Patent Application
20230339637
This application claims priority of European Patent Application No. 22382369.1, filed Apr. 20, 2022, the contents of which is incorporated herein by reference.
The invention belongs to the field of processing tuna and other pelagic species, specifically the processing and packaging or canning of these canned species.
In the industrial processing of tuna and other pelagic species, the use of canning equipment is widely used. There are different types of machines for packing or canning tuna and pelagic species, where operators manually feed the machine with blocks of tuna (in the form of loins, pieces and/or crumbs). Then the team compacts the blocks, cuts them into portions and inserts the portions into the cans, at production rates of between 200 and 500 cans per minute.
In the state of the art there is a plurality of disclosures directed to equipment and machines related to canning fish and specifically canning tuna and pelagic species, among which is the patent document ES 2119608 B1 that teaches a movement system actuated by toothed wheels and a motor-reducer with a frequency converter.
On the other hand, patent document ES 2351318 B1 discloses a machine that uses blocks of tuna, automating the feeding of raw material to the canning machine.
Document EP2204324 A1 discloses a machine that generates tuna advance with a conveyor band, and with rotating compaction chambers, equipped with a load cell to measure the compaction force.
Currently existing canning equipment and machines have the drawback that they do not have the ability to adapt the operating parameters of the machine precisely to the particular conditions of each production of the machine (e.g. raw material used, can format, covering liquid used, etc.), which may vary during the production itself, but instead follow an empirical trial and error operation, where an operator periodically adjusts machine parameters (e.g. change target fill weight) according to his experience. This subjective adjustment of the operating parameters of the machine is an inaccurate adjustment that produces variations in the drained weight of the cans, generating weights below what is defined, with the consequent rejection, or more usually, weight above what is defined to avoid non-compliance with the minimum requirements, with the consequent economic loss. In addition to the weights, the imprecise adjustment of the packing machine's operating parameters can also affect the quality of the canned product, and in particular the height (maximum, minimum and average) of the tuna cake in the can, characteristics that may affect the appearance of the product and that, in unbalanced conditions, can lead to product rejection.
In the state of the art, there is therefore, a need to implement a tuna canning system that has the ability to adjust the operating parameters of the machine objectively, precisely and adaptively to the particular and changing conditions of each production, with the aim of purpose of optimizing production.
This invention refers to a system and an adaptive method for canning tuna that solves the problems of the state of the art previously listed, allowing to adjust the operating parameters of a tuna canning machine in a precise, adaptive and optimized way to the particular conditions for each production. By means of these adjustments it is possible to optimize the production; for example, the gain of the production process can be improved (i.e. percentage of weight obtained at the end of the process with respect to the weight of raw material that enters the packing machine), the precision in the filling weight of the cans and/or the quality or appearance of the product.
The adaptive tuna canning system of the present invention comprises a tuna canning machine (or tuna packer) and an optimization unit, which is responsible for optimizing the machine settings for each production. The tuna canning machine comprises a control unit in charge of controlling the machine during the tuna canning process of a production.
The optimization unit comprises a data processing unit configured to collect historical data from past productions captured by one or more training tuna canning machines during the tuna canning process in past productions.
The data processing unit is configured to receive some input parameters of a new production of the tuna canning machine, establishing a target function for the new production, and calculating, from historical data of past productions, some parameters of optimal performance for new production by optimizing the target function. The tuna canning machine is configured with the optimal operating parameters and executes the canning of tuna in the new production using these parameters.
In one embodiment, the tuna canning machine comprises at least one sensor configured to measure one or more properties of the tuna during the canning process, such that the input parameters comprise said properties of the tuna. Advantageously, the system obtains optimal operating parameters of the tuna canning machine that adapt to the changing conditions of the raw material entering the packer during production. The variations produced in the properties of the input raw material can significantly affect the productivity of the machine, since these properties (such as humidity or temperature) influence the weight of the tuna and/or the compaction of the tuna performed by the machine before canning and weighing the can, thereby affecting the filling weight and the gain of the production line. Sudden changes and fluctuating situations in the properties of the tuna can produce significant deviations in the filling weights (thus increasing their standard deviation) that decrease the machine's productivity.
