JP6053782B2 - Combination weighing device - Google Patents

Description

  The present invention relates to a combination weighing device that combines a plurality of weighing values input from a plurality of weighing machines, selects an optimum combination of weighing machines, and discharges articles from the selected weighing machine.

  This kind of combination weighing device is always required to have high accuracy and at the same time high availability. High accuracy reduces the loss of articles weighed in combination, and high availability improves productivity. For this reason, in the past, inventions as disclosed in the following patent documents have been proposed with the aim of improving both accuracy (yield) and availability.

  However, with these devices, the closest combination to the target value was selected as the optimal combination for each weighing cycle, so even if the best accuracy for each cycle, the average accuracy and operation after a long time operation In terms of rate, it was not always the best.

  In general, the individual weighing hoppers used in a combination weighing device, when selected in combination, discharge the weighed article, and then a new article is fed into the next weighing cycle. While such a cycle is repeated in the individual weighing hoppers, the availability of the device is improved. However, in some weighing hoppers, if such a cycle is delayed, articles easily adhere to the weighing hopper, and even if they are selected in combination, it becomes a factor affecting the weighing accuracy. Moreover, since the probability that such a staying hopper article is selected is relatively small, it also affects the weighing accuracy.

  On the other hand, if there are multiple combinations where the combined total value falls within the set upper and lower limits (hereinafter referred to as the allowable range), many combinations that remain within the allowable range remain in the next weighing cycle. If you leave it, it will lead to an improvement in the operating rate. Furthermore, it may be preferable to select the second or third combination within the allowable range in order to improve accuracy and operating rate.

  However, since the conventional apparatus has not been selected in consideration of the future, there has been a problem that the average accuracy and operating rate after long-time operation are not always the best.

  An object of the present invention is to solve such a problem, and an object thereof is to provide a new combination weighing device capable of improving both the average accuracy and the operation rate after long-time operation.

  A combination weighing device according to the present invention is a combination weighing device that selects an optimum combination of weighing machines by combining weighing values obtained by a plurality of weighing machines for weighing articles, and each of the combinations is selected. When preparing a plurality of evaluation functions for evaluating the impact on accuracy and the impact on operating rate, and combining each measurement value, for each combination, calculate an evaluation value based on the evaluation function, A combination that maximizes the sum of the calculated evaluation values is selected as an optimal combination.

  Each of the evaluation functions includes a weighting coefficient that represents the degree of relevance of each evaluation function, and the weighting coefficient is determined by a computer that executes a genetic algorithm.

  The combination weighing device is provided with an operation unit. From the operation unit, it is possible to arbitrarily set whether to perform an operation focusing on accuracy or an operation focusing on the operation rate.

  In general, in combination weighing devices, when the measurement accuracy is improved, the operation rate is deteriorated, and when the operation rate is improved, the measurement accuracy is deteriorated. Thus, there is a contradictory relationship between accuracy and availability. There is a multi-objective optimization method as a method for improving such a relationship (trade-off relationship) at the same time.

  In the present invention, using this multi-objective optimization method, in order to simultaneously improve the accuracy and the operating rate that are in a trade-off relationship, the effect on the accuracy and the operating rate when each combination is selected are affected. The influence is expressed by a mathematical expression (evaluation function).

  According to the present invention, it is possible to improve both the average accuracy and the operating rate which are in a mutually contradictory relationship after long-time operation. Further, an operation with an emphasis on accuracy and an operation with an emphasis on the operation rate can be performed by an operation from the operation unit.

The figure explaining a Pareto optimal solution. Schematic of the principal part of the combination weighing device according to one embodiment of the present invention. 1 is a configuration block diagram of a combination weighing device according to an embodiment of the present invention.

  In the present invention, the following six evaluation functions are defined as an example.

One is to evaluate the influence of each combination on the accuracy and is defined by the following equation (1). The evaluation function of the equation (1) is a linear function that becomes 1 if the combined total value is equal to the target value, and that gradually decreases as the distance from the target value increases. However, the deviation from the target value is e _W , the upper limit value of the deviation is e _u , b ₁ = −1 / _eu , and b ₂ = 1.

