and acquiring an exterior air comprehensive evaluation index (Q1) by using the first exterior air quality detection data;
C) determining the magnitude relationship between the interior air comprehensive evaluation index Q0 and the exterior air comprehensive evaluation index Q1;
D) when the exterior air comprehensive evaluation index is less than the interior air comprehensive evaluation index, training an exterior fresh air volume control model according to the following method:
selecting first groups of corresponding first interior air quality detection data and corresponding first exterior air quality detection data as experimental data;
setting the ventilation volume level of a ventilation system of the high-speed train to a first level;
performing ventilation experiments on the train under the conditions of the first groups of experimental data, to obtain a minimum ventilation volume level that can reduce the air quality to a human health value within 5 minutes, as a ventilation volume level label corresponding to each group of experimental data; and
training an exterior fresh air volume control model by taking the first groups of experimental data as input and taking the ventilation volume level labels corresponding to the first groups of experimental data as output to obtain the exterior fresh air volume control model;

E) when the exterior air comprehensive evaluation index is not less than the interior air comprehensive evaluation index, training an interior air purification control model according to the following method:
selecting second groups of the corresponding first interior air quality detection data and the corresponding first exterior air quality detection data as experimental data;
setting the ventilation volume level of the ventilation system of the high-speed train to the first level, and setting the power level of an air purification device to a second level;
performing ventilation experiments on the train under the conditions of the second groups of experimental data, to obtain a minimum ventilation volume level and a minimum power level of the air purification device that can reduce the interior air quality to the human health value within 5 minutes, wherein the obtained minimum ventilation volume level is used as a ventilation volume level label corresponding to each group of experimental data, and the obtained minimum power level of the air purification device is used as a power level label of the air purification device corresponding to each group of experimental data; and
training an interior air purification control model by taking the second groups of experimental data as input and taking the ventilation volume level labels and the power level labels of the air purification device corresponding to the second groups of experimental data as output to obtain the interior air purification control model; and

F) using the trained exterior fresh air volume control model and the trained interior air purification control model as the train interior air quality monitoring and ventilation control model.

2. The modeling method for a train interior air quality monitoring and ventilation control model according to claim 1 , wherein the first interior air quality detection data and the first exterior air quality detection data both comprise one or more of CO 2 concentration, NO 2 concentration, SO 2 concentration, PM2.5 concentration, VOC concentration, and dust concentration.

3. The modeling method for a train interior air quality monitoring and ventilation control model according to claim 1 , wherein at least one of the first interior air quality detection data and the first exterior air quality detection data are obtained by a multi-point monitoring mode.

4. The modeling method for a train interior air quality monitoring and ventilation control model according to claim 1 , wherein
a calculation method of the interior air comprehensive evaluation indicator Q0 is:
Q 0=interior CO 2 concentration× p 1 +interior NO 2 concentration× p 2 +interior SO 2 concentration× p 3 +interior PM 2.5 concentration× p 4 +interior VOC concentration× p 5 +interior dust concentration× p 6 ; and

a calculation method of the exterior air comprehensive evaluation indicator Q 1 is:
Q 1=exterior CO 2 concentration× p 1 +exterior NO 2 concentration× p 2 +exterior SO 2 concentration× p 3 +exterior PM 2.5 concentration× p 4 +exterior VOC concentration× p 5 +exterior dust concentration× p 6 ;

wherein, P 1 , P 2 , P 3 , P 4 , P 5 , and P 6 are corresponding weights of pollutants.

5. The modeling method for a train interior air quality monitoring and ventilation control model according to claim 1 , wherein in step D), the exterior fresh air volume control model is trained by using a BP neural network algorithm, wherein the weight and threshold of the BP neural network are obtained by quantum particle swarm optimization with self-adaptive weights, comprising:
D1) using a position vector of each quantum particle individual in quantum particle swarms as the weight and threshold of the BP neural network, and initializing the position vector parameter of the quantum particle swarm individual to a random number of [−1, 1];
wherein the number of the quantum particle swarms ranges [25, 70], the number of particles in a quantum particle swarm ranges [5, 50], the maximum number of iterations ranges [400, 1000], the number of iterations for forming an elite swarm ranges [60, 180], the premature convergence determination threshold ranges [0.01, 0.45], and the worst particle variation ratio among the swarms δ% ranges [2%, 7%];

D2) setting a fitness function, and determining the position vector and the number of iterations t of an initial optimal quantum particle individual, wherein t=1;
substituting the weight and threshold corresponding to the position vector of the quantum particle individual into the exterior fresh air volume control model based on the BP neural network, determining the type of an identified vector label by using the exterior fresh air volume control model based on the BP neural network determined from the position vector of the quantum particle individual, and using the reciprocal of the mean square error of the outputted vector label and the actual vector label as a second fitness function;
D3) calculating a swarm fitness variance of each quantum particle swarm, and performing premature convergence determination;
if the swarm fitness variance of the quantum particle swarm is smaller than a premature convergence determination threshold γ, mutating δ% of worst fitness particles and swarm extreme particles in the quantum particle swarm, and using the current best fitness particles as global optimal quantum particle individuals;
D4) determining whether to form an elite swarm;
when the number of iterations is greater than the number of iterations for forming the elite swarm, extracting extreme values of each swarm by means of information sharing between the swarms to form the elite swarm, and turning to step D8), otherwise, turning to step D5);
D5) updating particle parameters of each swarm;
D6) recalculating and comparing the fitness value of each particle; if the fitness value is superior to the current individual extreme value, updating the individual extreme value; comparing the global extreme particles; if the fitness value of a particle is superior

Assignees

• Univ Central South

Inventors

• Liu Hui
• Li Yanfei
• Yang Rui
• Dong Shuqin
• Yu Chengqing

Classifications

• G06N3/0499

  Feedforward networks · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06N3/0985

  Hyperparameter optimisation; Meta-learning; Learning-to-learn · CPC title

• B61D27/009 Primary

  Means for ventilating only (ventilation in general F24F) · CPC title

• B60H1/0073

  Control systems or circuits characterised by particular algorithms or computational models, e.g. fuzzy logic or dynamic models · CPC title