JP2988306B2 - Elevator group control device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an elevator group management control device for allocating calls using a neural network, and more particularly, to a reduction in the allocation performance even when the traffic situation fluctuates greatly after the operation of the elevator. The present invention relates to an elevator group management control device that can be prevented.

[0002]

2. Description of the Related Art Conventionally, elevator group management control has mainly been a call allocation method using an evaluation function or a call allocation control by an expert system using fuzzy logic. A new method of assigning calls using a neural network has been proposed.

A neural network is a network imitating the human brain. A plurality of neural cell models (neurons) are connected in a complex manner, and the operation of each neuron and the form of connection between the neurons are properly determined to perform pattern recognition. The function and the knowledge processing function can be embedded. For example, "Nikkei Electronics" 1987
P115-P124 and 1 of August 10, issue (No. 427)
It is disclosed in the book "PDP model" published by Sangyo Tosho Co., Ltd. in February 989, etc. In particular, those in which neurons are arranged in a hierarchical structure can use an autonomous learning algorithm called "back propagation". There are features.

When this neural network is used, there is no need for a human to consider an assignment algorithm at all, and a decision system for deciding an optimum assigned car in response to various traffic conditions can be automatically generated. For example, Japanese Patent Application Laid-Open No. 1-275381 "Elevator group management control device" and Japanese Patent Application Laid-Open No. 3-31173 "Elevator group management control device" Japanese Patent Application No. 5-2438
No. 17, "Method of Learning Neural Network for Elevator Call Assignment" and the like.

FIG. 4 shows an example of a neural network for call assignment. As shown in FIG. 4, the neural network NN for call assignment includes an input layer NR1 corresponding to an input pattern (elevator system state data), an output layer NR3 corresponding to an output pattern (assignment suitability),
It is composed of neurons of an intermediate layer NR2 located between the input layer and the output layer.

The input pattern uses various data representing the state of the elevator system (floor and direction of landing call, position and driving direction of each car, car call, load state, etc.) as parameters necessary for call assignment. , Into a form that can be input to a neural network, and the number of neurons in the input layer corresponds to the total number of parameters.

When input data is given to each neuron in the input layer, a signal is sequentially transmitted to the output layer, and as a result, a certain value is output from each neuron in the output layer. In the output layer, there are neurons for the number of elevators, and for various input patterns, the neuron corresponding to the car that is optimal for allocation outputs “1” (maximum value), and the other neurons output “0”. Is learned in advance.

Accordingly, among the values of each neuron in the output pattern, a neuron that outputs a value closest to “1” (maximum value) indicates that it is optimal for allocation.
The unit corresponding to this neuron is selected as the assigned unit. Note that the number of neurons in the intermediate layer (one layer in the embodiment, but may be plural) is appropriately determined according to the number of elevators, the properties of the building, and the like.

Although not shown, a connection weight (synapse weight) indicating the strength of the connection between the neurons is set between the neurons. The connection weight is initially set to an appropriate value, but can be modified so that a more accurate call assignment can be performed using a learning algorithm called “back propagation”.

Since this back propagation is well known and will not be described in detail, a learning sample (an input pattern and a desired output pattern corresponding to the input pattern, that is, a teacher signal are paired). An algorithm that compares the output pattern for the same input pattern with the teacher signal and corrects the connection weights to minimize the error. Initially, all weights are initialized (for example, random The input pattern of the learning sample is given to each neuron of the input layer. The output pattern at this time and the output pattern of the learning sample (teacher signal)
Then, using the difference (error), the values of the connection weights are sequentially corrected from the output layer side so that the difference becomes smaller.

If this is repeated using a large number of learning samples until the error converges, the call assignment function at the same level as the teacher signal is automatically embedded in the neural network, and the learning input The same level of call assignment as the teacher signal can be performed not only for patterns but also for unknown input patterns.

[0012]

This neural network for assignment is installed in an actual site after performing a series of learning by using a learning sample created in advance based on simulations, actual measurement data on the site, and the like. However, if the traffic conditions of the building fluctuate due to changes in specifications or the movement of tenants after the operation of the elevator, the subsequent allocation performance may be significantly reduced. In such a case, it is sufficient to create a learning sample corresponding to the traffic condition after the change and re-learn the neural network. There is a problem that the performance continues to be greatly reduced.

The present invention has been made in view of such problems, and a group management system that does not significantly reduce the allocation performance even when the traffic situation of a building fluctuates greatly after the operation of the elevator. The purpose is to provide.

