Group Control elevator Scheduling Algorithm

1) Clarify the group control elevator SchedulingAlgorithmEvaluation Indicators
Because the length of the passengers' psychological waiting time, the speed of the elevator's response to the call ladder, the size of the passenger flow of the call Hall Station, and the number of passengers in the car are all vague concepts, it is difficult to define the relationship by exact quantity, or to describe it by using common logical rules.

In recent years, we have used the membership functions in Fuzzy Mathematics to convert complicated fuzzy problems into simple and clear forms for solution and control. fuzzy control uses fuzzy logic to make an effective judgment on the operation status of the elevator. However, it is very difficult to establish correct Fuzzy Rules and membership functions for complicated multi-variable systems, moreover, the subordinate functions and rules established through a large number of experiments are sometimes difficult to ensure accuracy and reasonableness. In addition, the weighting coefficient in the membership function is determined and cannot be changed based on the change in passenger flow.

In order to solve some problems in fuzzy control, the new invention applies the neural network control method to elevator control, without the need to establish an accurate mathematical model, and can provide accurate control policies to reduce the waiting time, reduce the anxiety of passengers, save energy, and reasonably and effectively schedule the optimal operation of the elevator.

(2) understand the concepts of the uplink peak mode, the downlink peak mode, and the dual-path operation mode, and find an algorithm that indirectly calculates the operation mode based on a series of input methods:

Uplink peak traffic mode: when the main passenger flow is the uplink direction, that is, all or most passengers enter the elevator and move up from the Building lobby, this situation is defined as the uplink peak traffic.

Downstream peak traffic mode: when the main passenger flow is downstream, that is, all or most passengers take the elevator down to the entrance hall to exit the elevator, this situation is defined as downstream peak traffic conditions.

Second-Road Traffic: when the main passenger flow is directed to or from a certain layer, and this layer is not the entrance hall, this situation is defined as second-road traffic. The second road traffic condition is mostly because there is a refreshments or conference room on the first floor of the building. At a certain time of the day, the floor attracted a lot of signals of arrival and departure. Therefore, the traffic condition on the second road occurs during the morning and afternoon breaks or during the meeting.

Four-way traffic: when the main passenger flow is directed to two specific floors, one of which may be the entrance hall, this traffic condition is defined as four-way traffic. During the noon break period, there will be peak traffic in both the upstream and downstream directions. At lunch, the main passenger flows are downstream, toward the entrance hall and dining room. When the lunch break ends, it mainly goes up from the lobby and dining room. Therefore, the four-way traffic usually occurs during the lunch break. Four-way transportation can be divided into pre-lunch transportation and post-lunch transportation modes. The two traffic modes differ from the upstream and downstream peaks that occur in the morning and evening. Although the main passenger flows are both upstream and downstream, however, there is also a considerable proportion of Inter-layer traffic and traffic in the opposite direction. The proportion of each traffic volume is also related to the length of lunch break, the location of the restaurant and the usage of the building. The main traffic flows and other flows must be considered during the four-way traffic, which is different from the upper and lower peak periods.

Balanced inter-layer traffic mode: when the number of upstream and downstream passengers is roughly the same, and the traffic demand between each layer is basically balancedAt this time, the traffic mode is in a normal two-way inter-layer traffic situation, it exists in the day's bigIn part of the time, a passenger usually requires the minimum waiting time and ride time.

Idle traffic mode: the idle traffic mode usually occurs on holidays, late night, and Dawn. At this time, the building has few passengers and the arrival interval of passengers is very long, in this situation, only some elevators in the group control system work, while other elevator cars are idle.

Traffic Pattern Recognition Based on Neural Networks

Traffic Pattern Recognition Based on statistical rules

(3) What scheduling algorithms are applicable to different running modes?

