A data monitoring and analysis method for an elevator energy feeding device

Abstract

This invention discloses a data monitoring and analysis method for an elevator power supply device, relating to the field of electrical data processing and analysis technology. The data monitoring method includes real-time acquisition of the power generation, power generation current, and temperature of the elevator power supply device; based on multiple preset current ranges and power generation durations, real-time recording of the number of times the power generation current persists for each duration within each range and the maximum current value, forming a multi-dimensional current event matrix; recording the number of overcurrent events in the first time period and the number of overtemperature events in the second time period, forming a multi-level protection matrix; and sending the multi-dimensional current event matrix, the multi-level protection matrix, and the cumulative power generation and output power to a cloud server in real time for analysis of the elevator and elevator power supply device's operating status. This embodiment provides data monitoring and analysis that integrates high-precision sampling, intelligent data aggregation, edge computing, and intelligent backend services.

Description

Technical Field

[0001] This invention relates to the field of electrical data processing and analysis technology, and in particular to a data monitoring and analysis method for an elevator energy supply device. Background Technology

[0002] In existing technologies, elevator energy recovery devices, as electrical equipment connected to the elevator system, primarily perform the single function of potential energy recovery. Most elevator energy recovery devices are not connected to a monitoring backend. Even those that are connected are only used to monitor their own operational status, providing data for equipment maintenance. However, the operational data of the elevator energy recovery device, such as bus voltage, feedback current, leakage current, and temperature, are also related to the elevator's operating status. Therefore, real-time and accurate monitoring of parameters such as current, voltage, and temperature of the elevator energy recovery device is crucial for assessing the operating status of the equipment (including the elevator energy recovery device and the elevator) and ensuring the safe and stable operation of the system.

[0003] However, traditional methods for monitoring and analyzing elevator energy supply device data typically have the following shortcomings: 1. Single sampling and monitoring dimension: It only focuses on instantaneous values ​​or simple average values, lacks detailed multi-dimensional records of current, and cannot accurately characterize load features.

[0004] 2. Simple protection and early warning logic: Overcurrent and overheat protection are usually based on fixed thresholds and simple delays, making it difficult to distinguish between short-term disturbances and real risks, and are prone to false triggering or failure to trigger.

[0005] 3. Weak data security mechanisms: Device operation authorization relies on fixed keys or simple protocols, which are at risk of being cracked and counterfeited.

[0006] 4. Insufficient edge computing capabilities: A large amount of raw data is directly uploaded to the monitoring backend, which puts pressure on communication bandwidth and the central server, and the real-time performance is poor.

[0007] 5. Insufficient ability to identify the operating status of equipment (including elevators and elevator power supply devices), and lack of effective methods to distinguish between normal and various abnormal states. Summary of the Invention

[0008] This invention provides a data monitoring and analysis method for an elevator energy supply device to solve at least one of the above-mentioned problems.

[0009] In a first aspect, embodiments of the present invention provide a data monitoring method for an elevator power supply device, applied to an elevator power supply device, the data monitoring method comprising: S110: Real-time acquisition of its own power generation, power generation current and temperature; S120. Based on multiple preset current ranges and multiple power generation durations, record in real time the number of times the power generation current lasts for each duration in each range and the maximum current value within a certain period of time, forming a multi-dimensional current event matrix. S130. Real-time monitoring of the number of times the generator current exceeds the current safety threshold, as the first count; real-time monitoring of the number of times the temperature exceeds the temperature safety threshold, as the second count; and taking the first count reaching the corresponding threshold as an overcurrent event, and the second count reaching the corresponding threshold as an overtemperature event; recording the number of overcurrent events in the first time period and the number of overtemperature events in the second time period to form a multi-level protection matrix, wherein the first time period is shorter than the second time period; S140. Calculate the cumulative power generation and power output within a certain period of time based on the power generation and power output current. S150. The multidimensional current event matrix, multi-level protection matrix, cumulative power generation and power output are sent to the cloud server in real time for the cloud server to analyze the operating status of the elevator and elevator power supply device.

[0010] Secondly, embodiments of the present invention provide a data analysis method for an elevator energy supply device, applied to a cloud server, the data analysis method comprising: By using a trained neural network model, the operating data of the elevator energy supply device is analyzed to identify the operating status of the elevator and the elevator energy supply device. The operational data includes: a multi-dimensional current event matrix and a multi-level protection matrix for the same time period, as well as cumulative power generation and power output; the multi-dimensional current event matrix includes the number of times the power generation current lasts for each power generation duration in each current range and the maximum current value; the multi-level protection matrix includes the number of overcurrent events in the first time period and the number of overtemperature events in the second time period, wherein the first time period is shorter than the second time period, and the number of times the power generation current exceeds the current safety threshold is considered an overcurrent event, and the number of times the temperature exceeds the temperature safety threshold is considered an overtemperature event. The operating status includes: normal status and various abnormal statuses; the normal status means that both the elevator and the elevator power supply device are normal, and the various abnormal statuses include elevator malfunction, elevator power supply device malfunction, and abnormal coordination between the elevator and the elevator power supply device.