  By this equation (1), the error of each combination total value is normalized within an allowable range.

  Next, the following four evaluation functions are defined in order to evaluate the degree of influence on the accuracy of each combination or the degree of influence on the operating rate.

1. 1. Evaluation function for stay count 2. Evaluation function regarding the number of selected weighing machines 3. Evaluation function for contribution degree Evaluation functions regarding the degree of dispersion Hereinafter, these evaluation functions will be sequentially described.

1: Evaluation function regarding stay count The combination weighing device supplies articles to the weighing hopper of each weighing machine and weighs them, and selects the optimum combination by combining the obtained weighing values. Further, the article is discharged from the weighing hopper of each selected weighing machine, and the article is supplied again to the discharged weighing hopper. Repeat such a series of cycles. Each weighing machine has its own counter. When the article has not been discharged, the count value of the counter is increased by one, and when the article is discharged, the counter value of the counter is reset to zero. When the count value of the counter is high, articles are staying in the weighing machine for many cycles. If it is left as it is, the number of combinations that fall within the allowable range decreases and the measurement accuracy deteriorates. In addition, when the number of weighing machines that retain articles increases, the number of combinations that fall within the allowable range is relatively reduced, and a combination failure is likely to occur. It affects accuracy and availability. Therefore, in order to positively discharge articles from a weighing machine with a high count value, an evaluation function of the following equation (2) is defined. However, and one count value in the weighing machine belonging to combination the largest of the C _M, the upper limit of the retention count value and C _L.

2: Evaluation function regarding the number of selected weighing machines The combination weighing device includes a plurality of weighing machines. In each combination, the smaller the number of weighing machines that fall within the combination allowable range, the more weighing machines that can be used in the next cycle. It mainly affects the utilization rate. Therefore, in order to make it easier to select a combination with a smaller number of weighing machines, an evaluation function of the following equation (3) is defined. However, the weighing machine number that falls within the allowable range and N _S, and the entire weighing machine number as the N _L.

3. Evaluation function regarding contribution When selecting an optimal combination from a plurality of combinations that fall within the allowable range, the operation rate is improved by leaving many combinations that fall within the allowable range even in the next cycle. Thus, for example, as a result of combining the measurement values of ten weighing machines, six combinations that fall within the allowable range are found as shown in Table 1. Here, the symbol “◯” represents a selected weighing machine that falls within the allowable range.

Then, the number of times belonging to any set is counted for each weighing machine, and the contribution of each weighing machine is defined by the following equation (4) based on the count value. Then, in this case, the contribution of each weighing machine is as shown in Table 1. Here, the weighing machine h th contribution and C _h, and the number of times that belong to n _h, the total weighing machine number and N _L.

  When the combination having the lowest total contribution is selected, the total contribution of the remaining weighing machines becomes the highest. Selecting such a combination will leave many combinations that fall within the allowable range in the next cycle, and as a result, the operating rate will be improved. Therefore, the following equation (5) is defined as an evaluation function for obtaining the total value of contributions of the remaining weighing machines.

  Table 1 shows the total value of the contributions of each set and the total value of the contributions of the remaining weighing machines, as shown in Table 2. Therefore, if only the case of Table 1 is seen without considering other evaluation functions, the set 4 is the optimum combination. Then, since the remaining weighing machines belonging to the set 6 are within the allowable range even in the next cycle, the operating rate is improved.

4: Evaluation function regarding degree of dispersion By the way, in the combination weighing device, it is known that the number of combinations that fall within the allowable range is more generated when the measurement values of each weighing machine vary. Therefore, in order to positively leave a weighing machine with a large degree of variation in the weighing values left unselected, the degree of dispersion of each remaining weighing machine is defined by the following equation (6). Here, the number of remaining weighing machines is n, the weighing value of the remaining weighing machines is h _i , and the average value of each weighing value h _i is μ.