The tuna canning machine may comprise a temperature sensor configured to measure the temperature of the tuna during tuna canning. Preferably, the temperature sensor is located at the entrance of the packer, coming into contact with the blocks of tuna (e.g., a sensor located on the raw material conveyor band, and protruding or flush with the top edge of the band), to obtain continuous temperature measurements on the tuna as it enters the packer input conveyor.
Additionally, or alternatively, the tuna canning machine may comprise an NIR sensor configured to continuously measure one or more properties of the tuna, such as humidity, protein, fat and ash content of the tuna. To do this, this NIR sensor uses diode array technology, and is based on the analysis of the near infrared, through the spectroscopic technique that naturally uses the electromagnetic spectrum. Spectrometry is the measurement of the amount of energy an element absorbs as a function of wavelength. The NIR sensor is located, preferably also at the entrance of the packing machine, so that the near infrared light is directed at the tuna or raw material that enters the machine. The light is modified according to the composition of the tuna and the NIR sensor detects this modified light. The spectral modifications are converted into information on the composition of the tuna through specific calibrations for each type of product. It is a technique used in other sectors (meat, fruit and vegetable or dairy) that allows the detection of contaminants and foreign matter in food and that allows the measurement of various parameters such as humidity, fats, proteins, sugars, fiber, etc.
In this invention, the NIR sensor is configured to measure some parameters that have been identified as relevant for the performance of the tuna canning process, since they can directly affect the absorption and gain obtained at the end of the process, thus influencing the determination of the optimal adjustment of the operating parameters of the machine. Notably, the following parameters have been identified: humidity, fats, proteins, and ashes. The NIR sensor is preferably configured to measure any combination of these parameters (e.g., only humidity, humidity, and fats, etc.). Any commercial NIR sensor can be used, on which a specific calibration is performed to be used on tuna.
The use of the properties of the tuna measured, preferably continuously (with a short sampling time), by means of the NIR sensor or the temperature sensor during production allows the machine to be optimally adjusted, since it takes into account truly relevant properties that affect the productivity of the packer. In addition, this property is usually very changeable, since the tuna introduced into the machine does not always have to be the same (type of species, type of packaging, type of processing, fishing area, etc.) nor is it introduced into the packer under the same conditions of temperature or humidity, for example.
The tuna canning machine may comprise a vibration sensor, configured to detect vibrations in some part of the machine, which may be associated, according to a vibration pattern, with unwanted operations of the machine. Using information captured by the vibration sensor, the machine can be configured to obtain a vibration pattern that corresponds to what is obtained when the machine is at peak performance.
A second aspect of the present invention relates to an adaptive method for canning tuna comprising the following steps:
This invention advantageously allows the machine to be adjusted with the optimal parameters calculated according to the characteristics of each production, either at the beginning of it (e.g., once before starting production or just started) or dynamically over manufacturing time, to adapt continuously and repeatedly to changing production conditions, such as the temperature and humidity properties of tuna. Thus, when the packer control unit detects a change in any of the production input parameters, such as a significant alteration (e.g. above a certain threshold) in the temperature or humidity of the tuna, the control unit can send the new production input parameters to the optimization unit, including the parameter value (e.g., temperature, humidity, etc.) altered, so that it calculates the optimal operating parameters, and sends them to the control unit for application, so that the tuna canning machine can maintain an optimal productivity throughout the production process.
The invention makes it possible to obtain optimal packer settings calculated by the optimization module. Optimum operating parameters are preferably obtained through supervised machine learning that includes training using, at least partially, historical production data. For this, the data types of past productions acquired by the training tuna canning machines must be identical, or at least similar, to the data types that are acquired in the new production of the tuna canning machine. Thus, the input parameters of the new production must be a data type included, at least partially, in the production data history. Training tuna canning machines are preferably machines of the same type (model, manufacturer) as the tuna canning machine on which the new production is carried out, which can be part of the set of training tuna canning machines.