[0011] Thirdly, embodiments of the present invention also provide a data monitoring and analysis system for an elevator energy supply device, comprising: The elevator energy feeding device is used to collect its own power generation, power generation current, and temperature in real time; according to multiple preset current ranges and power generation durations, it records in real time the number of times the power generation current lasts for each range and the maximum current value within a certain period of time, forming a multi-dimensional current event matrix; it monitors in real time the number of times the power generation current exceeds the current safety threshold, as the first count; it monitors in real time the number of times the temperature exceeds the temperature safety threshold, as the second count; it identifies the first count exceeding the first count threshold as an overcurrent event, and the second count exceeding the second count threshold as an overtemperature event; it records the number of overcurrent events in the first period of time and the number of overtemperature events in the second period of time, forming a multi-level protection matrix, wherein the first period of time is shorter than the second period of time; based on the power generation and power generation current, it calculates the cumulative power generation and power generation within a certain period of time in real time; and it sends the multi-dimensional current event matrix, the multi-level protection matrix, and the cumulative power generation and power generation to the cloud server in real time. The cloud server is used to analyze the operating data of the elevator power supply device using a trained neural network model, and to identify the operating status of the elevator and the elevator power supply device. The operating data includes: a multi-dimensional current event matrix and a multi-level protection matrix for the same time period, as well as the cumulative power generation and power output. The operating status includes: normal status and various abnormal statuses. The normal status means that both the elevator and the elevator power supply device are operating normally. The various abnormal statuses include elevator abnormality, elevator power supply device abnormality, and abnormal coordination between the elevator and the elevator power supply device.

[0012] In summary, this invention provides a data monitoring and analysis method for elevator power supply devices, integrating high-precision sampling, intelligent data aggregation, reliable and safe control, edge computing capabilities, and intelligent backend service capabilities. This method treats the elevator power supply device as an edge device and a cloud server as a cloud device. At the edge, through customized data structures and algorithms, it achieves refined statistics of electrical parameters, multi-level early warning, and protection. The multi-dimensional current event matrix enables refined multi-dimensional recording of current, accurately characterizing load features; the multi-level protection matrix effectively distinguishes between short-term disturbances and real risks, recording more realistic overcurrent and overheat protection data. These data structures and algorithms effectively reflect key information about device operation, not only reducing the amount of data uploaded, improving real-time performance, and alleviating the pressure on communication bandwidth and the cloud server, but also providing a more accurate data foundation for backend data analysis. The cloud-based backend then performs long-term analysis based on this data, learning operational data patterns, analyzing and judging the operating status of the elevator and elevator power supply device, and providing decision support for equipment monitoring and maintenance. Attached Figure Description

[0013] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0014] Figure 1 This is an architecture diagram of a data monitoring and analysis system for an elevator energy supply device provided in an embodiment of the present invention; Figure 2 This is a flowchart of a data monitoring method for an elevator energy supply device provided in an embodiment of the present invention; Figure 3 This is a flowchart of a data analysis method for an elevator energy feeding device provided in an embodiment of the present invention; Figure 4 This is a schematic diagram of the structure of a neural network model provided in an embodiment of the present invention; Figure 5 This is a flowchart of a data monitoring and analysis method for an elevator energy feeding device provided in an embodiment of the present invention; Figure 6 This is a flowchart of a dynamic authorization method provided in an embodiment of the present invention. Detailed Implementation

[0015] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0016] In the description of this invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0017] In the description of this invention, it should also be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0018] This embodiment provides a data monitoring and analysis method for elevator energy supply devices. To illustrate this method, the system architecture supporting its implementation will be introduced first. Figure 1 This is an architecture diagram of a data monitoring and analysis system for an elevator energy supply device provided in an embodiment of the present invention. Figure 1 As shown, the system includes an elevator power supply device and a cloud server (corresponding to the blue box part), while the rest is the original elevator system consisting of the elevator and the power grid.

[0019] In the original elevator system, the three-phase current supplied by the AC power grid is converted into DC power by rectifier unit D1, which powers the frequency converter Q1. The frequency converter powers motor M, which drives the elevator car (not shown in the diagram) to rise and fall. The frequency converter is connected to the control unit, which controls the elevator speed. Under the instructions of the control unit, the frequency converter adjusts the motor's frequency, torque, etc., thereby regulating the elevator speed. During elevator operation, motor M has two working states: one is the driving state, in which the car moves upward under the motor's drag, and the motor does work, converting electrical energy into potential energy; the other state is the generating state, in which the car (e.g., during descent) drives the motor to rotate, and the motor generates electricity, converting potential energy into electrical energy. The electrical energy is applied to both sides of the bus capacitor through the inverter, causing the voltage across the capacitor to rise. When the voltage exceeds the capacitor's capacity, the electrical energy is consumed through the braking resistor box R1 (equivalent to converting potential energy into heat energy).

[0020] In this embodiment, the elevator power supply device is connected to the busbar of the original elevator system, where the red busbar corresponds to the positive terminal P and the black busbar corresponds to the negative terminal N. After connecting the elevator power supply device, the electrical energy generated by the motor in generator mode will be fed back to the AC power grid through the elevator power supply device, reducing the power consumption of the braking resistor box. Specifically, the elevator power supply device can monitor the busbar voltage in real time. When the busbar voltage rises, it feeds the busbar electrical energy back to the AC power grid (for use by other equipment in the power grid), causing the busbar voltage to drop and not exceed the capacitor capacity, thereby reducing the energy consumption of the braking resistor.