Then, the following equation (7) is defined as an evaluation function in which the evaluation value increases as V increases. However, _{W T} is a combination target value.

  Each of the above evaluation functions is a combination of weighing values (weight), but some combination weighing devices perform a number combination. For example, when the sausage to be packed is to be weighed, the combination total value is within the allowable range and the number is the target number is discharged. The evaluation of such a number combination will be described next.

  For example, when a target number of articles is selected, the combination may not be established in any way depending on the number of articles put into each weighing machine. For example, when the target number is set to an odd number, if an even number is supplied to each weighing machine, a combination failure occurs. Here, this state is referred to as a tabu state, but when the state falls into this state, the operation rate decreases. Therefore, the margin of each combination until falling into such a taboo state is calculated. For example, Table 3 assumes the case where the number shown in Table 3 is input to each of eight weighing machines, and the target number is seven.

  In Table 3, if you select the combination of Units 3, 4, and 5 for the first time, the weighing machine that satisfies the target number is the combination 1 from the remaining weighing machines. There are two. However, since there are no more combinations beyond this, in the case of this selected route, it is possible to guarantee the combination of the target numbers up to two cycles ahead. Therefore, the margin of the selected route in this case is set to 2. Subsequently, it is the combination 3 that can be selected from the weighing machines except the No. 3, 4, and 5 units, and then the combination 4 can be selected. In this case as well, since up to two cycles are guaranteed, the margin is set to 2 in this case as well.

  Thus, the margin of each selected route is obtained for all combinations that satisfy the target number, and the one having the largest margin is defined as the margin T in the combination. For example, in the case of Table 3, the margin T in the combination of Nos. 3, 4, and 5 is 2. The following equation (8) is defined as an evaluation function in the case of the number combination using the margin T.

By defining the six evaluation functions in this manner, the influence on the accuracy and the influence on the operation rate are evaluated for each combination. However, since it is not known to what extent these evaluation functions are involved, whether the accuracy is maximized or the operating rate is maximized, the evaluation value V _{C of} each combination is obtained by the following equation (9). look, the evaluation value V _C is selected as the optimal combination largest one.

Here, a _i is a weighting factor and is set to an arbitrary value under the constraint condition of the following equation (10).

The coefficient a _i serves as an index indicating how much importance is given to which evaluation function. By adjusting these coefficients, it is possible to perform an operation focusing on accuracy and an operation focusing on the operating rate. However, since it is difficult for humans to determine each coefficient, a Pareto optimal solution is obtained using a computer that executes a genetic algorithm.

In general, a set of solutions that represent the limits of the trade-off relationship is called a “Pareto optimal solution”.
This Pareto optimal solution tries to improve any value (eg, accuracy) of the objective function.
It is a solution in which the value (operation rate) of another objective function is corrupted.

  FIG. 1 is a diagram for explaining the Pareto optimal solution. The solutions a, b, c, and d are Pareto optimal solutions, and the solutions e and f deviating from them are inferior solutions. The limit surface formed by the set of solutions a, b, c, and d is called a Pareto front. There are a plurality of Pareto optimal solutions, and it is possible to place importance on accuracy or on the operating rate depending on which solution is selected. For example, if the solution a is selected, the accuracy is improved, but the operating rate is lowered. If the solution d is selected, the accuracy is lowered, but the operating rate is improved.

Therefore, a genetic algorithm is used to obtain the Pareto optimal solution. The genetic algorithm is based on the process of natural evolution (chromosome selection, crossover and mutation) as a hint. This is an algorithm proposed by Holland. When using this algorithm, for example, the weighing value of each weighing machine obtained in 10,000 weighing cycles is recorded as sample data. Then, a solid (coefficient a _i ) is randomly generated to evaluate the fitness of each individual. In other words, 10,000 optimal combinations are obtained using equation (9), and the average accuracy and operating rate of the 10,000 pieces are obtained. And it leaves so that possibility that it will be selected as a solid with good fitness becomes high becomes high. That is, when changing individual (coefficient a _i), will leave the average accuracy and availability and well made towards individuals (coefficients a _i). Furthermore, to generate a new population (coefficient a _i) by the crossover and mutation, continue to evaluate the fitness of each solid (coefficient a _i). In this way, generational alternation by evaluation, selection, crossover, and mutation is repeated, and a Pareto optimal solution is obtained with a weighted weight coefficient a _i .

