Elevator intelligent dispatching and maintenance early warning system and method
By using LSTM models and multi-sensor data analysis, intelligent scheduling and early warning of elevators are achieved, solving problems such as unreasonable scheduling during peak hours, insufficient fault warnings, and unreasonable maintenance cycles, thereby improving elevator operating efficiency and safety.
Patent Information
- Application Number
- CN202511176649.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-08-21
- Publication Date
- 2025-10-28
AI Technical Summary
Traditional elevator dispatching algorithms cannot be adjusted dynamically, have weak fault warning capabilities, and have unreasonable maintenance cycles, resulting in low efficiency during peak hours, long fault handling times, and wasted maintenance resources.
The system uses an LSTM neural network model to predict passenger flow, combines multi-sensor data to calculate elevator health index, dynamically adjusts maintenance cycles, and adjusts elevator operation priority in emergency situations to achieve intelligent scheduling and early warning.
It improves the operating efficiency of elevators during peak hours, shortens fault handling time, optimizes maintenance resource allocation, and enhances the safety and emergency response capabilities of elevators.
Smart Images
Figure CN120841329A_ABST
Abstract
Description
Technical Field
[0001] This invention discloses an intelligent elevator scheduling and maintenance early warning system, which relates to the field of elevator management technology. Background Art
[0002] Inefficient peak-hour scheduling: Traditional elevator scheduling algorithms often use fixed patterns or simple response mechanisms, which cannot dynamically adjust according to changes in passenger flow. For example, during the morning rush hour in office buildings, a large number of passengers are concentrated on the lower floors, and the frequent back-and-forth of elevators leads to excessively long waiting times for passengers on higher floors, with an average waiting time of more than 5 minutes. During peak hours, elevator utilization is less than 60%.
[0003] Weak fault early warning capabilities: Existing elevator fault monitoring systems mostly rely on single sensors or simple threshold judgments, lacking comprehensive analytical capabilities. Statistics show that approximately 78% of elevator faults are sudden and cannot be predicted in advance, leading to safety hazards such as passengers being trapped when a fault occurs. The average time to handle each fault exceeds 45 minutes.
[0004] Inappropriate maintenance cycles: Currently, elevator maintenance often adopts a fixed cycle model (such as once a month), without considering the actual usage intensity and differences in the elevator's condition. This results in approximately 30% of maintenance being over-maintenance (unnecessary maintenance performed on elevators in good condition), while approximately 25% of elevators suffer from under-maintenance, accelerating equipment aging.
[0005] These problems not only affect the efficiency and safety of elevator use, but also increase operating costs and management difficulty, making an intelligent elevator management system urgently needed to solve them. Summary of the Invention
[0006] This invention aims to provide an intelligent elevator scheduling and maintenance early warning system. This system, by integrating technologies such as passenger behavior prediction, multi-sensor health monitoring, adaptive maintenance scheduling, and emergency dynamic priority adjustment, solves the problems of traditional elevator systems, including peak-hour congestion, sudden malfunctions, and low maintenance efficiency. Specifically, it includes:
[0007] The data acquisition module is used to collect elevator operating parameters and passenger flow data;
[0008] The intelligent scheduling module enables dynamic elevator scheduling based on passenger flow prediction.
[0009] The health assessment module calculates the elevator health index through multi-sensor data fusion.
[0010] The maintenance scheduling module dynamically adjusts the maintenance cycle based on health index and usage intensity.
[0011] The emergency response module adjusts the elevator's operating priority in emergency situations.
[0012] Furthermore, the intelligent scheduling module uses an LSTM neural network model to predict passenger flow and divides the elevators into different service groups based on the prediction results to achieve dynamic scheduling by zone.
[0013] Furthermore, the health assessment module collects multi-dimensional data through vibration, temperature, noise, and current sensors to calculate the elevator health index H, using the following formula:
[0014] H=α×V_score+β×T_score+γ×N_score+δ×C_score
[0015] Where V_score is the vibration score, T_score is the temperature score, N_score is the noise score, C_score is the current score, and α, β, γ, and δ are weighting coefficients and their sum is 1.
[0016] Furthermore, the maintenance scheduling module calculates the actual maintenance cycle according to the following formula:
[0017] M = M0 × (k / H) × (1 + λ × U)
[0018] Where M0 is the baseline maintenance cycle, k is a constant, H is the health index, λ is the usage intensity coefficient, and U is the usage intensity.
[0019] Furthermore, the emergency response module can identify emergencies such as fires and earthquakes, as well as situations where passengers are trapped, automatically adjust the elevator's operating priority, and link with the building's emergency system.