[0021] Specifically, this embodiment treats the elevator power supply device as an edge device, deploying edge acquisition equipment and an edge computing module within it. The edge acquisition equipment may include electricity meters, ammeters, voltmeters, temperature sensors, etc., collecting data such as electricity and current fed back to the AC power grid from the elevator power supply device at a high frequency. The edge computing module may include computing devices such as MCUs (Microcontroller Units) to analyze and process the high-frequency data collected by the edge acquisition equipment, obtaining more meaningful operational data which is then uploaded to the cloud server.

[0022] Optionally, the elevator power supply device can transmit data to a cloud server via a DTU (Data Transfer Unit). The cloud server performs more in-depth analysis and processing of the received operational data to further monitor the operating status of the elevator and the elevator power supply device. Simultaneously, the cloud server can also issue operational control strategies to the elevator power supply device based on the analysis and processing results, ensuring the safe and efficient operation of the elevator power supply device.

[0023] Based on the above systems, Figure 2 This is a flowchart illustrating a data monitoring method for an elevator power supply device according to an embodiment of the present invention. This method can be executed by the elevator power supply device in the aforementioned system, or it can be executed independently by other electronic devices outside the aforementioned system. Figure 2 As shown, the method specifically includes: S110: Real-time acquisition of power generation, power generation current, and temperature of the elevator power supply device.

[0024] The data collected here is the elevator's own operational data, such as power generation (i.e., power supplied), power generation current (i.e., power supplied current), voltage, and temperature, which serve as the data source for the entire method. This data is collected very frequently, and the amount of data is substantial.

[0025] Existing technologies directly send the collected raw data to the backend. Due to the large data volume, the data received by the backend suffers from significant lag, substantial data loss during transmission, and discontinuous data. This embodiment deploys an edge computing module in the elevator power supply device. This module analyzes and processes the collected raw data, selectively uploading it. This reduces the amount of data transmitted while ensuring that critical information is not lost, thus improving the accuracy of subsequent analysis and monitoring. The analysis and processing process of the raw data by the edge computing module will be described in detail below through steps S120-S150.

[0026] S120. Based on multiple preset current ranges and multiple power generation durations, record in real time the number of times the power generation current lasts for each duration in each range and the maximum current value, forming a multi-dimensional current event matrix.

[0027] This step involves analyzing and processing the raw data to construct a multidimensional current event analysis model. Current events are recorded using a matrix (as shown in Table 1) with "current threshold - duration" as the dimension, in order to accurately characterize the load conditions of the elevator and the elevator power supply device.

[0028] Referring to Table 1, multiple current thresholds (e.g., 5A, 6A, ..., 30A, where A represents the current unit ampere) and multiple duration levels (e.g., 3s, 10s, 30s, where s represents the duration unit second) can be preset, forming a current range between each pair of adjacent current thresholds. By monitoring the current in real time, the frequency of events and the maximum current value falling within each "current range + duration level" combination are recorded, generating a three-dimensional data profile describing the load characteristics.

[0029] Since the duration of the current corresponds to the power generation period, each duration can also be called a power generation duration. In this embodiment, this matrix, with each current interval and each power generation duration as dimensions and the number of corresponding current events and the maximum current value as elements, is called a multidimensional current event matrix. This matrix can reflect the operating status of the elevator and elevator energy supply device, such as operating frequency, peak hours, and load size, providing data support for energy-saving effect prediction and equipment operation and maintenance (including elevator and elevator energy supply device operation and maintenance).

[0030] Table 1 Optionally, the specified time period can be 1 day, 1 hour, or other durations. Referring to Table 1, a multi-dimensional current event matrix can be generated for each time period (e.g., daily, hourly). Each column of the matrix corresponds to a current range; for example, the value 5 in the first column corresponds to the current range [0, 5A), the value 6 in the second column corresponds to the current range [5A, 6A), and so on. Each row of the matrix corresponds to the duration of a specific current level; 3s, 10s, and 30s correspond to short, medium, and long duration levels, respectively. The elements of the matrix are the number of times the current lasts for the corresponding duration within the current range and the maximum current value. Taking the element in the first row and first column as an example, whenever the current lasts for 3 seconds within the range [0, 5A) during the current time period, a current event corresponding to the element in the first row and first column is recorded. Finally, the total number of current events and the maximum current value among all current events are recorded at the element position in the first row and first column.

[0031] Furthermore, the current range and power generation duration in multidimensional current events can be determined manually based on experience, or by referring to relevant standards and the elevator's operating conditions. Optionally, if relevant standards stipulate that the power generation duration of an elevator for a set number of floors must not exceed a certain threshold, then multiple power generation durations can be preset based on the number of floors traversed in a single power generation by the elevator (which is also the power generation duration of the energy feeding system). For example, assuming the highest floor in the current building is F, and the standard requires that the power generation duration of the elevator traversing floor F not exceed 30 seconds, then three durations can be defined: 3 seconds, 10 seconds, and 30 seconds, corresponding to the durations traversing a small number of floors, a medium number of floors, and a maximum number of floors, respectively. Optionally, multiple current ranges can be preset based on the power generation current of the elevator energy feeding device under different loads and floors. The power generation current is related to both the elevator load and the number of floors traversed; the power generation current values ​​under different load and floor combinations can be pre-calculated to determine representative current thresholds for dividing the current ranges.