1. Elevator Group Control Scheduling Algorithm Based on Expert System [8]

The elevator group control system is a complex nonlinear system with a large amount of uncertain and incomplete information. It is difficult to accurately describe such a complex system engineering problem using mathematical methods, but it requires * experience and knowledge that has not yet formed a scientific system. Expert systems are the methods for knowledge representation, use, and acquisition to deal with engineering and technical problems of complex systems. It is enlightening and can use the knowledge and experience of experts to make heuristic reasoning for uncertain or inaccurate problems. It has certain superiority in solving the elevator group control system.

The elevator group control scheduling algorithm based on the expert system is to establish rules based on the experts' experience and use these rules to compare various possible elevator operation routes and select the best route from them, this increases the passenger's transportation capacity and reduces the waiting time. It is applicable to the interlayer mode, but not to the upper peak, because the range of the planned optimal route under the interlayer traffic is large, but it does not predict the acceleration/deceleration time of the car.

This method increases the flexibility of system control, but there are also deficiencies. The entire control process is completely dependent on the knowledge source, obviously, the availability and perfection of expert knowledge determine the performance of elevator scheduling methods, which will be affected by the comprehensiveness of knowledge sources. At the same time, in order to meet the diversity of service requirements, the control rules will inevitably increase a lot, but it may not fully reflect the problem.

2. Elevator Group Control and Scheduling Technology Using Fuzzy Control [9]

Fuzzy logic is an effective method in mathematics and logic for dealing with blurred boundary objects and problems.The basic idea of fuzzy control is to summarize the control policies of human experts on specific controlled objects or processes.A series of control rules expressed in the form of "If (condition) Then (Action)". The control function set is obtained through fuzzy inference, which acts on the controlled object or process, processing Control System variable information with unknown language expressions [7].

The Application of Fuzzy Control Technology in the elevator control system shows a high degree of superiority. The elevator system contains many fuzzy and incomplete information, which must be described in a fuzzy set. Computers cannot accept vague answers, but can use fuzzy logic for reasoning. It can imitate the reasoning ability of the human brain and simplify many complicated questions. The fuzzy control method has the following advantages:

(1) fuzzy control is implemented based on the operator's control experience, without the need to establish a mathematical model. It is an effective way to solve the uncertainty of the system.

(2) fuzzy control has strong robustness. Changes in the controlled object parameters have little impact on fuzzy control and can be used for nonlinear, time-varying, and time-delay system control.

(3) controlled query tables are obtained from offline computation, improving the real-time performance of the control system.

(4) The control mechanism is in line with the intuitive description and thinking logic of People's role in Process Control.

The average waiting time of the elevator group control system with fuzzy logic is reduced, which is much better than that of the conventional elevator group control system. However, because fuzzy control itself does not have the learning function, many rules are difficult to determine and rely on Heuristic knowledge such as expert knowledge or blackboard structure, the control system cannot keep up with the changes of building traffic, it is difficult to achieve optimal scheduling under the control target requirement.

3. Neural Network-based group control scheduling method [10] [11] [12]

The Application of Expert Systems, fuzzy logic, and other artificial intelligence technologies in the elevator group control system greatly improves the service quality and efficiency of the ladder group, but their critical weakness is that they cannot improve control algorithms through learning. Therefore, when the passenger flow changes, the system cannot fully adapt to this change, resulting in the following problems:

If the actual situation of a building is different from that of an expert, the rules based on this assumption cannot bring better results. The system performance is determined by the knowledge and experience of experts and has certain limitations; it is difficult to adjust the fuzzy level membership function, and a lot of simulation is required. Once the rules are integrated into the system, it takes a lot of time and effort to change, and it can only be done manually.

In order to adapt the elevator group control system to changes in conditions, the system makes real-time automatic adjustments and provides better services in various situations, so that the system has the self-learning ability, artificial Neural Networks are introduced into the elevator group control system. The advantages of the introduction of Shenjing network technology to the elevator group control system include: when the building conditions imagined by experts are different from those of the actual building, the elevator group control system with neural network has the learning ability. A suitable model is established using nonlinear and learning methods for high-speed reasoning and short and long-term prediction of elevator traffic.