The settings calculated by the optimization module are optimal operating parameters that are applied to the tuna canning machine to optimize the new production. Depending on the established target function, one or another aspect of the production process will be improved. In one embodiment, the target function is based on maximizing the number of cans produced per mass unit of input tuna. The target function defined in this way improves the gain and accuracy of the filling weights, which directly influences the productivity of the line. Other objective functions can be used, such as maximizing gain, minimizing the standard deviation of fill weights, maximizing the quality of product appearance, or a combination of several factors applying a relevance factor (i.e., weights) to each of them.
The application of the optimal settings on the machine can be done manually by an operator and/or automatically by the packer's own control unit. For example, there may be a combination of manual and automatic adjustments, where for some machine adjustments may be necessary, due to their physical nature (e.g., change of nozzle used in the packer), manual change to be made, while other adjustments can be implemented automatically by the control unit (e.g., automatic change of target packing weight).
Next, a series of drawings that help to better understand the invention, and that are expressly related to an embodiment of said invention, which is presented as a non-limiting example of it, are very briefly described.
The tuna canning machine 2 includes a control unit 10, implemented for example, by means of a computer in communication with a PLC, in charge of controlling the machine (including, e.g. controlling of parameters or configurable settings of the machine) during the tuna canning process of a production.
The optimization unit 3 comprises a data processing unit 4, which is configured to:
Once the optimal operating parameters 9 have been obtained, the optimization unit 3 sends them to the tuna canning machine 2 for their application in the new production. The tuna canning machine 2 is configured, either automatically and/or manually, to execute the canning of tuna in the new production based on the optimal operating parameters 9 received.
The at least one training tuna canning machine 7 may include the tuna canning machine 2.
In
The optimal operating parameters 9 obtained by the optimization unit 3 are applied in the new production of the tuna canning machine 2. There are different ways for this. For example, the optimization unit 3 can send these parameters to the control unit 10 (as shown in
Optimum operating parameters 9 may include machine parameters or settings that are manually made by a machine operator (due to, e.g., a physical change to be made to the tuna canning machine 2). In this case, the calculated optimal operating parameters 9 can be represented on a screen 14, so that an operator can configure the tuna canning machine 2 according to said parameters.
The optimization unit 3 is preferably configured to calculate the optimal operating parameters 9 through supervised machine learning that includes a training stage using the production data history 5.
The tuna canning machine 2 preferably comprises at least one sensor 11 configured to measure one or more properties of the tuna 12 during the canning process, where the properties of the tuna 12 form part of the input parameters 8 of the new production.
The at least one sensor 11 may comprise a temperature sensor configured to measure the temperature of the tuna entering the machine during the canning process and/or a NIR sensor (near-infrared spectral region) configured to measure, by means of spectroscopy techniques of the near infrared, at least one tuna property. In one embodiment, the NIR sensor is configured to measure at least one of the following properties of the tuna: humidity, protein, fat and ash content of the tuna.
Adaptive tuna canning system 1 can be applied to any type of tuna canning machine 2.
This invention also refers to an adaptive tuna canning method, as represented in the flow chart of
Method 100 may comprise measuring 170 one or more properties of tuna 12 during the canning process in tuna canning machine 2; where the input parameters 8 of the new production comprise the properties of the tuna 12 measures. The properties of the tuna 12 measured may comprise, e.g., the temperature, humidity, protein, fat and/or ash content of the tuna. Measurements of the properties of tuna 12 can be made only once, e.g., before starting new production or during new production.