[0020] Furthermore, a method for intelligent elevator scheduling and maintenance early warning is also provided, including the following steps:
[0021] (1) Collect elevator operating parameters and passenger flow data;
[0022] (2) Realize dynamic elevator scheduling based on traffic flow prediction;
[0023] (3) Calculate the elevator health index by integrating data from multiple sensors;
[0024] (4) Dynamically adjust the maintenance cycle based on health index and usage intensity;
[0025] (5) Adjust elevator operation priority in emergency situations.
[0026] Furthermore, in step (2), the dynamic scheduling of elevators based on traffic flow prediction includes:
[0027] Predict passenger flow on each floor at different times using an LSTM model;
[0028] Based on the prediction results, the elevators are divided into different service groups;
[0029] Monitor traffic changes in real time and dynamically adjust scheduling strategies.
[0030] Furthermore, in step (3), the calculation of the elevator health index includes:
[0031] Standardize the vibration, temperature, noise, and current data;
[0032] Calculate the scores V_score, T_score, N_score, and C_score for each indicator;
[0033] The health index H is calculated using a weighted summation formula.
[0034] Furthermore, in step (4), the dynamic adjustment of the maintenance cycle includes:
[0035] Obtain the baseline maintenance cycle M0 and the current health index H;
[0036] Calculate the elevator service intensity U;
[0037] The actual maintenance cycle is determined by applying the maintenance cycle calculation formula;
[0038] A maintenance plan is generated based on the actual maintenance cycle.
[0039] Furthermore, in step (5), adjusting the elevator operation priority in an emergency includes:
[0040] Identify the type and level of the emergency;
[0041] Suspend non-essential elevator services;
[0042] Prioritize dispatching elevators to the floor where the emergency occurred;
[0043] In conjunction with the building's emergency response system, optimize evacuation routes.
[0044] Beneficial effects:
[0045] Improve elevator operating efficiency: Through intelligent scheduling algorithms, the average waiting time for passengers during peak hours can be reduced, and the utilization rate of elevators can be improved.
[0046] Enhanced fault early warning capabilities: Enables early warning of potential faults, shortens fault handling time, and significantly improves elevator operation safety.
[0047] Optimize maintenance resource allocation: Reduce over-maintenance and avoid under-maintenance through adaptive maintenance scheduling, thereby reducing maintenance costs.
[0048] Enhanced emergency response capabilities: In emergency situations, elevator emergency response time is shortened, improving the overall safety of the building. Attached Figure Description
[0049] Figure 1 System overall architecture diagram. DETAILED DESCRIPTION
[0050] Example 1
[0051] This invention aims to solve three main problems existing in the prior art: first, the elevator scheduling efficiency is low during peak hours and cannot be dynamically adjusted according to passenger flow; second, the elevator fault early warning capability is weak and it is difficult to detect potential problems in advance; and third, the maintenance cycle is fixed, resulting in resource waste or insufficient maintenance.
[0052] To address the aforementioned technical problems, this invention provides an intelligent elevator scheduling and maintenance early warning system, comprising the following four innovative aspects:
[0053] LSTM-based passenger flow prediction and intelligent scheduling module
[0054] By analyzing historical data and real-time monitoring, a passenger flow prediction model was established.
[0055] A dynamic partitioning scheduling algorithm is adopted to allocate elevator resources in advance based on prediction results.
[0056] To achieve adaptive adjustment of the elevator group control system and reduce passenger waiting time.
[0057] Elevator health status assessment module with multi-sensor fusion
[0058] Integrates multi-dimensional sensor data including vibration, temperature, noise, and current.
[0059] The elevator health index calculation model is constructed, and the formula is as follows:
[0060] H=α×V_score+β×T_score+γ×N_score+δ×C_score
[0061] Where H is the elevator health index (0-100), V_score is the vibration score, T_score is the temperature score, N_score is the noise score, C_score is the current score, α, β, γ, and δ are the weighting coefficients of each index, and α+β+γ+δ=1.
[0062] Health-based adaptive maintenance scheduling module
[0063] The maintenance cycle is dynamically adjusted based on the elevator's health index and usage intensity.
[0064] The formula for calculating the maintenance cycle is as follows:
[0065] M = M0 × (k / H) × (1 + λ × U)
[0066] Where M is the actual maintenance cycle, M0 is the baseline maintenance cycle, k is a constant (usually taken as 80), H is the health index, λ is the usage intensity coefficient, and U is the usage intensity (0-1).
[0067] Dynamic priority adjustment module in emergency situations
[0068] Identify emergencies such as fires and earthquakes, as well as special situations such as passengers being trapped.
[0069] Automatically adjust elevator operation priority to ensure emergency response efficiency.
[0070] Enables linkage with building emergency systems to optimize evacuation routes.
[0071] Example 2
[0072] Office building morning rush hour intelligent dispatch scenario
[0073] Step 1: Passenger Flow Forecast
[0074] The system automatically activates the morning rush hour prediction mode at 6:30 AM on weekdays.
