JP3467910B2 - Tunnel ventilation control method - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to tunnel ventilation control performed by increasing or decreasing the number of operating blower devices in a tunnel, and more particularly to a method of tunnel ventilation control for providing an optimum number of blower devices. It is a thing.

[0002]

2. Description of the Related Art As shown in FIG. 2, in a tunnel 51, in order to suppress air pollution, an air blower 52 is provided to ventilate the tunnel itself by ventilation. The blower device 52 is, for example, an axial fan attached to a ceiling portion in the tunnel or the like, and a plurality of blower devices 52 are arranged at predetermined intervals in the tunnel longitudinal direction. For example, blowers are installed at 20 places in a 4000 m tunnel. The ventilation capacity depends on the number of blowers 52 and each blower capacity, but in this example, each blower 52 is on / off controlled to either rated operation or stop. In this case, the blower 5
Ventilation capacity can be increased / decreased by adjusting the number of operating units in 2. Energy consumption also increases or decreases as the number of operating vehicles increases or decreases.
On the other hand, the degree of air pollution inside the tunnel is constantly changing depending on the intensity of traffic and various other conditions. Therefore, in order to maintain the environment in the tunnel at a constant standard without wasting energy, it is necessary to adjust the number of operating blowers according to the intensity of traffic and the degree of air pollution.

In the prior art, traffic volume, air transparency, harmful gas concentration, etc. are used as data for determining the number of operating blowers. The traffic volume is a count of the number of vehicles passing within a certain period of time, and is an index of the amount of exhaust gas generated by adding speed and vehicle type to this.
Air transparency and harmful gas concentration are physical quantities that quantitatively indicate the degree of air pollution. In addition, in order to make the control of the number of operating vehicles correspond to the change in the degree of air pollution in advance, the predicted value after a certain time may be used for the traffic volume and the air transparency. For example, a mathematical model is constructed so that the predicted value of air transparency is output when the predicted value of traffic volume and the current value of air transparency are input, and the predicted value of air transparency is output, and the air transparency is constant. Determine the number of operating units to obtain the ventilation capacity that falls within the standard.

[0004]

However, the traffic volume and the air transparency do not always correspond to the degree of air pollution. For example, the air pollution is not so severe even though there is a lot of traffic, or the air transparency is high. Even if the value is high, the concentration of harmful gas may be high. Therefore, the number of operating blowers uniquely determined from the output of the mathematical model is not necessarily the optimum number of operating units, and energy may be wasted or air pollution may exceed a certain standard.

Further, today, in a system in which the input / output relationship is complicated or ambiguous like the relationship between the traffic volume and the air transparency and the actual degree of air pollution, the control amount is controlled by a neural network, AI, fuzzy theory, etc. It is known how to make decisions based on artificial intelligence. As a result, it is possible to configure an artificial intelligence number-of-driving-vehicles determining unit that inputs traffic volume and air transparency and outputs the number of operating vehicles. However, even though these artificial intelligences can handle complex ambiguous input-output relationships,
It is inconvenient if the input itself is incorrect. When a predicted value is used to input traffic volume or air transparency, if the prediction is inaccurate, the optimum number of operating vehicles cannot be obtained after all.

SUMMARY OF THE INVENTION An object of the present invention is to solve the above problems and provide a tunnel ventilation control method for providing an optimal number of blowers to be operated.

[0007]

In order to achieve the above object, the present invention provides a plurality of air blowers in a tunnel,
In the tunnel ventilation control method that controls the operation of the blower according to the intensity of traffic and the degree of air pollution, the traffic volume is predicted and the air transparency is measured, and the predicted value of the traffic volume and the current value of the air transparency are also calculated. And stop all the
And the number of operating units in the tunnel are input to a physical model that simulates air pollution in the tunnel, and the air transparency is calculated for each operating number of operating units from the stop of all blowers to the operation of all of the blowers in this physical model. Output the predicted value of
The predicted deviation of the predicted value of the air transparency is corrected by a neural network that has learned the predicted deviation of the physical model using the output results, and the corrected predicted value of each of the operating vehicle number candidates is used, and air pollution is an environmental standard. The number of operating units is determined by fuzzy theory so that the energy consumption of the air blower is small and the number of times of stopping and operating the air blower is small.

The traffic volume prediction value is obtained by predicting the traffic volume at each time by a statistical traffic volume model that statistically predicts the accumulated past traffic volume according to conditions such as days of the week, while the current traffic volume is calculated every predetermined short time. From the time series of the difference between the measured current traffic volume and the predicted value of the traffic volume, regression prediction of the difference between the current traffic volume and the predicted value of the traffic volume by the autoregressive model is performed, and the statistical predicted value is calculated. May be corrected with the above regression prediction value.

