FI89895C - Procedure for streamlining control of elevator doors - Google Patents

Description

39895

METHOD FOR IMPROVING THE CONTROL OF ELEVATOR DOORS

The present invention relates to a method according to the preamble of claim 1 for enhancing the control of elevator doors.

In modern express elevators, door control is not optimal. The actual control of the movement of the door leaves is good in modern, fully electronically controlled doors. Instead, the timing of the doors, which in this context means the opening time of the door, and if the door approaches an obstacle and reopens when closed, the so-called reopening time, are less well controlled. The biggest drawbacks are related to the selection of the closing time 15.

Especially doors that try to close too quickly in traffic jams cause confusion and thus additional downtime on the levels lasting several seconds. To prevent this from happening, the basic timing settings for the door 20 are long, resulting in the doors being open for an unnecessarily long time, usually several seconds. One second of additional downtime per stop causes a loss of up to 5% in the capacity of the express lift 25 During quiet traffic, the effect of unnecessarily long door opening times on transport capacity is naturally insignificant, but unnecessarily long opening hours annoy passengers.

A method is known in which elevator control predicts door times using information about the type of arrival floor, i.e. whether it is a base floor or a normal floor, and whether a parallel call is printed on the level, from which it can be concluded that passengers are aiming for the elevator car 35 when the door opens. Such a system works correctly when set up satisfactorily during heavy traffic, but in other types of traffic and in special traffic situations it does not work satisfactorily due to a lack of information in the decision-making situation.

A method is also known in which the control of the elevator door 5 mainly utilizes the signal of the safety list so that the door tends to close immediately, but turns immediately if the list identifies the passenger. This method has the disadvantage that passengers are particularly frightened by the rapidly approaching door and hesitate and even stop, resulting in a considerable reduction in capacity. For this reason, this method is hardly used in new, high-speed doors.

A method is also known in which passengers themselves use the door opening and closing buttons. This works satisfactorily 15 especially in office buildings in countries where such a culture of use has developed. It is clear that such a system only works in part because some passengers do not take control of the doors, and with such a passenger standing in front of the buttons, other passengers have to wait until the door closes-20 by itself.

Methods are also known for optimizing door opening hours and movements by considering passengers at the level. The problem then is to leave the full or almost full 25 baskets.

Determining the correct closing time is complicated due to several factors. Some of them are non-linear in nature and are also often interdependent.

U.S. Patent No. 4,491,199 discloses a method for enhancing elevator door control in which passengers in an elevator car are monitored by a sensor and those from the sensor are monitored! the data is used to control the door. Still the publication predicts the number of passengers leaving the elevator car, and 3,89895 this information is used to control the door. GB patent 1,109,005 discloses a method in which passengers are also monitored outside the elevator car and in the doorway.

The method according to the invention can eliminate the disadvantages of the prior art and improve the optimization of door times. Characteristic features of the method according to the invention for enhancing the control of elevator doors are set out in the claims below.

10

In the method according to the invention, the processing of complex, nonlinear dependencies is easier due to the properties of the neural network than in known solutions.

15 The greatest benefit of improved door time optimization is achieved in a situation where one or a few people are trying to leave the rear of a full basket. This situation is most common in congestion, which is also generally the most critical situation in terms of transport capacity, and is therefore generally used as a sizing criterion. If the passenger does not get to the door in time, there will be confusion, and if the door times are long, the door will be open unnecessarily.

In this situation, the predicted opening time of the door is dependent - 25 nen mm. the number of persons in the basket and the commands they give, the intentions of the persons in the basket to leave, the type of traffic prevailing, the type of arrival layer, the direction from which the basket enters the level, the number of persons on the level and the intentions to ascend 30 baskets.

In addition, human behavior is not always the same, so a system that seeks to predict when the correct door 35 will close must be dynamic. In addition, it must operate on a strong statistical basis.

4 89395

In the following, the invention will be described in more detail by way of example with reference to the accompanying drawings, in which

Figure 1a shows an elevator door.