Alternatively, the measurements can be performed repetitively during the canning of tuna in the new production. In this case, as shown with dashed lines in the flow diagram of
In one embodiment, the input weigher 21 is an automatic weighing system installed in a section of the conveyor before reaching the packer, with the function of weighing the tuna continuously. In another embodiment, the input weigher 21 is a scale configured to obtain punctual measurements of input tuna. The scale is used if the tuna is received in batches (e.g., boxes, bags, carts), where an operator located at the entrance of the machine must take each of these sets, place it on the scale and, once that weight has been recorded, deposit the tuna, emptying the container into a fish receiving tray that is connected to the packer's inlet conveyors. In this case, the tuna container element must be previously used to tare the scale, so that weight will be subtracted in each new weighing and the weight of the container itself will not be collected as part of the raw material.
The input tuna weight is used to later calculate the diminishment generated during packaging, that is, the amount of tuna mass that is lost in the process. This loss is one of the fundamental parameters to know to evaluate the performance of a line or factory, in order to take into account, the liquid that contained the tuna and that is lost in the compacting process, or the small crumbs and pieces that escape at different points of the machine (due to gaps, friction, etc.). For the calculation of the losses to be correct, all the product that is introduced into the machine must be weighed, as well as the weight of all the manufactured cans obtained by means of the dynamic weigher 24. To calculate the loss, the total net weight of the canned tuna is subtracted from the total weight of the tuna put into the machine. Therefore, when the target function is related to gain and filling weight accuracy, the input weigher 21 and the dynamic weigher 24 are important to obtain the optimal operating parameters 9.
The cans 29 accepted by the dynamic weigher 24 (i.e., those that have acceptable weights) are transported to the following stages of the production line until the final product 33 is obtained, with the added covering liquid, and the can duly closed and sterilized. Subsequently, the draining process 34 is carried out in the laboratory of a certain number of cans produced, randomly selected from a production, obtaining an average drained weight, which is defined as the arithmetic mean of the weights obtained after the process of draining the covering liquid. Knowledge of the drained weight and packing weight of a can (or the average packing weight of the cans produced in that production), together with the machine and raw material conditions, and settings at the time of manufacture, allows obtaining relevant production parameters, such as absorption and gain.
By providing the control unit 10 with access to Internet 32, the optimization unit 3 can collect the data collected during production. Subsequently, in the server 30 the data is structured, stored and processed to prescribe the optimal settings of the machine (i.e., optimal operating parameters 9). The optimization unit 3, in addition to providing relevant information about the process, has the ability to discover which are those variables that influence the final result of the production and, according to given input conditions, recommending to the user how to adjust them so that the objectives set according to a target function 13 (e.g. weight gain and precision objectives) are the best possible, based on the behavior obtained in past productions, previously received and stored in memory (e.g. in a database 31), which corresponds to a production data history 5. As explained above, said production data history 5 may correspond to past production data 6 of the same tuna canning machine 2, or other similar tuna canning machines located in different factories or locations.
The optimization unit 3 has the objective of prescribing a series of optimal parameters, providing recommendations of the machine parameters on which it is possible to act, and that, based on past production histories, they are identified as optimal for specific input conditions. To do this, optimization unit 3 performs the following steps:
In the tuna canning process, different critical variables can be considered for the production process. In one embodiment, the following parameters related to weight and gain are considered:
The following critical parameters related to the performance of the production line can also be considered:
Once the critical variables have been identified, the desired final objectives are defined. The final objective is related to maximizing the profitability of the production line. Depending on the critical variables considered, different final objectives can be expressed:
The first objective is focused directly on the final product, and on the optimization of the raw material. This objective, related to gain, implicitly implies the optimization of diminishment and absorption. Accuracy in weight translates into optimizing the standard deviation of the weights, trying to obtain the minimum possible value. The lower this deviation, the lower the target packaging weight may also be, reducing the margin that must be maintained to meet the established minimums. The second objective focuses on other aspects of the production line, related to the use of the machine, minimization of breakdowns, preventive maintenance, etc.
Once the critical parameters, and the variables to be optimized to meet the defined objectives have been defined, they are grouped into a single target function that encompasses them, and which will be the one that the algorithm will try to optimize.