[0009]

With the above structure, the physical model imitates air pollution in the tunnel and outputs a predicted value. In other words, if all the air blowers are stopped, what happens to the current air transparency due to subsequent changes in traffic volume?
・ Outputs the predicted value of air transparency for each candidate for the number of operating units, which will be what happens if all units are operating. Since this prediction does not always correspond to the air pollution that is occurring in the actual tunnel, there is a misalignment. Here, since the neural network learns the tendency of the physical model to make a prediction deviation based on the actual results, the prediction deviation can be well corrected. Therefore, the predicted value of the air transparency for each candidate number of operating vehicles is an accurate prediction of the air transparency due to air pollution that may occur in an actual tunnel. Therefore, the number of operating vehicles determined by the fuzzy theory using the corrected predicted value for each candidate for number of operating vehicles becomes optimum.

Traffic forecasting by a statistical traffic model is well known. However, the traffic volume does not always follow the statistics, and there is a deviation from the actual traffic volume. Therefore, the actual traffic volume is measured and used for correction, but what is needed is not the current traffic volume but the predicted value. Therefore, the difference between the traffic volume and the predicted value of the traffic volume after the present is regression-predicted from the measured short-term fluctuation of the traffic volume, and the statistical prediction value is corrected by the regression prediction value. This makes it possible to predict traffic volume that is close to reality.

[0011]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described in detail below with reference to the accompanying drawings.

As shown in FIG. 1, the tunnel ventilation control system according to the method of the present invention mainly comprises a traffic volume predicting unit 1 for predicting traffic volume and a V for predicting air transparency (VI).
The I prediction unit 2 and the operating vehicle number determining unit 3 that determines the operating vehicle number.

The traffic volume prediction unit 1 is a statistical traffic volume model 1
1 and traffic forecast correction autoregressive model (AR model) 12
Consists of.

The statistical traffic volume model 11 has a database in which past traffic volume data by vehicle type is accumulated, and these traffic volume data are statistically classified according to conditions such as the day of the week, and the traffic volume forecast value is calculated according to the day of the week. Can be output. The statistical traffic model 11 of this embodiment inputs a time and outputs the past average value of the day corresponding to the time as a predicted value. Therefore, a statistical prediction is obtained that there is this amount of traffic at this time on this day of the week.

The traffic volume prediction correction autoregressive model 12 is used to perform regression prediction of the difference between the traffic volume and the traffic volume prediction value after the present from the time series of the difference between the traffic volume measurement value and the traffic volume prediction value. is there. Traffic counter is a traffic counter (TC)
Is going on. The traffic counter counts the number of vehicles that have passed within a predetermined time by vehicle type. In this embodiment, the counting result (TC measurement value) is sampled every 5 minutes and stored in a memory (not shown) as a time series.

The traffic volume prediction unit 1 is adapted to correct the statistical prediction value with the regression prediction value. Therefore, the corrected predicted value of the traffic volume is a predicted value that takes into consideration that the actual traffic volume deviated from the average data will change thereafter. V
The traffic forecast value for each vehicle type is output to the I predictor 2.

The VI prediction unit 2 includes a VI prediction physical model 21.
And VI prediction correction neural network 22.

The VI prediction physical model 21 simulates air pollution in a tunnel, and inputs a traffic volume prediction value, an air transparency (VI) measurement value, a wind direction wind speed (WS) measurement value, and a fan number candidate. , The air transparency predicted value (VI predicted value) is output. The traffic volume prediction value is given from the traffic volume prediction unit 1. The air transparency is obtained by providing a light source and an optical sensor in a tunnel, transmitting light at a predetermined distance, and measuring the transparency of air from the amount of the light transmitted.
The wind direction is measured by arranging an anemometer in the tunnel and measuring the air flow in the tunnel.

The air flow in the tunnel is the air flow due to the inflow of natural wind outside the tunnel and the air flow accompanying the traveling of the vehicle, plus the air flow due to the operation of the blower. . Ventilation also follows the flow of air in the tunnel, with natural diffusion overlapping.

The fan number candidate is a variable that designates a case where all the blowers (fans) are stopped, that is, 0 to 1 to 2 ... 21. This VI prediction physical model 21
Is the total amount of exhaust gas that takes into account the estimated traffic volume and the exhaust volume for each type of vehicle, calculates the total exhaust gas volume, and determines how much pollution will progress with the given ventilation capacity. To predict.

The VI prediction correction neural network 22 is for learning the tendency of the VI prediction physical model 21 to be misaligned using the input / output results. Of course, re-learning can be accumulated.