[0032] Furthermore, the multidimensional current event matrix contains important information related to the equipment's operating status. For example, this matrix can reflect the power generation and frequency of the power supply device; a decrease in power generation may indicate equipment malfunction. Simultaneously, the power generation frequency is related to the number of elevator movements over a period of time, providing information about equipment operating efficiency and lifespan. This is why this matrix is ​​constructed in this embodiment.

[0033] S130. Real-time monitoring of the number of times the generating current exceeds the current safety threshold, as the first count; real-time monitoring of the number of times the temperature exceeds the temperature safety threshold, as the second count; and taking the first count reaching the corresponding threshold as an overcurrent event, and the second count reaching the corresponding threshold as an overtemperature event; recording the number of overcurrent events in the first time period and the number of overtemperature events in the second time period to form a multi-level protection matrix, wherein the first time period is shorter than the second time period.

[0034] This step establishes a multi-level protection mechanism based on frequency accumulation through analysis and processing of raw data. For overcurrent and overheating events, a multi-level protection matrix is ​​proposed to record the frequency of events within different time windows (i.e., different durations), providing data support for equipment operation and protection.

[0035] In one specific implementation, the current safety threshold may include two thresholds: a current warning threshold and a current protection threshold, wherein the current protection threshold is greater than the current warning threshold. Similarly, the temperature safety threshold may also include two thresholds: a temperature warning threshold and a temperature protection threshold, wherein the temperature protection threshold is greater than the temperature warning threshold. Accordingly, the generation of the multi-level protection matrix may include the following steps: Step 1: In response to the first detection that the generated current exceeds the current warning threshold, overcurrent and overtemperature event monitoring is initiated. The current warning threshold corresponds to a safety boundary for the elevator power supply device. When the current exceeds this threshold, a safety warning may be issued. Therefore, this embodiment monitors the generated current exceeding the current warning threshold in real time. When the current exceeds the threshold for the first time within a certain period, the monitoring and recording of overcurrent and overtemperature events are initiated.

[0036] Step 2: Monitor the number of times the current exceeds the current warning threshold in real time. and the number of times the current exceeds the current protection threshold. Simultaneously, the system monitors in real time the number of times the temperature exceeds the aforementioned temperature warning threshold. and the number of times the temperature exceeds the current protection threshold. After initiating overcurrent and overtemperature event monitoring, this embodiment still defines overcurrent and overtemperature events by continuously monitoring the number of times the current and temperature exceed the safety boundaries. The recorded data... , , , This serves as the basis for judging subsequent overcurrent / overtemperature events.

[0037] Step 3, whenever Reaching the threshold Trigger an overcurrent warning event; whenever Reaching the threshold This triggers an overcurrent protection event, in which... The overcurrent warning threshold and overcurrent protection threshold correspond to overcurrent warning events and overcurrent protection events, respectively. Whenever... Accumulated to threshold When an overcurrent warning event is triggered and recorded, the recording will be completed. Reset to zero and begin counting the next overcurrent warning event. Overcurrent protection events are similar, but the difference is that the safety risk of the current protection threshold is greater than that of the current warning threshold; therefore, the threshold for the number of overcurrent protection events is [not specified]. Compare More stringent. Similarly, for over-temperature events, the over-temperature warning threshold and over-temperature protection threshold correspond to over-temperature warning events and over-temperature protection events, respectively, as well as different trigger thresholds: whenever... Reaching the threshold This triggers an over-temperature warning event; whenever Reaching the threshold This triggers an over-temperature protection event, in which... .

[0038] Step 4: Within a short time period starting from the startup time, record the number of overcurrent warning events and overcurrent protection events, and write them into the multi-level protection matrix; simultaneously, within a long time period starting from the startup time, record the number of overtemperature warning events and overtemperature protection events, and write them into the multi-level protection matrix. Since current changes are short-term changes, while temperature changes require long-term accumulation, this embodiment sets different time windows for recording overcurrent and overtemperature events. For ease of distinction and description, the short time period can be referred to as the first time period, and the long time period as the second time period. Record the number of overcurrent warning events and overcurrent protection events in the first time period, and the number of overtemperature warning events and overtemperature protection events in the second time period, and write these numbers into the multi-level protection matrix.

[0039] Table 2 shows an example of a multi-level protection matrix, in which the warning current threshold and the protection current threshold are 25A and 35A, respectively, the warning temperature threshold and the protection temperature threshold are 55℃ and 75℃, respectively, and the first time period and the second time period are 2s and 30s, respectively.

[0040] Table 2 Referring to Table 2, when the current exceeds the current warning threshold of 25A for the first time within a certain period, it indicates that an overcurrent or overtemperature event is about to occur, and overcurrent and overtemperature event monitoring is initiated. Then, within a short 2-second window starting from the initiation time, the number of times the current exceeds 25A is monitored. When this number reaches a certain threshold (e.g., 10 times), an overcurrent warning event record is triggered, and the matrix element corresponding to 25A is incremented by 1 (10000 in the table is the maximum upper limit of this matrix element). The count is then reset to zero, and the accumulation restarts, waiting to trigger the next overcurrent warning event record. Simultaneously, within the same 2-second window, the number of times the current exceeds 35A is monitored. When this number reaches a certain threshold (e.g., 4 times), an overcurrent protection event record is triggered, and the matrix element corresponding to 35A is incremented by 1 (10000 in the table is the maximum upper limit of this matrix element). The count is then reset to zero, and the accumulation restarts, waiting to trigger the next overcurrent protection event record. It can be seen that the record triggering condition corresponding to the protection threshold (threshold 4 times) is more stringent.