Measured gain as the ratio between the final drained weight, and the weight at the entrance to the packer, the standard deviation of the machine does not influence said value, since with a greater standard deviation the accumulated weight of tuna at the entrance will increase, but also the accumulated weight of tuna after draining. Considering that you want to produce a certain number of cans, if the machine has a higher standard deviation, both the input weight, and the output weight will increase to produce the same number of cans (as shown in the chart in
However, if instead of considering output weight, the number of cans manufactured is counted, both gain and precision in weight can be grouped together, being able to take as a final objective to optimize the number of cans produced per mass unit of input tuna in the packer (e.g. number of cans produced per kilogram of input tuna), since this variable is affected both by the standard deviation of the weight (accuracy in weight), and by the gain. In particular, as illustrated in
On the other hand, in addition to achieving the objectives of weight gain and precision, the appearance of the product can also be considered, either the appearance when it comes out of the package and/or the final appearance obtained. The quality of the required appearance can be a more or less a relevant factor. In the event that appearance is not considered an important factor, accuracy in weight and final gain can be prioritized first. In cases where appearance is key, it can also be included in the final objective, although to achieve this it is necessary in many cases to increase the packaging weight.
With regard to the appearance of the final packaging, this can be objectively assessed by means of an artificial vision system, based on the elements detected in the image (residues of skin, sangacho, thorns, etc.) and the height of the tuna cake, and its distribution inside the can.
The final appearance of the product can be evaluated manually, although an artificial vision system could alternatively be used to objectify the measurement. The final appearance is defined during the draining process, wherein a certain number of cans are opened, the user observes their appearance, and manually enters the evaluation they consider into the system. In order to quantify the final appearance, a scale can be defined, e.g., a scale of five values in which value 1 represents an unacceptable or poor appearance, and value 5 represents an optimal appearance.
When it comes to maximizing the number of cans produced per mass unit of input tuna (that is, maximizing gain and weight accuracy), the appearance of the product can also be taken into account (be it the appearance of the final packaging, the final appearance or a combination of both). In that case, it has to be established how much importance should be given to each of the two factors. Two options can be considered:
If the final appearance is applied, in both cases it will be essential that during the registration of drained weights in the system by the user, the appearance obtained in the open cans is also registered, by means of the defined scale, in order to later have information on whether the system is meeting the desired objectives with the machine settings applied in each production.
In the second option, the target function is made up of two terms, the one that represents the number of cans manufactured per mass unit of input tuna (e.g. for each Kg of input tuna), and the appearance. The relevance factors that must be applied to each term (which we can call the gain factor and the appearance factor) can be considered to be two percentages, and both must add up to 100%. Thus, for example, if appearance is of little or very little importance for a given product, prioritizing maximizing the number of cans, an importance percentage of 10% or 20% could be set for the appearance factor and 90% or 80%, respectively, for the gain factor. These percentages must be set by the user for each product to be manufactured and are added as two new parameters to be entered manually at the beginning of a production.
On the other hand, in addition to establishing the importance of each term, these can also be converted to relative values between 0 and 1. In this way, the maximum function can be determined, with a value equal to 1, at which the optimization algorithm has to get as close as possible, being able to compare the results obtained in different productions. For example, the number of cans manufactured per Kg of tuna will be higher the lower the target final weight, without meaning that the gain will be higher. Obtaining the relative values, it is possible to have an idea of the profitability obtained in each case, and different productions can be compared, regardless of the rest of the conditions.
To transform the gain value, an ideal maximum gain can be calculated, considering perfect conditions. These conditions are set, for example, as standard deviation=0, diminishment=0%, ideal absorption=130% (a value higher than the usual values is set).
Under these conditions, a target packing weight can be established:
Ideal packing weight=(drained weight/ideal absorption)*100
In addition, the fact that the deviation is zero implies that the machine always packs the exact weight of tuna in each can, so that the input weight of tuna for a can would be equal to the packing weight. Thus, the maximum number of cans that could be manufactured under ideal conditions for each Kg of tuna would be:
Number of cans=1000 gr/Ideal packing weight, wherein:
Ideal packing weight=target drained weight*1.30 (ideal absortion)
As an example, for a required drained weight of 40 grams, the target pack weight would be (40/1.30)=30.77 grams, and the maximum number of cans that can be manufactured with 1 Kg of tuna at the inlet of the packer, considering null diminishment, it would be (1000/30.77)=32.5 cans.
