REFRIGERANT LEAK PREDICTION IN SUPERMARKETS USING EVOLVED NEURAL NETWORKS
Dan W. Taylor1,2, David W. Corne1,
1 University of Reading, UK, 2 JTL Systems Ltd, UK
,
ABSTRACT
The loss of refrigerant gas from commercial refrigeration systems is a major maintenance cost for most supermarket chains. Gas leaks can also have a detrimental effect on the environment. Existing monitoring systems maintain a constant watch for faults such as this, but often fail to detect them until major damage has been caused. This paper describes a system which uses real-world data received at a central alarm monitoring centre to predict the occurrence of gas leaks. Evolutionary algorithms are used to breed neural networks which achieve usefully high accuracies given limited training data.
1. INTRODUCTION
In recent years, large supermarket chains in the UK have become increasingly aware of refrigeration systems in their stores. This has happened for a number of reasons, most notably because refrigeration is one of the largest costs when setting up and running a store and there are a number of ways in which the associated systems can be optimised to save money. Now, with the added pressures placed upon those who operate commercial refrigeration systems by environmental legislation, such as the Kyoto and Montreal protocols and ever increasing energy costs, the optimisation of refrigeration systems is more important than ever. See [1] for a more detailed review of this.
Individual cabinets and cold-rooms within a typical UK supermarket are part of a complex system of interdependent items of machinery, electronic and computer control units and many hundreds of metres of pipe-work and cabling.
Unlike the small, sealed refrigerators which can be found in most of our homes, the refrigeration systems to be found in supermarkets are fed with refrigerant via a network of piping which runs under the floor of the store. Refrigerant is allowed to evaporate within cabinets to absorb heat, and the resulting hot gas is then pumped to condensers outside the store. Large electrically powered compressors, situated away from the shop floor, are used to facilitate this.
As might be expected, the presence of refrigerant gas in this large, complex mechanical system inevitably leads to the occasional leak. A larger supermarket will have around 100 individual cooled cases and the associated refrigeration system can hold around 800kg of refrigerant. Refrigerant costs around £15 per kilogram and can have detrimental effects if leaked into the atmosphere. It is therefore imperative that leaks from refrigeration systems be minimised, both from financial and environmental points of view.
JTL Systems Ltd (www.jtl.co.uk) manufacture advanced electronic controllers which control and co-ordinate refrigeration systems in supermarkets. These systems, as well as controlling cabinet temperature, gather data on various parameters of the store-wide refrigeration system. These data are used to optimise the operation of machinery and schedule defrosts, whilst also being used to generate alarms.
Alarms are essentially warnings of adverse conditions in equipment. Alarms are transmitted, via a modem link, to a central monitoring centre. At this monitoring centre, trained operators watch for serious events and call the appropriate store staff or maintenance personnel to avert situations where stock may be endangered.
Gas losses have been highlighted by JTL and their customers (major supermarket chains) as a very important area in which to concentrate resources. There are essentially two types of gas leak:
· Fast: Equivalent to a burst tyre on a car: a large crack or hole in piping or machinery causes gas to be lost quickly and the refrigeration system to be immediately impaired. Fast leaks can be detected immediately at the JTL monitoring centre and the appropriate action taken.
· Slow: Equivalent to a slow puncture: gas slowly leaks from the system causing a seemingly unrelated series of alarms and alterations in the performance of the system. This type of leak is more frequent and can be much harder to detect. JTL’s customers tend to lose more money through slow leaks than through fast leaks.
This paper details work undertaken to develop systems which use alarm data, gathered from refrigeration systems in supermarkets, to predict/detect the occurrence of slow gas leaks. There is a clear commercial requirement for such a system as it will allow pre-emptive maintenance to be scheduled, thus minimising the amount of gas allowed to leak from the system.
The prediction/classification technique used in this paper is an extension of that presented in [2]. Neural networks are trained using a combination of evolutionary algorithms and traditional back propagation learning. This training scheme has been shown to be marginally more effective than evolved rule-sets and back propagation used in isolation.
A description of the data available for prediction systems and the various pre-processing operations performed upon it can be found in section 2. Section 3 goes on to describe the EA/BP hybrid training system in more detail. In section 4 we outline the various experiments performed and their results and finally, a concluding discussion and some suggested areas for further exploration can be found in section 5.
2. ENGINEERING AND ALARM DATA
There are two important data sets which must be combined in order to create training data suitable for the task in hand. These are outlined here, along with details of how they were combined and used to produce appropriate training data.
2.1. ALARM DATA
As previously mentioned, when adverse conditions are detected by on-site monitoring and control hardware they are brought to the attention of operators at JTL’s dedicated monitoring centre. The Network Controller, which is the principle component of the in-store refrigeration control system, uses its in-built modem to send a small package of data to the monitoring centre via the telecommunications network. This data package is known as an Alarm and contains useful information, including:
· The id number of the unit which raised the alarm
· The nature of the alarm conditions and any related information – such as temperature or pressure readings
· The time at which the alarm was first raised
Information from alarms is copied to a large relational database called “Site Central”. Alarm data has been archived here for almost three years and well over two million individual alarm records are stored. These alarms correspond to 40,000 control/monitoring units at 400 stores monitored by JTL for its customer.
A few human experts can diagnose problems with refrigeration systems using this alarm data. Some types of fault, gas loss in particular, have a well defined, but often quite subtle, pattern of events that can be detected by those in the know. Due to training and resource issues at the monitoring centre, staff have neither the time nor the expertise required to watch for these patterns.
