Author Email: firstname.lastname@example.org
Sensors connected to the Internet of Things (IoT) generates billions of data points that can lead to real-time insights. The sensor data stream-in at an interval of one second, which is equivalent to 86400 rows of data per day. A correct predictive analytics of sensor stream information can be used to estimate missing values or to replace incorrect readings captured because of the events of malfunctioning sensors or broken communication channel. It can also be used to anticipate situations that help in various decision makings, including maintenance of a stable internal cold storage temperature and operations. In this paper, a predictive analytics for anomaly detection in the Internet of Things enabled smart cold storage warehouse is proposed using Apache Spark Service and IBM SPSS Modeller which can generate a predicted value for temperature readings. Then a set of decision rules based on both the sensed data and the predicted temperature readings is developed to generate an alert for spikes and dips in temperature. This system can be helpful to cold storage owners to achieve zero wastage of perishable food items during cold storage warehousing.
Anomaly Detection, Internet of Things, Predictive Analytics, Smart Cold Storage
A grain saved is a grain produced. These golden words remain as a mere proverb when one visualizes the quantum of post-harvest wastages and losses of agricultural produce due to inadequate and inefficient storage facilities. The reason for such huge post-harvest losses mainly attributes to lack of scientific storage facilities and improper transportation, poor front-end infrastructure, such as inadequate warehousing facilities. It has been estimated that 40% of fruits and vegetables grown in India gets wasted every year . The major reasons are a lack of storage infrastructure clubbed with old-fashioned storing methods to stock the produced capacity . Moreover, a dearth of continuous electricity and absence of any warning systems add to the troubles of cold storage owners. Evidently, there is a lack of comprehensive technology for the entire produce range of vegetables, fruits, flowers, meat and cereals .
The combination of many small things can make a very big difference. Sensors connected to the Internet of Things generate billions of data points that can lead to real-time insight. With the use of cutting-edge analytics tools that scale on big data, it is likely to extract significant information that supports decision makings, estimate missed values or replace incorrect readings . In a predictive analytics blending, historical data and real-time sensor data with external data, such as weather data helps to predict estimate missed values, or to replace inappropriate readings due to faulty sensors or damaged communication channel.
Our proposed system uses predictive analytics techniques to make real-time predictions, learning from internal and external temperature and weather conditions to make suggestions to maintain optimal internal cold storage temperature. This could potentially be an aid for farmers for storing perishable goods over a longer period. The goal is to maintain an optimal temperature inside cold storage facility to retain the quality of food items over a longer period.
In this paper, we have discussed that machine learning methods can be used along with IBM Watson IoT Platform in order to provide a proactive solution for Cold storage warehousing using the Internet of Things applications. Our proposed system is flexible in nature and is also capable to deal with dynamic environments as opposed to the contemporary methods.
Different data streams have a different type of error in prediction and this error affects the event. In future, we aim to work on the modeling of our system in order to predict more complex events occurred due to change in temperature readings. We also look forward to test our system on other Internet of Things scenarios and high-velocity data.
234 total views, 2 views today