Thus, the first factor of the target function is the following, where a is the relevance factor assigned by the user to the gain factor:
α*(number of cans per Kg/maximum number of cans per kg)
Regarding the second term of the target function, it is necessary to transform the appearance value to a factor between 0 and 1, simply dividing the appearance value obtained by the maximum possible, which according to the scale defined in the previous example will be 5. Therefore, the second factor of the target function is:
γ*(obtained appearance/maximum appearance)
Thus, putting both parts together, the target function can be expressed with the following formula:
Wherein:
The higher the value calculated for the target function, the higher the gain obtained in the corresponding production.
Next, two different algorithms are detailed that can be used for the automatic identification of the optimal settings (i.e., identification of the optimal operating parameters 9): an algorithm based on iterative searches and an algorithm based on machine learning models.
With respect to the algorithm based on iterative searches, whether the appearance is initially set as the minimum requirement that the product to be manufactured must meet, or if its measurement is included in the target function to be maximized, the process to find the recommended settings is similar in both cases and is explained below.
The identification and compilation of all the parameters that are part of the process have already been carried out previously:
Although the target packing weight could be considered at first as a production parameter, it can also be considered as a modifiable setting that can be acted upon to increase gain, in the event that tuna canning machine 2 implements an automatic adjustment (for example, by self-adjusting the advance of the machine), so that the average pack weight of all accepted cans 29 approaches the target pack weight indicated by the user.
For each registered production, all the related variables are stored, both those that are collected directly from the plant, and those that are calculated later, also including the result of the target function as one more variable. With these data, a history is formed in the form of a table or matrix, wherein each row represents a production, and each column represents a different variable. As 5 the number of registered productions increases, the table grows, adding new data from past productions 6, already made with a given tuna canning machine 2. The more information there is, and the greater variability in the values that the considered parameters take, the better the optimal settings can be prescribed to optimize the gain and appearance parameters.
Once enough information has been collected in the production data history 5, the algorithm can make prescriptions about the optimal settings that can be modified to maximize the final objective. To do this, as illustrated in the example in
Regarding the algorithm based on machine learning models, machine learning techniques are applied for the prediction and prescription of parameters. To use this type of algorithm, it is necessary to have a sufficient history, that is, there must be a large number of registered productions, with some variability in the parameters collected. Optionally, to obtain a greater quantity and variability of the data, these can be collected from multiple tuna canning machines, located for example, in different factories, cities, countries, etc.
In the first place, the prediction system of the final target value is developed, given a series of input parameters, which will include both the parameters that define the production, the properties of the raw material, and certain machine settings. This is a regression problem, since the expected output is a continuous variable, which is itself a subfield of supervised machine learning. There are different techniques to solve this type of problem, among which linear regression algorithms, logistic regression, decision trees and deep learning algorithms stand out.
Regardless of the algorithm applied, the process is similar, and is mainly made up of a training stage, a validation stage, and a test stage. The training stage is where most of the available historical data is used. The algorithm is given a multitude of records, with their corresponding inputs and the expected output for each of them, so that the machine learning the algorithm can learn based on historical data and identify patterns that would be impossible to extract manually. It is important that the variability of the data is high since, if not, the result could be biased to a specific case, making the system unable to generalize when the input parameters change, providing incorrect outputs.
In the validation stage, a set of production records is used, of which the expected output is known, and which have not been used to train the algorithm. For each of them, the output returned by the already trained model is obtained, and the results are analyzed in order to validate the system, adjust the model parameters, etc. Finally, in the test stage, the model is executed directly with new data, which the algorithm has not previously used, and whose output is not previously known.
