task ThermoBot [8].
This project targets giving keen answers for screening toxic sewage gases and takes a shot at an arrangement of live sewage level discovery and observation. At whatever point a specific edge is crossed, an alarm is sent to whoever is inspecting the conditions from a remote area.
3.3 Methodology
The proposed system can be used to detect the sewer gas level as well as cracks in the drainage pipe system. The overall architecture of the proposed system is shown in Figure 3.1.
The system has three tasks: 1. Sewer gas detection using proposed consequent Tristate pattern, 2. Crack detection, 3. Alert is given when the level exceeds the threshold.
3.3.1 Sewer Gas Detection
A sewer system is an underground system of pipes commonly used to transport wastewater from homes and businesses either to a treatment facility, where the water is treated and released into natural water bodies like lakes and streams or in a river to permanently drain out from the area. The sewer manhole is one of the most important parts of the sewer system. A sewer manhole is a structure through which a person can gain access to the underground wastewater collection system. Manholes are not designed for someone to work in regularly, but workers may need to enter the manhole to complete their jobs such as cleaning, repair, inspection, etc. The neglect of sewage systems is responsible for the deaths of thousands of sewage cleaners throughout the year from accidents and various diseases such as hepatitis and typhoid due to sudden or sustained exposure to hazardous gases like carbon monoxide, hydrogen sulphide, and methane. A better knowledge of hazards in the surroundings is necessary for the prevention of poisonous gases. These gases have to be kept on track so that an enormous rise in the normal level of effluents should be known and corrective measures taken.
Figure 3.1 Architecture of the proposed system.
The first step of the proposed system is to detect sewer gas level. A gas sensor can be used to detect the level of the gas. The level of Carbon Monoxide, Hydrogen sulphide and Methane has been measured using a gas sensor. The sensor data is fed to the proposed Tristate pattern. This pattern trains the data based on continues 3 ones and predicts the threshold for each gas. If the level exceeds the threshold value then it gives the alert.
3.3.1.1 Proposed Tristate Pattern
The gas sensor data is given to the input of the proposed system. The proposed system provides 1 for 3 consequent ones and .5 for 2 consequent ones and .25 for one. The Table 3.1 below shows the structure of the Tristate pattern.
Table 3.1 Tristate pattern.
| Patterns | Score |
|---|---|
| Consequent 3 ones | 3 |
| Consequent 2 ones | 2 |
| Single one | 1 |
Finally, the sum of the values has been calculated using the formula below:
Where C,H,M denotes Carbon Monoxide, Hydrogen sulphide and Methane sensor datas and n,m, p denotes the range of the datas.
All these major components are connected in the arduino board with connecting wires and fitted in the required places (underground drainage pipes) where damaged occurred. It will easily detect the gases and also the leakage of pipes in that specific area. It also gives a notification to the authorities about the leakage by a message and it is able to identify the location by GSM module and fix the problem as soon as possible.
3.3.2 Crack Detection
A semi-automated battery-operated (Figure 3.2) cart is designed with ultrasonic sensors attached to the front. This ultrasonic sensor sends high-frequency sound waves through the transmitter head, which then goes and reflects back on hitting the surface of the track, and comes back to the receiver head. This is then converted to distance.
A threshold distance is set, by measuring the distance between the sensor and the surface of the track. At the time of inspection, if the distance exceeds or is less than the threshold distance, it indicates that a crack is present on the track. When a crack as such is detected, the cart stops there. This is the first level of testing.
As a second level of testing, once the cart stops, a camera attached to the front of the cart is triggered on. This camera then takes a picture of the track where the crack is present. This picture is then sent to the Amazon S3 bucket and image processing is performed on the captured image. This is done so as to detect if it is a major or a minor crack.
In case of a major crack, the cart will store the GPS location and return to the previous checkpoint. In case of a minor crack, the cart will continue with its inspection. In either case, the GPS location and the details of the crack and the status are stored in Amazon EC2 and are sent immediately to the concerned authorities to be rectified as soon as possible.
Figure 3.2 Flow diagram of the automated system.
Cart development and sensor:
For the development of the cart, 12V DC motors can be used. Motor controller L298N is used to control the movement of the cart. Ultrasonic sensor HC-SR04 sends high-frequency sound waves and calculates the distance of the reflected wave. This sensor detects the minute defects with high accuracy. This is a non-destructive testing method.
Raspberry Pi 3:
RPi is a low-cost, small credit-card-sized computer. RPi3 is faster than Arduino.
Cloud server and Image processing:
Host server is AWS EC2. Date, time, message status, ID, sensor value, latitude and longitude are stored in the EC2 database. A picture of the defect is taken and stored in AWS S3 bucket. Gaussian Mixture Model can be used for image processing which comes under supervised learning.
3.3.3 Experimental Setup
Gas Sensor
A gas sensor (Figure 3.3) is utilized to identify the nearness of a hazardous LPG spill in a vehicle or in a work station, stockpiling tank condition.