[0014]

In order to achieve the above object, according to the present invention, a first allocating means for allocating by a neural network, a second allocating means for allocating by a conventional method, and a hall call are generated. Allocation method selection means for selecting one of the first allocation means and the second allocation means at a preset rate each time the processing is performed, allocation performance by the first allocation means, and the second allocation means Allocation performance evaluation means for comparing and evaluating the allocation performance according to, and ratio setting value changing means for changing the setting value of the ratio according to the result of the evaluation,
A learning means for re-learning the neural network when a ratio setting value of the first allocating means becomes equal to or less than a predetermined value.

[0015]

In the elevator group management apparatus according to the present invention, each time a hall call occurs, the elevator is assigned by the first assigning means using a neural network according to a preset ratio or by a conventional method other than the neural network. The assignment is performed by selecting whether the assignment is made by the second assignment means, and when the call is serviced, the assignment performance (such as the waiting time of the call) is stored. After performing this for a predetermined number of hall calls or for a predetermined period, the performance (average waiting time, etc.) of the first and second allocating means is compared. Is higher, the ratio allocated by the first allocator is increased, and conversely, if the allocation performance by the second allocator is better, the ratio by the second allocator is increased. This is repeated, and when the ratio of the first allocating unit becomes equal to or less than the predetermined value, the learning of the neural network is performed again.

[0016]

FIG. 1 is a block diagram showing the overall configuration of the present invention. In the drawing, reference numeral 1 denotes a hall call button provided on each floor (only one floor is shown, and other floors are shown). (Abbreviated), 2 is a hall call signal, A1 is an operation control device for managing the operation of the first unit, similarly A2 to An are operation control devices for the second to nth units, 3 is the state of each unit (car) A car information signal indicating the position, the driving direction, whether the vehicle is running or stopped, whether the door is open or closed, a car call, a load, a service floor, an abnormality, etc.) 10 is an elevator system state comprising the hall call signal 2 and the car information signal 3 A group management device that assigns a hall call to an optimal car based on data and outputs the assigned car call as an assignment signal 4. Each of the operation control devices A1 to An controls the hall call assigned by the assignment signal 4 and the car of the own car. Basket call registered within To control the operation of the car to sequentially response.

The group management device 10 comprises a microcomputer or the like, and uses an input / output interface 11, a first allocating means 12 composed of a neural network, and a conventional method other than the neural network, for example, an evaluation function or a fuzzy rule. A second allocating unit 13 using a conventional allocating method, a learning unit 14 for learning a neural network, and either the first or the second allocating unit according to a preset ratio when a hall call occurs. An allocation method selecting means 15 for selecting, a first or second
Allocation performance evaluation means for comparing and evaluating the performance from the results allocated by the allocation means, and ratio setting value changing means 17 for changing the set value of the ratio in accordance with the evaluation result of the evaluation means 16. Each of these means is realized on software in a microcomputer.

In the above configuration, the operation procedure of the present invention will be described with reference to the flowchart of FIG. First, in step S1, it is confirmed whether or not the assignment by the neural network is possible, that is, whether or not the neural network is in a state where the assignment is not possible due to learning or the like. In step S2, it is determined whether or not a new hall call has occurred. If a hall call has occurred, in step S3, an allocation method is selected, that is, the first allocation means using a neural network or the second allocation method using a conventional method. Means or
The selection is made according to a preset ratio. For example, assuming that the set ratio is 5: 5, the first assigning means and the second assigning means are alternately selected each time a hall call occurs.

In step S4, the assignment is performed by using the selected assigning means.
After confirming that the call has been serviced in step 5, an index indicating the assigned performance at that time, for example, the actual waiting time of the call, is calculated and stored in step S6. In step S7, it is determined whether or not a predetermined period has elapsed, and if not, steps S2 to S7 are repeatedly executed. In step S7, it may be checked whether or not the number of assigned hall calls has reached a predetermined number instead of the predetermined period.

Steps S2 to S7 are repeated in this way, and after a predetermined period has elapsed, or when the number of assigned calls reaches the predetermined number, in step S8, the allocation performance of each allocation method during that period is compared and evaluated. The comparison of the allocation performance is performed, for example, by calculating the average waiting time for each allocating means from the waiting time of each call stored in step S6 and comparing the average waiting times. Then, in step S9, it is determined which allocation performance is better, in this example, which average waiting time is shorter. If the first allocation means using the neural network has better performance, the process proceeds to step S10. Then, the above-mentioned preset ratio is changed so that the ratio of the allocation by the first allocation means is increased (for example, 6: 4).