5

Figure ib shows a passenger in an elevator car.

Figure 2 shows the implementation of the neural network.

Figure 3 shows a system according to the invention in block diagram form.

Figure 4 shows a typical exit situation.

In the method according to the invention, in the opening of the elevator doors 1 and at least partially inside the elevator car there is a sensor 2 which forms, for example, a two-person deep observation field 3 in the elevator car near the doors provided by information and other sensors and devices connected to the door time optimizer. the data are used to form a continuous or intermittent exit profile. The monitoring field is usually substantially inside the elevator car, but depending on the sensor and its location, the monitoring field may also comprise 2 * 5 parts of the area in front of the elevator doors in the elevator lobby.

In this context, the exit profile means the passenger queue according to Fig. 4, shown as a function of time, identified in the doorway or in its immediate vicinity. 30 Combining this information, which therefore comes from the door traffic sensor, with the exit model selected from the exit pattern profiles, a prediction is obtained as to whether it is probable that the car still has the passengers leaving the floor. The forecast is used to generate door opening maintenance 35 and closing and reopening commands. In addition, the best door speed for each time is deduced from the same data.

5 39395

Certain fully dominant sensor data, such as door open and close button data and the traditional short detection safety list, can be connected either to the optimizing device 5 of the present invention, which has a very high weight in the decision, or as a logic input in a normal door control unit. the information overrides, if necessary, the instructions given by the device according to the invention in each case.

10 This will lead to at least the following improvements: shorter opening hours and more friendly door behavior.

The implementation of the system according to the invention is briefly described below:

The door traffic detector, which operates as a neural network-based data processing device or function according to Fig. 2, contains a number of models for different door traffic situations according to the above-mentioned exit profile principle. They can be standard patterns or the result of learning in your own teaching space.

According to Figure 2, the neural network consists of three layers 4-6. The first layer 4 forms an input layer which divides the second layer 5 of the input vector 7 formed from the inputs of the sensor devices, which forms a hidden layer. A vector is formed at the output 6 of the neural network, from which a signal 8 is obtained to form the door control. Each neuron contains e.g. blocks for calculating output and modifying network behavior during teaching, and weighting factors that determine network behavior.

The sensor devices used are adapted to the module of Figure 3 using different transducers depending on the quality of the original signal 6 39895. Block 10 illustrates an inward sensor adapter, block 20 a photocell adapter, and block 30 a basket outward sensor adapter.

5 Block 40 describes e.g. an adapter for the location, load, direction information and remaining calls of the elevator car, i.e. an adapter for the data coming from the computer controlling the elevator, the data entry speed of which is not critical for the performance of the method.

10

Block 50 describes an adapter for a sensor (so-called mechanical or ecetronic safety-1) at the front edge of a door in an elevator car.

15 Blocks 60 and 70 illustrate an adapter for other information from the elevator car. These include e.g. door open and close buttons. The adapters are connected to the control computer 80.

To illustrate the operation, the following example mode-20 are discussed:

Figure 4 shows a typical situation in which a half-full car in a typical elevator of a larger office building with a maximum number of people of 16 leaves 25 people. Other situational data include: congestion, car halfway between the gap going up, three more indoor calls printed and load data suggesting that there are still 8 people in the car.

The first exit 12 has already stood ready in front of the door and starts the exit immediately when the door opening is about 800 mm. He is identified by the monitor immediately when the field monitoring is started, for example when the door is opened.

3 5 7 39395

For simplicity, this discusses a solution in which the coverage of the observation field inside the car is only one person's "deep" lane near the door. If more precise sensor device technology is used, a wider sensor field, even covering the entire body 5, is useful if the information obtained can detect the flow of passengers towards the door opening.

The detection devices 10 in the slot and / or outside the door function mainly to increase the reliability of the information of the method which is the subject of the invention by facilitating e.g. getting a better picture of the movement. Of course, all detection devices can also function in other or new existing tasks, such as a traditional infrared photocell usually operating in a doorway, which information is used in modern door controls to identify people that may not be used to control the closing of a door.