[0041] Over-temperature warning events and over-temperature protection events are similar, used for temperature warning and temperature protection respectively, except that the cumulative counting time window is longer (e.g., 30 seconds). For ease of distinction and description, this embodiment refers to the number of times the generator current exceeds the current safety threshold as the "first count" (in the above embodiment). , All are the first counts), and the number of times the temperature exceeds the temperature safety threshold is referred to as the second count (in the above embodiments). , (All are the second number).

[0042] This multi-level recording method, which accumulates data based on frequency, effectively filters out noise interference and improves the accuracy of protection warnings. Simultaneously, this data also contains information related to the equipment's operating status, which can be used to assess the equipment's safe operating condition and provide data for setting operating parameters and evaluating equipment maintenance. For example, under normal operating conditions, the current safety boundary condition will not be triggered. The triggering frequency can indicate the power matching degree between the energy feeding equipment and the elevator. If the triggering frequency is relatively high, and there is a short-term increase in bus voltage, it may correspond to incomplete energy recovery or insufficient power of the energy feeding device, requiring replacement with a higher-power energy feeding device. Temperature reflects the long-term operating status of the equipment. If the equipment temperature is close to the upper limit for a long period, it indicates that the equipment is constantly operating under overload conditions, which may indicate insufficient power, requiring replacement with a higher-power model; it may also indicate a high ambient temperature around the equipment, such as a malfunctioning elevator room exhaust fan or a broken air conditioner, requiring improved ventilation conditions in the elevator room. This is why this embodiment records and uploads this data.

[0043] S140. Calculate the cumulative power generation and power output within a certain period of time based on the power generation and power output current.

[0044] In addition to the multidimensional current event matrix and the multi-level protection matrix, the cumulative power generation and average power generation over a certain period are also important operational data. Power generation can be obtained and accumulated through direct energy meters, while power generation can be calculated by multiplying current and voltage. Optionally, the certain period can be 1 day, 1 hour, or other durations; the cumulative power generation and average power generation can be calculated once for each period (e.g., daily, hourly).

[0045] Power generation and output also contain important information related to the equipment's operating status. For example, among equipment in the same area, of the same model, and operating conditions, equipment with significantly lower power generation may be malfunctioning; and for the same elevator at different times, the operating efficiency is lower during periods with significantly lower power generation, and so on. Therefore, these are also indispensable data in this embodiment.

[0046] S150. The multidimensional current event matrix, multi-level protection matrix, cumulative power generation and power output are sent to the cloud server in real time for the cloud server to analyze the operating status of the elevator and elevator power supply device.

[0047] By analyzing and processing the raw data at the edge, effective information such as multi-dimensional current event matrix, multi-level protection matrix, cumulative power generation and power output are obtained and uploaded to the cloud server in a timely manner for the cloud server to conduct more in-depth and long-term data analysis.

[0048] Optionally, the elevator power supply device can package multi-dimensional statistical data, protection action count, power supply and power consumption event count (e.g., a current greater than 0.1A for 0.1 seconds is considered a valid power supply pulse) into a fixed time interval (e.g., 600 seconds) and send the packaged data in real time.

[0049] It is worth emphasizing that the above data analysis and recording are completed inside the elevator energy supply device. The elevator energy supply device only needs to transmit the analyzed and processed data to the backend cloud server, without having to upload a large amount of raw data. This ensures the accuracy of the data and the collection interval, while reducing the amount of data transmission and saving network bandwidth and server storage space.

[0050] Figure 3 This is a flowchart illustrating a data analysis method for an elevator energy supply device according to an embodiment of the present invention. This method can be executed by a cloud server within the aforementioned system, or it can be executed independently by other electronic devices outside the aforementioned system. Figure 3 As shown, the method specifically includes: S210. Using the trained neural network model, analyze the operating data of the elevator energy supply device to identify the operating status of the elevator and the elevator energy supply device.

[0051] The operational data includes: a multi-dimensional current event matrix and a multi-level protection matrix for the same time period, as well as cumulative power generation and power output.

[0052] The operating status includes: normal status and multiple abnormal statuses; the normal status means that both the elevator and the elevator power supply device are normal, and the multiple abnormal statuses include at least two of the following: elevator malfunction, elevator power supply device malfunction, and abnormal coordination between the elevator and the elevator power supply device.

[0053] Furthermore, each abnormal state can be further subdivided. For example, elevator abnormalities can be subdivided into: improper counterweight balance coefficient (the elevator operates unbalancedly for extended periods, the traction machine performs extra work to overcome unnecessary weight differences, resulting in significantly increased power consumption during unloaded / lightly loaded downward movement or heavily loaded upward movement), track problems (improper guide rail installation, worn guide shoes, etc., leading to high running resistance, slippage or excessive tension of the wire rope / belt, and low transmission efficiency), poor lubrication system (lack of lubricant in key components such as the main unit, bearings, and guide wheels, increasing friction), and unreasonable control system (unreasonable elevator operating curve settings, poor sensing system, long door opening and closing times, unreasonable elevator control system logic, etc.). Regardless of the abnormal state classification method used, the abnormal states to be identified in this embodiment include at least two abnormality types.