Conversely, if the conventional method has better performance, the process proceeds from step S9 to step S11, and this time, the ratio of allocation by the second allocation means is increased (for example, 4: 6). ), Change the percentage setting. It should be noted that the rate of increase / decrease of the ratio at one time may be a constant value, or may be a value corresponding to a difference in allocation performance.

As a result, when the process proceeds to step S10, the process returns to step S2. However, when the process proceeds to step S11, and the ratio of the allocation by the neural network decreases, the value becomes equal to or less than the predetermined value in step S12. Check if it is not, and if it is below the specified value,
In step S13, the subsequent allocation is entirely switched to the conventional method, and in step S14, re-learning of the neural network is performed. That is, if there is a large fluctuation in the traffic situation of the building, the allocation performance of the neural network temporarily decreases, so the rate of allocation by the neural network gradually increases while the above procedure is repeated. It is conceivable that the temperature decreases and eventually falls below a predetermined value. In such a case, it is determined that it is no longer appropriate to use the current neural network for assignment, and the neural network is relearned. During this time, since the assignment by the neural network cannot be performed, the assignment is entirely switched to the conventional method, and thereafter, when the learning of the neural network is completed and the assignment becomes possible, the process proceeds from step S1 to S2 again, and The procedure will be repeated.

Next, an example of a learning procedure when the neural network performs re-learning will be described with reference to a flowchart of FIG. First, it is determined in step S21 whether a new hall call has occurred. If a call has occurred, in step S22,
The elevator system state data at the time of occurrence of the call is converted into a neural network input pattern. In step S23, it is confirmed whether or not the call has been serviced. If the call has been serviced, in step S24, the serviced car is converted to an output pattern of a neural network as an assigned car. At this time, if the initially assigned number is different from the actually assigned number due to an assignment change or reassignment, etc., that is, if the initial assignment is not the best as a result (however, since the operation has been switched to the dedicated operation, In the case of the case where the service cannot be performed, the actually serviced car is converted into the output pattern as the assigned car.

In step S25, the input pattern and the output pattern are stored as a learning sample in pairs. The assignment at the time of creating the learning sample is performed using, for example, a conventional method.

After step S26, the above procedure is repeatedly executed until the number of learning samples reaches a predetermined number. In step S27, learning of the neural network is performed using the collected learning samples. It is modified into a neural net corresponding to the situation and can be used again for call assignment.

[0026]

According to the present invention, even if the traffic condition of the building fluctuates greatly and the neural network cannot temporarily cope with it, the allocation by the conventional method can be secured during that time, so that the allocation performance is greatly improved. Can be prevented.

[Brief description of the drawings]

FIG. 1 is a block diagram showing the overall configuration of the present invention.

FIG. 2 is a flowchart showing an operation procedure of the present invention.

FIG. 3 is a flowchart showing a procedure for learning a neural network.

FIG. 4 is a diagram showing an example of a neural network for call assignment.

[Explanation of symbols]

 A1 to An 1 to n-th operation control device 1 hall call button 2 hall call signal 3 car information signal 4 allocation signal 10 group management device 11 input / output interface 12 first allocation means 13 second allocation means 14 learning means 15 Assignment method selection means 16 Assignment performance evaluation means 17 Ratio setting value change means

──────────────────────────────────────────────────続 き Continued on the front page (72) Inventor Mami Nakagawa 1-28-10 Sho, Ibaraki-shi, Osaka Inside Fujitec R & D Co., Ltd. No. stock Company Fujitec technology Institute in (72) invention's shares Company Fujitec technology research Institute within the examiner Hiroshi Shimizu Naoki Ota Ibaraki, Osaka Sho No. 1-chome 28 No. 10 (58) investigated the field (Int.Cl. ^6, (DB name) B66B 1/18-1/20

Claims (2)

(57) [Claims] When a plurality of elevators are put into service on a plurality of floors and a hall call occurs, various data representing the state of the elevator system at that time are converted into a form that can be used as an input pattern of a neural network. First grouping control means for assigning by a neural network in an elevator group management control device which selects the most suitable car from the output pattern and assigns the car to the hall call. A second assigning means for assigning according to an assignment method other than the neural network, and selecting one of the first assigning means and the second assigning means at a preset ratio each time a hall call occurs. Allocation method selection means, allocation performance by the first allocation means, and the second allocation Allocation performance evaluation means for comparing and evaluating the allocation performance of the means, rate setting value changing means for changing the ratio setting value according to the evaluation result, and a ratio setting value of the first allocation means being equal to or less than a predetermined value. And a learning means for re-learning the neural network. 2. The group of elevators according to claim 1, wherein the assignment method other than the neural network is a call assignment method using an evaluation function or a call assignment method using an expert system using fuzzy logic. Management control unit.