20

The second outlet 13 has stood close to the door (e.g. at a rear slope) and enters the observation field (in the case of one inner field) practically immediately after the first.

25

The third exit 14 has stood behind, which he has reached during the loading phase. Because the car has been almost full as it leaves the base floor, and since the car has emptied an average of 30 people per stop during the previous three stops, she has not yet started to push forward and thus enters the observation field about 1 to 2 s after the second exit.

In this situation, the door traffic sensor infers from the above traffic situation data and the detected exit situation that the fourth exit is unlikely to arrive again, and 8 39895 informs the door control, which initiates the closing of the door.

From the remaining number of passengers, which in this case 5 is 8, from the remaining calls, in this situation 3, uplink and congestion information, it can be assumed from the simulations that the number of departures is 2-3.

10 If the prediction turns out to be incorrect and the fourth person leaving is reasonably fast, he will still have time in the identification field in time to prevent the door from closing and re-issue the opening command. If the above-mentioned known solutions with short door times had been used, the the person most often got stuck in the basket and had to continue the journey to the wrong floor.

It will be clear to a person skilled in the art that the above-mentioned situation is one of the few most typical situations in which the decision to close the door 20 is complicated and in which the method which is the subject of the invention is of great benefit.

In many other situations, e.g. when the basket is almost empty, decision-making is considerably easier, but also in these situations the method according to the invention is useful. For example, when a passenger moves slowly due to an injury or burden, or when a passenger is a slowly moving elderly or child, the method according to the invention achieves an advantage.

Ύ0

It is still clear to a person skilled in the art that the above-mentioned situations are so multifaceted that it is difficult for a traditional algorithm to reach an optimal solution. Therefore, the solution according to the invention uses an if · neural network solution, in which the processing of complex, nonlinear dependencies is easier due to its properties.

It can be mentioned that the use of known solutions 5 would, in the exemplary situation, lead to a slower closing of the door by several seconds, as similar situations in the past would have caused complaints, as a result of which the door times would have been extended.

It will also be apparent to those skilled in the art that the various embodiments of the invention are not limited to the example described above, but may vary within the scope of the claims set forth below.

Claims (6)

10: 9895 A method for improving the control of elevator doors by monitoring passengers in an elevator car with one or more sensors (2) at least in the monitoring area (3) near the door or doors (1), and predicting the number of passengers leaving the elevator car using in the control of a door or doors, characterized in that the number of 10 passengers leaving the elevator car is predicted as self-learning by means of a neural network. 2. The first patent claim, which is incorporated into the invention, is described in the preamble to paragraph 15 of this Regulation. Method according to Claim 1, characterized in that the exit profile of the passengers leaving the elevator car is formed, by means of which the maintenance instructions for opening, closing and reopening the doors are formed. 3. An application is made to a patented simulation of the genome of the simulant. Method according to one of the preceding claims, characterized in that the exit profile is implemented using 20 continuous simulations. 4. For example, the passage of a patented device, including the passage of an indicator, is intended to be used in the elevator and the elevator car. 25 Method according to one of the preceding claims, characterized in that the passengers are monitored by means of sensors in addition to the elevator car in the slot of the door and / or outside the elevator car 25. Method according to one of the preceding claims, characterized in that the neural network operates in a neural network computer. 30 Method according to one of Claims 1 to 4, characterized in that the neural network-type operation is simulated in the control computer of the elevator system or in the computer performing the door control or in the auxiliary computer connected to the computer. il ’39895. 1. Effect for the effect of an elevator lift, the passage of which is indicated by an elevator or an indicator (5) the same information is provided for the purposes of the information provided for the purpose of the information provided, for example, for the purposes of the elevator 10. 5. For the purposes of this Regulation, the patent law, including the terms of the invention, is equivalent to that of the invention. * 6. For example, one of the patent simulators 1 to 4, 30 of which is a function of a neural system simulator with a styrofoam or a computer with a styrofoam or a computer with an assistant computer.