The VI prediction unit 2 includes a VI prediction physical model 21.
The predicted value of is corrected by the value output from the VI prediction correction neural network 22. Therefore, the corrected VI predicted value for each operating vehicle number candidate is a VI predicted value that accurately predicts air transparency due to air pollution that may occur in an actual tunnel depending on the operating vehicle number.

The operating number determining unit 3 comprises a fuzzy logic (optimal fan number determining fuzzy logic) 31,
Enter the predicted value, candidate number of fans, and CO concentration, and use fuzzy theory to prevent air pollution from exceeding the environmental standard, reduce the energy consumption of the entire fan, and stop the fan to reduce fan motor consumption. -It outputs the optimal number of fans so that the number of operations is reduced. CO (carbon monoxide) is one of the harmful gases, and C
The O concentration is measured every predetermined time by providing a CO concentration meter, a gas analyzer, etc. in the tunnel. The signal of the optimum number of fans is output to an operation combination device (not shown).

[0024]

The present invention exhibits the following excellent effects.

(1) Since the predicted value of the air transparency becomes accurate, it is possible to give the optimum number of operating blowers.

[Brief description of drawings]

FIG. 1 is a control diagram of a tunnel ventilation control system for implementing the method of the present invention.

FIG. 2 is a partial sectional view of a tunnel having ventilation equipment.

[Explanation of Codes] 1 Traffic Volume Prediction Unit 2 VI Prediction Unit 3 Driving Number Determining Unit 11 Statistical Traffic Volume Model 12 Traffic Volume Prediction Correction Autoregression Model 21 Physical Model (VI Prediction Physical Model) 22 Neural Net (VI Prediction Correction Neural NET) 31 Fuzzy theory (Fuzzy logic to determine the optimal number of fans)

─────────────────────────────────────────────────── ─── Continuation of front page (72) Shigeki Murayama Inventor Shigeki Murayama 1-15-3 Toyosu, Koto-ku, Tokyo Ishikawa Shima Harima Heavy Industries, Ltd. Toni Technical Center (72) Inventor Yoshitake Wada Isogo-ku, Yokohama, Kanagawa Prefecture Shin-Nakahara-cho 1 Ishikawajima Harima Heavy Industries Co., Ltd. Yokohama Engineering Center (72) Inventor Norio Takahashi 3-2-16 Toyosu, Koto-ku, Tokyo Ishikawa Shimaharima Heavy Industries Co., Ltd. Toyosu General Office (72) Inventor Nagano Honors, 3-2-16 Toyosu, Koto-ku, Tokyo Ishikawa Shima Harima Heavy Industries Co., Ltd. (72) Inventor Yasuyuki Sakamoto 3-2-16, Toyosu, Koto-ku, Tokyo Ishikawa Shima Harima Heavy Industries Co., Ltd. In-office (72) Inventor Hiroyasu Onozuka 3-2-16 Toyosu, Koto-ku, Tokyo Ishikawa Shimaban (56) Reference JP-A-6-185300 (JP, A) JP-A-7-62998 (JP, A) JP-A-6-42298 (JP, A) JP-A-7- -127395 (JP, A) JP-A-5-141200 (JP, A) (58) Fields investigated (Int.Cl. ⁷ , DB name) E21F 1/00 F24F 7/007 F24F 7/06

Claims (2)

(57) [Claims] 1. A tunnel ventilation control method for arranging a plurality of blowers in a tunnel and controlling the operation of the blowers according to the intensity of traffic and the degree of air pollution.
Predict the traffic volume and measure the air transparency, and predict the traffic volume, the current value of the air transparency, and stop all vehicles.
The number of operating units up to switching is input to a physical model that simulates air pollution in the tunnel, and the air is added to this physical model when operating with candidate operating units from stopping all units to operating all units. The predicted value of the transparency is output, and the predicted deviation of the predicted value of the air transparency is corrected by a neural network that has learned the predicted deviation of the physical model using the results of these inputs and outputs, and correction for each of the operating vehicle number candidates Using the predicted value, determine the number of operating units by fuzzy theory so that air pollution does not exceed the environmental standard, the energy consumption of the blower is low, and the number of times of stopping and operating the blower is small. A tunnel ventilation control method characterized by the above. 2. The predicted value of the traffic volume is predicted at each time by a statistical traffic volume model that statistically predicts the accumulated traffic volume according to conditions such as days of the week, and at the same time, at a predetermined short time. From the time series of the difference between the measured current traffic volume and the predicted value of the traffic volume, regression prediction of the difference between the current traffic volume and the predicted value of the traffic volume by the autoregressive model is performed, and the statistical prediction value is set to the above. The tunnel ventilation control method according to claim 1, wherein the tunnel ventilation control method is corrected with a regression prediction value.