The system works in such a way that, for a given set of conditions (input parameters 8 defining a new output), the already trained machine learning model 73 is executed using these parameters as input to the algorithm (inference), performing a random generation 71 of possible settings to establish a set of combinations of settings 72 that include possible values that the machine settings can take, obtaining a prediction of the value of the target function, target value 74, for each combination of settings 72, as represented in the chart of
A user modifiable parameter that is very relevant to the gain obtained at the end of production is the target pack weight, which is the target weight set in the dynamic weigher and/or in the control unit 10 for accepted cans 29. This variable is normally entered manually by the user, estimating it according to the minimum drained weight that the final product must meet, and according to values also estimated for packaging diminishments and coverage liquid absorption. According to an embodiment of this invention, the selection of this value can be advantageously automated, as well as the selection of the rest of the modifiable configurations 54 of the machine. Following the same approach described, the optimization algorithm can prescribe the optimal target packaging weight, which maximizes the profitability of a particular production. The optimal target packing weight calculated by the optimization unit 3 can be sent to the tuna canning machine 2 for manual application by an operator or, alternatively, it can be received and applied in an automated way in the control unit 10.
When a sufficiently large number of data have been collected in the production data history 5, a total self-adjustment of the machine can be reached, without the need for user intervention. To ensure optimal automated operation, it is advisable to have a large number of productions with different input parameters, settings, and even different machines and factories.
The adaptive tuna canning system of this invention can be applied to any tuna canning machine.
The temperature sensor 22 comes into contact with the tuna cakes to measure the temperature when the tuna cakes have already been compacted by a density control system 85, implemented by means of a group of conveyors that exert pressure on the tuna cakes. The density control system 85 can be optionally assisted by a rammer 88, which is responsible for controlling and homogenizing the density of the tuna, trying to distribute the tuna on the conveyor bands in a more homogeneous way to reduce the possible gaps that may appear.
The NIR sensor 26 is responsible for measuring the degree of humidity, fat, protein and/or ash content of the raw material 20 that enters the packer through a feeding conveyor system 84. The vibration sensor 83, located by example in the indexer-reducer assembly 87 of the packer, it can be used to detect undesired operations of the machine according to a vibration pattern.
An advance conveyor system 90 is responsible for moving the tuna cakes, already compacted by a density control system 85, towards a nozzle 91 through an intermittent start-stop movement, which generates an intermittent advance, controlled, for example, by a servomotor. Next, the nozzle 91 is responsible for shaping the tuna cakes with a certain shape that adapts to the empty cans 80, and a pusher 92 introduces the tuna portions into the empty cans according to an adjustable pressure and path. The base of the pusher (the part of the pusher that contacts the portion or tuna cake) can have perforations through which a blowing system of the machine can inject air and/or water vapor at the beginning of the movement of recoil of the pusher, in order to prevent the portion of tuna from sticking to the pusher 92 when it starts to recoil.
The tuna canning machine 2 shown in
The artificial vision system 93 may also comprise a lighting system and a color camera (e.g., linear RGB camera) configured to capture color images of the canned tuna portions as they progress through the output conveyor system 86. The artificial vision system 93 is preferably configured to detect defects (e.g., detection of sangacho, dark areas, foreign objects, holes in the tuna, or the presence of raw material in the sealing area) in the canned tuna portions by analyzing the images captured by the color camera and/or by analyzing the height measurements made by the laser profile sensor.
The right column of
In the example in
As described above, through the optimization of a target function 13, the optimization unit 3 can calculate and prescribe optimal operating parameters 9 for the new production. The various parameters employed (raw material parameters 102, final product parameters 104, machine parameters 106) by the adaptive tuna canning system 1 may vary and will depend on the particular embodiment. Likewise, the chosen target function and the optimal operating parameters 9 to be optimized, can also be selected in an initial configuration of the system. For example, in one embodiment it can be determined to optimize only the target packaging weight, keeping the rest of the configurable parameters unchanged.
| Number | Date | Country | Kind |
|---|---|---|---|
| 22382369.1 | Apr 2022 | EP | regional |