As the receipt of an alarm is a discrete event, without any duration, it was necessary to present our prediction system with a series of categorised alarm totals. This gives us a list of small valued integers. We create a vector of n samples, each of length t. This covers a period of n´t hours. For each sample period we create a three-tuple of values corresponding to the sum totals of alarms occurring within that sample period, in each of three categories:
· Plant alarms (compressors/machinery)
· Coldroom alarms (large cooled storerooms)
· Cabinet alarms (refrigerated cabinets on the shop floor)
Thus, for a vector where n = 3 and t = 8, we have three values, corresponding to plant, coldroom and cabinet alarm totals for each of our eight sample periods, spanning 24 hours altogether.
2.2. ENGINEERING DATA
In order to train prediction systems to recognise alarm patterns associated with a gas loss it is important to have a large set of training data, containing patterns corresponding to previous gas loss events. The record of gas leaks for the period between 1st Jan 2000 and 5th April 2002 was obtained from a maintenance logging system. This data records the date on which an engineer attended a site and what action was taken, containing altogether 240 engineering visits corresponding to gas losses over the two year period.
Sadly the engineering logs do not record an exact time for the gas loss event, only the date on which the engineer visited. This means that choosing an input vector for our classifier which immediately precedes the gas loss event is not possible. As a compromise the input vector’s last sample ends at 00:00 the day the engineer visited. So the gas loss could have occurred between one second and twenty four hours after the end of our input vector.
Our inability to select an input vector which immediately precedes a gas loss event is compounded by the fact that slow gas leaks take place over a period which varies in length from hours to days. Our system must therefore behave more like a classifier than a prediction system; deciding whether a gas loss is currently occurring or not (so that action can be taken to avoid the loss increasing), rather than predicting that a gas loss will occur at a given time.
It is also worth noting that when generating training data patterns we were unable to distinguish between fast and slow gas losses because this is not recorded in the engineering logs.
2.3. GENERATION OF TRAINING DATA
Training data was generated using the engineering and alarm data sets. These training data correspond to all recorded occurrences of gas loss at monitored stores for a two year period, amounting to 240 patterns in all. Our classifier, in order to be correctly trained, also needs a set of training patterns corresponding to sites which are operating normally (or have problems not relating to gas-loss). The ‘normal operation’ data were generated in a similar way, using the alarm data. Identically structured vectors of n samples were created for randomly selected sites. These vectors end at randomly selected dates and times. The dates and times used for these training patterns were generated according to two important constraints:
· The date/time selected must be within the period chosen for examination
· The corresponding alarm totals vector must not overlap any recorded gas loss event at the site
Using this scheme we generated an additional 256 training data patterns which we expect not to correspond to gas leaks in stores. This gives us a total of 496 training data patterns. The desired output of the neural network is a single Boolean value where 1 indicates gas loss and 0 indicates no gas loss. Thus we have 240 training patterns for which we desire an output of 1 and 256 patterns for which we desire an output of 0.
To aid the neural network’s training, we scaled the input vectors by multiplying each alarm total in a training pattern by 0.1, so an input value corresponding to 10 alarms is presented to our classifier as 1.
Due to the extremely small quantity of training data available to us it was necessary to generate test and training data partitions using the 0.632 bootstrap method [3]. We sample an n length dataset n times at random with replacement to create the training data set and used the remaining, unselected patterns for test data. This gives us a training data set which is, on average, 63.2% the size of the original data set. Accuracies on training data are quite optimistic while, conversely, test data accuracies are rather pessimistic. To counteract this we calculate the overall error value thus [3]:
E = (0.368 * Etest) + (0.632 * Etrain)
To compensate for any atypical results that may be generated due to a particular partitioning, we generated 5 differently partitioned sets of training and test data from our original data set. Training runs are then performed on these data sets for a specified number of generations and the mean error rate calculated (see section 4).
3. EVOLVING NEURAL NETWORKS
Using evolutionary algorithms (EAs) to train neural networks is a mature research topic. Yao [4] provides a watershed review, while recent interesting practical examples are reported in [5] and [6]. It is generally found that EAs (or combinations of EAs and backpropogation) achieve better results than backpropogation [7] alone [2,4,5,6]. The system we use to evolve neural networks is quite similar algorithmically to EP-Net [8, 9], although we currently only evolve weights against a fixed topology.
3.1. NETWORK REPRESENTATION
The neural network representation used is based around a connection matrix and a weight vector. These two simple data structures are capable of representing networks with high levels of complexity (including recurrent and partially recurrent networks, although these are not investigated here). The simple network shown in figure 1 is used as an example to illustrate our encoding.
Figure 1: A simple neural network model with 6 neurons (N0 to N5) and 8 weights (W0 to W7)
We use four different types of neuron in our model. Inputs are simple placeholders for values to be input. Outputs are similarly placeholders and have no activation function, and can receive only one incoming connection, the weight value of which is set to 1. Bias neurons have a constant output value of 1 and cannot accept incoming connections. Finally, sigmoid (or hidden) neurons are standard neurons with a sigmoid activation function.
0 / 1 / 2 / 3 / 4 / 50 / x / x / x / x / x / x
1 / x / x / x / x / x / x
2 / x / x / x / x / x / x
3 / 0 / 1 / 2 / x / x / x
4 / 3 / 4 / 5 / 6 / x / x
5 / x / x / x / x / 7 / x
Table 1: An example connection matrix.