[0075] Based on the analysis of three months of historical morning rush hour data using an LSTM model, combined with factors such as the day's weather and date type,
[0076] Predict passenger flow on each floor between 7:30 and 9:00, with a focus on floors 8-12 (where companies are concentrated) where peak passenger flow is expected.
[0077] A heatmap of traffic flow was generated: the upward demand index for floors 8-12 is 0.85 (out of 1.0), while the average for other floors is 0.32.
[0078] Step 2: Dynamic Partition Scheduling
[0079] The system divides the six elevators into three groups:
[0080] Group A (2 sections): Focus on serving floors 8-12, starting from floor 1 at 7:30 AM with designated waiting areas.
[0081] Group B (2 units): Serving layers 1-7 and 13-18, using conventional scheduling.
[0082] Group C (2 units): As a mobile group, providing dynamic support based on real-time traffic flow.
[0083] "Pre-stop mode" activated: Group A elevators will arrive at the 1st floor at 7:20 AM to wait, avoiding elevator shortages during peak hours.
[0084] Real-time adjustment: When the number of people waiting on the 9th floor exceeds 5, one elevator will be automatically dispatched from Group B to provide support.
[0085] Step 3: Real-time monitoring and adjustment
[0086] Traffic predictions are updated every 5 minutes, and scheduling strategies are fine-tuned based on actual conditions.
[0087] When a sudden surge of passengers (more than 15 people) occurs on the 10th floor, the system will automatically extend the dwell time of elevator A on the 10th floor.
[0088] At 8:45, based on the decreasing traffic flow trend, elevators in Group A were gradually returned to the regular dispatch mode.
[0089] This embodiment, through accurate prediction and dynamic zoning, reduces the average waiting time during peak hours on floors 8-12 from 4.2 minutes to 1.8 minutes, reduces the elevator's ineffective mileage by 35%, and avoids the problem of passengers on other floors being excessively ignored, resulting in an overall satisfaction increase of 60%.
[0090] Example 3
[0091] Elevator potential fault warning scenario
[0092] Step 1: Multi-dimensional data collection
[0093] The system collects elevator operation data in real time:
[0094] Vibration sensor: Vertical vibration amplitude of the car is 0.12mm (normal range 0-0.1mm)
[0095] Temperature sensor: Motor temperature 58℃ (normal range 30-55℃)
[0096] Noise sensor: Operating noise 68dB (normal range 50-65dB)
[0097] Current sensor: Motor operating current 4.2A (normal range 3.5-4.0A)
[0098] The data is updated every 10 seconds and stored on edge computing nodes.
[0099] Step 2: Health Index Calculation
[0100] The system performs standardized scoring for each indicator:
[0101] Vibration score V_score = 75 (slightly above the limit)
[0102] Temperature score T_score = 70 (slightly above the limit)
[0103] Noise score N_score = 80 (slightly exceeds the limit)
[0104] The current score C_score is 78 (close to the upper limit).
[0105] The formula for calculating the health index is as follows:
[0106] H = 0.3 × 75 + 0.25 × 70 + 0.2 × 80 + 0.25 × 78 = 22.5 + 17.5 + 16 + 19.5 = 75. A health index of 75 is below the threshold of 80, so the system marks it as "mildly abnormal".
[0107] Step 3: Fault Early Warning and Handling
[0108] The system analysis revealed an abnormal pattern, which was likely caused by "motor bearing wear".
[0109] Generate an early warning report and push it to the maintenance personnel's mobile device:
[0110] Abnormal indicators and trend charts
[0111] Potential Fault Point Analysis
[0112] Recommended solutions
[0113] After the maintenance personnel confirm the appointment, an off-peak time (22:00 that evening) will be scheduled for the inspection.
[0114] Inspection revealed premature wear in the motor bearings, which were promptly replaced, preventing potential malfunctions.
[0115] This embodiment successfully issued an early warning 48 hours before the malfunction occurred through multi-dimensional monitoring and health index assessment, avoiding the serious consequences of elevator shutdown for more than 8 hours. At the same time, the maintenance was scheduled during non-operational hours, eliminating the impact on users.
[0116] Example 4
[0117] Adaptive maintenance scheduling scenario
[0118] Step 1: Maintenance Parameter Acquisition
[0119] The system retrieves relevant parameters for elevator number 3:
[0120] Current health index H = 82
[0121] The standard maintenance cycle M0 = 30 days
[0122] Usage intensity U = 0.75 (above average, due to serving high-floor units and high usage rate)
[0123] Use a strength coefficient λ = 0.2
[0124] Simultaneously, parameters of elevator No. 5 in the same building were obtained for comparison:
[0125] Health Index H = 95
[0126] Use strength U = 0.4
[0127] Other parameters are the same.