[0054] More complex data monitoring and analysis algorithms are deployed within the cloud server to analyze and identify the operating status of elevators and elevator energy supply devices. Optionally, this embodiment provides a neural network model to serve as the algorithm. The model input is multi-dimensional operating data of the elevator energy supply device during the same period, and the model output is the operating status of the elevator and elevator energy supply device during that period. The model prediction results will be provided to the back-end maintenance personnel for further verification and analysis.

[0055] In one specific implementation, the neural network model can employ, as follows: Figure 4 The structure shown includes a feature preprocessing layer, a basic feature extraction layer, a normal identification branch, and an anomaly type identification branch. The data processing in the model may include the following steps: Step 1: Denote the multi-dimensional current event matrix for the same time period as A1, the multi-level protection matrix as A2, and the cumulative power generation and output as A3. Then, input A1, A2, and A3 into their respective feature preprocessing layers to process them into their corresponding feature vectors. The main function of feature preprocessing is to perform data preprocessing and dimensionality transformation, balancing the data volume of each feature vector and preventing low-dimensional operational data (such as A3) from being overwhelmed in subsequent operations. For example, A1 is a multi-dimensional matrix and can use a CNN-structured feature preprocessing layer; A2 and A3 can use a fully connected (FC) structured feature preprocessing layer, keeping the dimensionality of the three preprocessed feature vectors within a set range.

[0056] Step 2: Concatenate the preprocessed feature vectors into a single vector and input it into the basic feature extraction layer to extract fundamental electrical features from the data. Specifically, the main function of the basic feature extraction layer is to extract shallow features, learn clean and stable basic signals and preliminary data patterns, such as the preliminary correlation between current, temperature, and power. At this stage, the basic electrical features mainly include basic data characteristics and cannot yet directly distinguish categories. Optionally, the basic electrical feature extraction layer can adopt structures such as convolutional networks, transformers, or fully connected (FC), and this embodiment does not impose specific limitations.

[0057] Step 3: Input the basic electrical features into the normal recognition branch, extract a deep feature, and identify whether the same time period is in a normal or abnormal state based on this deep feature. Optionally, combine... Figure 4The normal identification branch can include a deep feature extraction layer and a prediction layer. The deep feature extraction layer is used to further process the basic electrical features and extract deep features that can distinguish between normal and abnormal states. The prediction layer is used to output the probability that the current running data belongs to a normal or abnormal state based on the deep features. Optionally, the normal identification branch can adopt an MLP (Multilayer Perceptron) or other structures, and finally output a two-dimensional probability vector.

[0058] Step 4: If the probability of a normal state is high, it indicates that the device is in a normal state, and the identification ends. If the probability of an abnormal state is high, the basic electrical features are input into the abnormal type identification branch to extract another deep feature, and the abnormal state type is identified based on this other deep feature. Optionally, combined with... Figure 4 The anomaly type identification branch can also include a deep feature extraction layer and a prediction layer. The deep feature extraction layer is used to further process the basic electrical features of the abnormal state data to extract deep features that can distinguish various anomaly types. The prediction layer is used to output the probability that the current running data belongs to various anomaly types based on the deep features. Optionally, the anomaly type identification branch can also adopt an MLP or other structures, and finally output a K-dimensional probability vector, where K is the number of anomaly types.

[0059] For ease of distinction and description, this embodiment refers to the features extracted from the normal identification branch as the first deep feature, and the deep feature extraction layer and prediction layer in the normal identification branch as the first deep feature extraction layer and the first prediction layer, respectively; while the features extracted from the anomaly identification branch are referred to as the second deep feature, and the deep feature extraction layer and prediction layer in the anomaly identification branch are referred to as the second deep feature extraction layer and the second prediction layer, respectively.

[0060] Specifically, the second deep feature in this embodiment removes common features among anomaly types, enabling a more explicit representation of the differential features between anomaly types. For example, various anomaly types exhibit common phenomena such as overcurrent, overtemperature, and deviation from normal, but the specific patterns of overcurrent, overtemperature, and deviation from normal differ for each anomaly type, and may even include other differential features beyond the common phenomena. Common features are the representation of these common phenomena in the feature space. After removing common features, the differential features of each anomaly type become more apparent, making it easier to distinguish between different anomaly types. This is particularly suitable for situations in this embodiment where samples of each anomaly type are scarce and the original differential features are not obvious.

[0061] Furthermore, to achieve the above objectives, the neural network model can be trained in the following manner: Step 1: Pre-collect operational data of the elevator power supply device under various operating conditions to form a sample set. Each sample includes operational data of the elevator power supply device during the same time period and is labeled with its operating status.

[0062] Step 2: Pass each sample through the basic feature extraction layer and the normal recognition branch in sequence. By constraining the minimization of the first deep feature difference of normal samples, the convergence of the first deep features of abnormal samples, and the maximization of the first deep feature difference between normal and abnormal samples, the basic feature extraction layer and the normal recognition branch are trained so that the normal recognition branch can distinguish between normal and abnormal states.