4. Scheduling Method of Fuzzy Neural Networks [13]

Fuzzy Neural Networks combine neural networks with fuzzy logic and take advantage of their respective advantages. they overcome the difficulty of determining the structure of the artificial neural network and the disadvantages of the non-self-learning function of fuzzy logic. Fuzzy Neural Networks are capable of acquiring and learning knowledge. The process of network learning is the process of optimizing fuzzy logical rules. By learning to adjust the network weight, you can optimize the rules based on the premise of a fuzzy rule, the relative importance of a fuzzy set, and the relative importance of each rule. On the other hand, it provides an understandable model structure for interpretation and reasoning and can describe knowledge in a clear way.

The establishment and use of a fuzzy neural network are divided into the following steps: using expert knowledge to roughly form a fuzzy model and some fuzzy rules and fuzzy inference methods. Based on this fuzzy model, a neural network is constructed; train the sutra network. Some specific traffic conditions are collected as samples, and corresponding algorithms are used for learning. The necessary weights of the neural network are adjusted to obtain optimized Fuzzy Rules and then applied to the network.

5. Genetic algorithm-based group control scheduling algorithm [14]

Since the 1940s s of this century, Biological simulation has become an integral part of computer science. Genetic Algorithms are a new scientific paradigm that takes complex problems as objects in the connotation of uncertainty, non-linearity, and irreversible time. The genetic transfer algorithm is mainly used to simulate the natural selection in the biological field and the evolution of natural genetic mechanisms, and is used to simulate the process at the molecular level, because of the high adaptability of genetic algorithms to problems and the capability of embedding parallel exploration, genetic algorithms have been widely used in optimization.

In the elevator group control system, as a practical and robust optimization method, genetic algorithms can be used for dynamic partitioning and parameter optimization. It not only adapts to the overall needs of the entire building, but also adapts to the different needs of the elevator on each floor, achieving personalized control of different floors. However, the biological basis of genetic algorithms is clear and the mathematical basis is not perfect. Currently, there are still problems with search efficiency and real-time performance. Therefore, it is one of the best tools to seek satisfactory solutions, however, it is unrealistic to solve the optimal solution to complex problems.

6. Elevator Group Control Scheduling Algorithm Based on Multi-objective Optimization [15]

Multi-objective problems are decision-making problems with multiple implementation goals. multi-objective optimization is to study the nature of multi-objective decision-making problems, and to convert multiple goals pursued by decision-making problems into single goals through modeling, then the processing method for obtaining the optimal solution. The main optimization objectives of the elevator group control system include reducing the average waiting time of passengers, reducing the average waiting time of passengers, and reducing the energy consumption during system operation, therefore, the elevator group control algorithm is a typical multi-objective decision-making problem.

For a multi-objective optimization problem, if an evaluation function can be constructed based on the preference information provided by the designer, so that the solution designer's most satisfactory solution is equivalent to the optimal solution of the new target function after the solution is evaluated, the problem of this multi-objective problem is quantifiable, the multi-objective optimization theory is to study the conditions of such evaluation functions and how to construct them.

Because today's elevator group control scheduling algorithms mostly use certain statistical rules to solve the sub-optimal solution to the multi-objective problems of the elevator group control system, and then distribute and schedule them. Modeling is difficult, learning takes a long time, and control is not timely. When optimization conditions are broken, the optimal scheduling scheme cannot be provided. Multi-objective Optimization can obtain the optimal target function through modeling, which is the best solution to the elevator group control decision-making problem.

(4) What are the significant improvements in efficiency after the system adopts this new algorithm.

(5) Based on these algorithms, how can we estimate the maximum number of elevators required for a given passenger flow? How to schedule?

(Original address Http://blog.csdn.net/binbin127/archive/2009/07/09/4331090.aspx)