[0128] Step 2: Maintenance Cycle Calculation
[0129] Calculate the maintenance cycle for elevator No. 3:
[0130] M3 = 30 × (80 / 82) × (1 + 0.2 × 0.75) ≈ 30 × 0.9756 × 1.15 ≈ 30 × 1.122 ≈ 33.7 days
[0131] The system has adjusted the next maintenance cycle for elevator No. 3 to 34 days.
[0132] Calculate the maintenance cycle for elevator No. 5:
[0133] M5 = 30 × (80 / 95) × (1 + 0.2 × 0.4) ≈ 30 × 0.8421 × 1.08 ≈ 30 × 0.909 ≈ 27.3 days
[0134] The system has adjusted the next maintenance cycle for elevator No. 5 to 27 days.
[0135] Step 3: Maintenance Plan Optimization
[0136] The system integrates the maintenance plans for elevators No. 3 and No. 5, scheduling them on different dates within the same week.
[0137] Considering the building's event schedule, avoid the company's annual meeting period at the end of the month.
[0138] Maintenance reminders have been generated: Elevator No. 3 will undergo maintenance in 34 days (on a non-working day), and Elevator No. 5 will undergo maintenance in 27 days.
[0139] After maintenance is completed, the system records the maintenance content and effects, and updates the health index model parameters.
[0140] This embodiment, through adaptive adjustment, appropriately extended the maintenance cycle for elevator No. 3 (generally healthy but frequently used), avoiding over-maintenance; while shortening the maintenance cycle for elevator No. 5 (well-healthy but less frequently used), ensuring timely inspection of critical components. Overall maintenance costs were reduced by 18%, while the annual elevator availability increased to over 99.5%.
Claims
1. An intelligent elevator scheduling and maintenance early warning system, characterized in that, include: The data acquisition module is used to collect elevator operating parameters and passenger flow data; The intelligent scheduling module enables dynamic elevator scheduling based on passenger flow prediction. The health assessment module calculates the elevator health index through multi-sensor data fusion. The maintenance scheduling module dynamically adjusts the maintenance cycle based on health index and usage intensity. The emergency response module adjusts the elevator's operating priority in emergency situations.
2. The system according to claim 1, characterized in that, The intelligent scheduling module uses an LSTM neural network model to predict passenger flow and divides the elevators into different service groups based on the prediction results to achieve dynamic scheduling by zone.
3. The system according to claim 1, characterized in that, The health assessment module collects multi-dimensional data through vibration, temperature, noise, and current sensors to calculate the elevator health index H, using the following formula: H=α×V_score+β×T_score+γ×N_score+δ×C_score Where V_score is the vibration score, T_score is the temperature score, N_score is the noise score, C_score is the current score, and α, β, γ, and δ are weighting coefficients and their sum is 1.
4. The system according to claim 1, characterized in that, The maintenance scheduling module calculates the actual maintenance cycle according to the following formula: M = M0 × (k / H) × (1 + λ × U) Where M0 is the baseline maintenance cycle, k is a constant, H is the health index, λ is the usage intensity coefficient, and U is the usage intensity.
5. The system according to claim 1, characterized in that, The emergency response module can identify emergencies such as fires and earthquakes, as well as situations where passengers are trapped, automatically adjust the elevator's operating priority, and link with the building's emergency system.
6. A method for intelligent elevator scheduling and maintenance early warning, characterized in that, The following steps are involved: (1) Collect elevator operating parameters and passenger flow data; (2) Realize dynamic elevator scheduling based on traffic flow prediction; (3) Calculate the elevator health index by integrating data from multiple sensors; (4) Dynamically adjust the maintenance cycle based on health index and usage intensity; (5) Adjust elevator operation priority in emergency situations.
7. The method according to claim 6, characterized in that, In step (2), the dynamic scheduling of elevators based on traffic flow prediction includes: Predict passenger flow on each floor at different times using an LSTM model; Based on the prediction results, the elevators are divided into different service groups; Monitor traffic changes in real time and dynamically adjust scheduling strategies.
8. The method according to claim 6, characterized in that, In step (3), the calculation of the elevator health index includes: Standardize the vibration, temperature, noise, and current data; Calculate the scores V_score, T_score, N_score, and C_score for each indicator; The health index H is calculated using a weighted summation formula.
9. The method according to claim 6, characterized in that, In step (4), the dynamic adjustment of the maintenance cycle includes: Obtain the baseline maintenance cycle M0 and the current health index H; Calculate the elevator service intensity U; The actual maintenance cycle is determined by applying the maintenance cycle calculation formula; A maintenance plan is generated based on the actual maintenance cycle.
10. The method according to claim 6, characterized in that, In step (5), adjusting the elevator operation priority in an emergency includes: Identify the type and level of the emergency; Suspend non-essential elevator services; Prioritize dispatching elevators to the floor where the emergency occurred; In conjunction with the building's emergency response system, optimize evacuation routes.