[0063] Combination Figure 4 This step of the training Figure 4 In the first path of the model, each sample sequentially passes through a feature preprocessing layer, a basic feature extraction layer, and a normal identification branch, outputting a probability vector indicating whether each sample belongs to the normal or abnormal category. Since the abnormal states in this embodiment are diverse and the samples are imbalanced (many normal samples, few samples of each abnormal type), it is difficult to extract features from both normal samples and samples of each abnormal type simultaneously through a unified feature extraction branch. This can easily lead to slow model convergence and difficulty in effectively distinguishing each abnormal type. Meanwhile, considering that normal samples are numerous and easiest to identify, this embodiment first uses a separate normal identification branch to extract the essential differences between normal and abnormal samples, enabling the normal identification branch to distinguish between normal and abnormal states. In this stage of training, all samples of abnormal types are considered abnormal samples, increasing the number of samples and inter-class balance, accelerating model convergence, and promoting accurate identification.

[0064] It is worth mentioning that, in practical applications, the feature preprocessing layer can also be removed from or merged into part of the basic feature extraction layer. Therefore, the model training method of this embodiment can be applied to... Figure 4 When the neural network model shown includes a feature preprocessing layer, the basic feature extraction layer mentioned in the training method includes the feature preprocessing layer by default (unless otherwise specified).

[0065] Optionally, the parameters of the basic feature extraction layer and the normal recognition branch can be updated using the following loss function: (1) in, This represents the value of the loss function. Represents the sample set, This represents the normal sample set. Indicates an abnormal sample set, and Indicates the sample index; Indicates sample The running status label, in In the middle, normal samples Abnormal samples ; This represents samples that correctly identify branch predictions. The probability of being in a normal state; and Representing samples respectively and samples The first depth feature, express and Similarity; , , and For weight parameters, , , and It can also include an operation of averaging the number of samples to keep the magnitude of each loss term balanced.

[0066] pass Minimizing the updates of the basic feature extraction layer and the normal recognition branch parameters enables the normal recognition branch to extract the essential differential features between normal and abnormal samples, thereby enabling it to distinguish between normal and abnormal samples. Among these, This is used to constrain the accuracy of identifying normal and abnormal samples (the cross-entropy loss function is used here, but it can also be replaced with other supervised loss functions that have the same effect). This is used to constrain the minimization of the first depth feature difference among all normal samples. This is used to constrain the minimization of the first depth feature difference for each outlier sample. This is used to constrain the maximization of the first deep feature difference between normal and abnormal samples. Optionally, to extract common features of abnormal states, the depth can be increased. The value of makes the first depth features of the abnormal samples tend to be consistent.

[0067] Step 3: After training, take the center of the final first depth feature of each abnormal sample as the common feature of the abnormal state. After training in Step 2, the first depth features of all abnormal samples are sufficiently similar, and these features correspond to the common manifestations of all abnormal states. Take the center of these features and use them as the common features of the abnormal states.

[0068] Step 4: Input each abnormal sample into the trained basic feature extraction layer and the untrained abnormal type recognition branch. By constraining the second deep features of each abnormal sample to be orthogonal to the common features, minimizing the difference in the second deep features of samples of the same abnormal type, and maximizing the difference in the second deep features between samples of different abnormal types, the abnormal type recognition branch is trained so that it can distinguish between different abnormal types.

[0069] Combination Figure 4 This step of the training Figure 4 The second path can directly extract the final basic electrical features of each abnormal sample in step two, input them into the abnormal type identification branch, and output the probability vector of each abnormal sample belonging to each abnormal type; or a new batch of abnormal samples can be passed through a trained feature preprocessing layer, a trained basic feature extraction layer, and an untrained abnormal type identification branch (without passing through the normal identification branch) in sequence, and output the probability vector of each abnormal sample belonging to each abnormal type.

[0070] Because the number of samples for each anomaly type is small, the model struggles to fully learn the features of each anomaly type with limited samples, increasing the difficulty of anomaly type identification. Therefore, this step removes (or suppresses) the common features among anomaly samples in the anomaly type identification branch, which is equivalent to amplifying the differential features between each anomaly type. This makes it more likely that the model can quickly learn the differences between different anomaly types with limited samples, improving the model's convergence speed and identification ability. At the same time, this step excludes normal samples from the identification range, which reduces the types of samples to be identified and the interference from large samples, further reducing the difficulty of identification.

[0071] Optionally, the parameters of the basic feature extraction layer can be fixed, and the parameters of the anomaly type recognition branch can be updated using the following loss function: (2) in, This represents the value of the loss function. , and All are indexes of exception types. Indicates the exception type The sample set, Indicates sample Status labels; For the sample When it is a real exception type, ; Not a sample When it is a real exception type, ; Samples representing the branch prediction for anomaly type identification Belongs to the exception type The probability of; and Representing samples respectively and samples The second depth feature, express and Similarity; Common characteristics of abnormal states Transpose of; , , and For weight parameters, 、 、 and It can also include the operation of averaging the number of samples.