  Genetic algorithms include Fonseca et al. MOGA (Multiobjective Genetic Algorithm), Deb et al. NSGA-II (Non-Dominated Sorting Genetic Algorithm-II), Zitzler et al. SPEA2 (Strength Pareto Evolutionary Algorithm 2), and the like. In the present invention, NSGA-II is used, but it is not limited to this genetic algorithm.

  Hereinafter, a combination weighing device according to an embodiment of the present invention will be described with reference to the drawings.

  FIG. 2 is a schematic diagram of a main part of a combination weighing device according to an embodiment of the present invention. In this figure, the combination weighing device 100 includes a dispersion feeder DF, a plurality of radiation feeders RF, a plurality of pool hoppers PH, a plurality of weighing hoppers WH, and a collective chute CS. The dispersion feeder DF is disposed at the upper center of the apparatus. The plurality of radiation feeders RF are arranged radially around the dispersion feeder DF so as to surround the dispersion feeder DF. The plurality of pool hoppers PH are arranged in the lower stage of each radiation feeder RF. The plurality of weighing hoppers WH are arranged below the plurality of pool hoppers PH. The number of weighing hoppers WH and pool hoppers PH is the same. The collective chute CS is disposed below the plurality of weighing hoppers WH.

  The dispersion feeder DF disperses the article dropped thereon in the circumferential direction by the vibration of the electromagnetic feeder DV.

  The radiation feeder RF conveys the article conveyed from the dispersion feeder DF to the tip of the trough TR by the vibration of the electromagnetic feeder RV, and discharges the article to the lower pool hopper PH.

  The pool hopper PH temporarily stores articles discharged from the radiation feeder RF. When the lower hopper PH opens and closes, the pool hopper PH opens and closes the gate g based on a command from the control unit CU, and discharges the articles stored in the pool hopper PH to the lower weigh hopper WH.

  A weight sensor WS is attached to the weighing hopper WH. The weight detected by the weight sensor WS is input to the control unit CU and used for the combination calculation. Since each hopper PH, WH has a known configuration, a gate opening / closing mechanism, a support mechanism for the hopper PH, WH, and the like are omitted here.

As shown in FIG. 3, the control unit CU includes a CPU 10 and a ROM 11, a RAM 12, and a hard disk 13 that are controlled by the CPU 10. The CPU 10, the ROM 11, the RAM 12, the hard disk 13, and the like are connected to each other via a bus line such as an address bus or a data bus. Further, the control unit CU is connected via an interface 14 to the operation unit RU having a dispersion feeder DF, a radiation feeder RF, a pool hopper PH, a weighing hopper WH, and a touch panel function. Further, the operation unit RU is connected to the computer C that executes the genetic algorithm, and the above-described weighting factor a _i is updated.

Various programs are stored in the ROM 11. When the CPU 10 reads and executes various programs stored in the ROM 11, the calculation of the equations (1) to (8), the management of the stay count value, and the gate opening / closing control for the pool hopper PH and the weighing hopper WH are performed. Is called. In addition, the hard disk 13 stores evaluation functions of equations (1) to (8). The weighting factor a _i used is periodically updated by the computer C executing the genetic algorithm.

