Security scale loss prevention strategy applied to self-service cash register
Technical Field
The invention provides a weighing self-learning algorithm, which is particularly applied to a loss prevention mode of a supermarket self-service cash register system, and belongs to the field of data structures and algorithms.
Background
With the development of intelligent retail, self-service cash registers, namely equipment for realizing self-service shopping in supermarkets and shopping malls, are gradually introduced into supermarkets. The equipment can enable a customer to scan commodities and settle accounts for payment in a self-service mode, a queuing process is omitted, and customer experience is improved. For a supermarket, the one-to-one binding relationship between a cashier and cash is released, so that the labor is saved, and the cost of the cashier is saved. In the self-service cash registering system, the loss prevention function is paid attention to, and when customer misoperation or omission commodity, self-service cash registering equipment can automatic identification to in time make warning and operation suggestion, consequently need provide an effectual loss prevention scheme in self-service cash registering machine.
Disclosure of Invention
The invention provides a loss prevention strategy for a safety scale. And sensing the weight of the product by using the self-service weighing platform, and calculating the self-learning weight of the corresponding commodity by cumulatively recording the weight of each commodity, wherein the self-learning weight is used as a reference for subsequent weighing and loss prevention.
The invention supports three weighing comparison modes:
1. not comparing the weight of the goods: i.e. no damage prevention setting is made.
2. Comparing the weight change of the goods: once the customer has brushed the commodity two-dimensional code, only detect weighing platform and whether have weight change. And if the change is changed, the damage prevention check is passed, otherwise, the damage prevention check is not passed.
3. Comparing the weight of the commodity with the self-learning weight: a set of sample data weights of the same commodity is taken, and a weight self-learning algorithm (combined with standard deviation and range) is used for calculating the self-learning weight of the commodity. And comparing whether the weight of the commodity is correct or not by using a weight comparison algorithm. If the return is correct, the damage prevention check passes, otherwise, the check does not pass.
The self-service checkout flow of the supermarket is shown in figure 1.
See figure 2 for the loss prevention inspection flow.
Drawings
Fig. 1 shows a self-service cash-collecting flow in a supermarket.
Fig. 2 loss prevention inspection flow.
Detailed Description
1. Initializing a loss prevention mode: a comparison mode is set, and commodity weight is not compared, commodity weight change is compared, and commodity weight and self-learning weight are compared.
2. When comparing the weight of the goods with the self-learning weight is selected, each good has 4 states:
1) ultra-light goods: since the electronic scale weighs 50g at the minimum, less than 50g is an ultra-light commodity, and weight learning and comparison are not performed.
2) Overweight goods: since the electronic scale weighs 30kg at maximum, over 30kg is an overweight commodity, and weight learning and comparison are not performed.
3) In weight learning: when the damage prevention inspection is carried out, data when the same commodity is weighed for the first 10 times is used as sample data, and the state of the commodity is in weight learning.
4) Weight learning is completed: and calculating the self-learning weight of the commodity by using a weight self-learning algorithm according to the commodity sample data.
3. The weight self-learning algorithm is a weight loss prevention algorithm which combines standard deviation and range difference and is suitable for self-service cash-receiving scenes of supermarkets. Because the electronic scale weighs and peels off and has certain error, the same commodity weighs differently each time, in order to be close to the actual weight as much as possible, the maximum value and the minimum value of 10 sample data are removed by using a range algorithm, then the standard deviation of the remaining 8 sample data is calculated, and the formula is as follows:
where N is 8 and μ is the average number of samples. And if the standard deviation sigma is larger than 5, the current sample data has high dispersion, and the weighed weight of the electronic scale is unstable. At this time, 8 sample data after the maximum value and the minimum value are removed and the latest two sample data are used as the latest 10 sample data, the standard deviation is calculated by using the weight self-learning algorithm again, the self-learning is finished by analogy until the standard deviation is less than 5, and the average number of the current sample is used as the self-learning weight of the commodity. The commodity state is weight learning completion.
4. After the commodity weight self-learning is completed, calculating the weighing weight of each commodity, comparing the weighing weight with the self-learned weight, considering the offset when weighing each time, and obtaining a comparison weight formula as follows:
when the self-learning weight of the commodity is less than 100 g:
self-learned weight x 95% < actual weighing of the good < self-learned weight x 105%
When the self-learning weight of the commodity is more than 100 g:
self-learned weight x 90% < actual weighing of the commodity < self-learned weight x 110%.