  When the CPU 10 receives a discharge request signal from a packaging machine (not shown) or starts with its own cycle timer, the CPU 10 inputs the measured values from the weight sensors WS of the respective weighing machines M1 to Mn and stores them in the RAM 12. Subsequently, calculations of formulas (1) to (9) are executed based on the stored measurement values. At that time, if there is a margin in the calculation time, the calculations of the above equations (1) to (9) are executed for all combinations. On the other hand, when the operation speed is increased and there is no allowance in the calculation time, a combination in which the combined total value falls within the allowable range is extracted first, and then, for each of the extracted combinations, the above formulas (1) to ( The calculation of 9) is executed. At this time, if it is a weight combination, the calculation of the expression (9) is executed by the linear sum of the remaining expressions excluding the expression (8) used in the number combination. If the number is a combination, the calculation of Expression (9) is executed with a linear sum including Expression (8).

  When the CPU 10 selects the largest evaluation value as the optimum combination by this calculation, a discharge command is transmitted to the selected weighing machines M1 to Mn, and then the discharge command is sent to the corresponding pool hopper PH with a slight delay. Send. Then, the weighing hoppers WH of the weighing machines M1 to Mn that have received the discharge command open and close to discharge the articles, and then the pool hopper PH opens and closes to supply the articles to the empty weighing hopper WH. Subsequently, when the pool hopper PH becomes empty, a drive command is transmitted to the radiation feeder RF and the dispersion feeder DF, and a new article is supplied from the corresponding radiation feeder RF to the empty pool hopper PH. Thus, the combination weighing of one cycle is completed, and then the next weighing cycle is started by the next discharge request signal or the start signal from the cycle timer.

On the other hand, the measured values of the weighing machines M1 to Mn in each cycle and the total weight of the optimum combination are recorded on the hard disk 13. The computer C periodically accesses the hard disk 13, takes in the recorded data, executes a genetic algorithm based on the data, and updates the weighting coefficient a _i while calculating the average accuracy and the operating rate. To go.

When the weighting factor a _i is updated in this way and the accuracy and operating rate for providing the Pareto optimal solution are specified, the operation unit RU displays, for example, a Pareto front as shown in FIG. 1 on the operation screen. When the operator touches the Pareto front and designates desired accuracy and operating rate, a weighting factor a _i giving the specified accuracy and operating rate is specified and stored in the RAM 12. The CPU 10 executes the above-described equations (1) to (9) based on the weighting factor a _i stored in the RAM 12. Thereby, the accuracy and the operation rate can be changed to desired values. Each weighting factor a _i varies depending on the type of goods to be weighed, the transport characteristics, operating conditions, etc., but when tested with snacks, the accuracy was set to 0.2%. It has been confirmed that the rate is improved by 3%.

As mentioned above, although one Embodiment of this invention was described, it is not limited to this, Other forms are employable. For example, in this embodiment, the computer C that executes the genetic algorithm is externally attached. However, the computer C is incorporated in the operation unit RU, and the weighting coefficient a _i is updated while driving. You can also display the operating rate.

  The external computer C may be connected by either wired or wireless. Further, the external computer C may be a computer provided in a remote data center. Moreover, you may be comprised so that the function of the external computer C might be borne in a cloud.

  Further, the execution of the genetic algorithm may be configured to be executed when the combination weighing device is not operating (for example, at night) in order to reduce the processing load.

100 combination weighing device M1-Mn weighing machine
RU operation unit
C Computer executing genetic algorithm

Japanese Patent No. 3360895 JP 2009-47519 A

Claims (3)

A combination weighing device that selects the optimum combination of weighing machines by combining weighing values obtained from multiple weighing machines that weigh items, and the effect on the accuracy and operation of each combination selected. Prepare multiple evaluation functions for evaluating the impact on the rate, and when combining each measurement value, calculate the evaluation value based on the evaluation function for each combination, and the sum of the calculated evaluation values is the maximum A combination weighing device characterized by selecting a combination to be an optimal combination. Each of the evaluation functions includes a weighting coefficient that represents the degree of involvement of each evaluation function,
2. The combination weighing device according to claim 1, wherein the weighting factors are determined by a computer executing a genetic algorithm.   The combination weighing device according to claim 1 or 2 is provided with an operation unit, and from the operation unit, it is possible to arbitrarily set whether to perform an operation focusing on accuracy or an operation focusing on an operating rate. A combination weighing device characterized by that.