[0072] pass Minimizing the update parameters of the anomaly type identification branch allows the second-depth features of each anomaly type to remove common features of the anomaly state, and to better and faster distinguish anomaly types through differential components. Among these, This is used to constrain the accuracy of anomaly type identification (the identified type tends to be consistent with the true type; the cross-entropy loss function is still used here, but it can also be replaced with other supervised loss functions with the same effect). The common features of abnormal states were removed from the second feature to constrain it. This is used to constrain the minimization of the second-depth feature differences among samples of the same anomaly type. This is used to constrain the maximization of the differences in second-depth features between samples of different anomaly types.

[0073] Furthermore, the basic feature extraction layer and normal identification branch can be pre-trained in step two. Then, each sample (whether normal or abnormal) is input into the pre-trained basic feature extraction layer to obtain the basic electrical features of each sample. The basic electrical features of each normal sample are then input into the pre-trained normal identification branch, while the basic electrical features of each abnormal sample are simultaneously input into both the pre-trained normal identification branch and the untrained abnormal identification branch. By constraining the minimization of the first deep feature difference of normal samples, the convergence of the first deep features of abnormal samples, the maximization of the first deep feature difference between normal and abnormal samples, the minimization of the second deep feature difference of samples of the same abnormal type, the maximization of the second deep feature difference between samples of different abnormal types, and the orthogonality of the second deep features of each abnormal sample to the center of the first deep features of all abnormal samples, the parameters of the basic feature extraction layer, the normal identification branch, and the abnormal identification branch are simultaneously updated. This ensures that the normal identification branch can distinguish between normal and abnormal states, while the abnormal identification branch can also distinguish between different abnormal types. Optionally, a joint loss function can be used. The parameters of the basic feature extraction layer, normal identification branch, and abnormal identification branch are updated. Among them, the parameter update of the abnormal identification branch is larger, while the parameters of the basic feature extraction layer and normal identification branch are fine-tuned. Then take the center of the first depth feature of all abnormal samples in the same batch.

[0074] Figure 5 This is a flowchart of a data monitoring and analysis method for an elevator energy supply device provided in an embodiment of the present invention. The method can be executed by the elevator energy supply device and the cloud server in the above system, and specifically includes the following steps: S10. The elevator power supply device collects its own power generation, power generation current, and temperature in real time; based on multiple preset current ranges and multiple power generation durations, it records in real time the number of times the power generation current lasts for each duration in each range and the maximum current value within a certain period of time, forming a multi-dimensional current event matrix; it monitors in real time the number of times the power generation current exceeds the current safety threshold, as the first count; it monitors in real time the number of times the temperature exceeds the temperature safety threshold, as the second count; and it treats the first count reaching the corresponding threshold as an overcurrent event and the second count reaching the corresponding threshold as an overtemperature event; it records the number of overcurrent events in the first period of time and the number of overtemperature events in the second period of time, forming a multi-level protection matrix, wherein the first period of time is shorter than the second period of time; based on the power generation and power generation current, it calculates the cumulative power generation and power generation within a certain period of time in real time; and it sends the multi-dimensional current event matrix, the multi-level protection matrix, and the cumulative power generation and power generation to the cloud server in real time.

[0075] S20. The cloud server uses a trained neural network model to analyze the operating data of the elevator power supply device and identify the operating status of the elevator and the elevator power supply device. The operating data includes: a multi-dimensional current event matrix and a multi-level protection matrix for the same time period, as well as the cumulative power generation and power output. The operating status includes: normal status and various abnormal statuses. The normal status means that both the elevator and the elevator power supply device are normal. The various abnormal statuses include elevator abnormality, elevator power supply device abnormality, and abnormal coordination between the elevator and the elevator power supply device.

[0076] The method in this embodiment is based on the same inventive concept as the data monitoring method and data analysis method described above. Any limitations in the embodiments of the data monitoring method and data analysis method described above are applicable to this embodiment, and will not be repeated here.

[0077] Furthermore, in another specific implementation, a "one device, one code" dynamic security authorization method is proposed, which improves the encryption method based on fixed feature codes and realizes random feature codes and dynamic authorization. Optionally, a unique random feature code is written into the elevator energy supply device at the factory, one code per device. When the cloud server authorizes (authorizes the elevator energy supply device to upload operating data), it first remotely reads the feature code, then combines it with the current authorization time information, uses an encryption algorithm to generate a dynamic key, and sends it to the elevator energy supply device. The elevator energy supply device uses the same algorithm and its own feature code for verification. After successful verification, it performs data monitoring and uploading. Authorization needs to be updated every hour. If the timeout or no authorization is obtained, data monitoring and uploading will stop (DTU communication is maintained). The specific process is as follows: Figure 6 As shown, this method binds the device (elevator super-powered device), time, and authorization commands, effectively preventing message attacks and device spoofing.

[0078] Reference Figure 6 The above method specifically includes the following steps: After the elevator power supply device is powered on, it sends a heartbeat packet to the cloud server. The heartbeat packet contains a unique random feature code of the elevator power supply device (i.e., Figure 6 (The random number in the packet). After receiving the heartbeat packet, the cloud server first checks if the elevator power supply device exists in the cloud server's internal device list. If it exists, it extracts the unique random signature from the heartbeat packet, encrypts the signature and the current time information (i.e., the authorized time) according to the encryption algorithm, generates a key sequence, and sends the key sequence to the elevator power supply device. If the elevator power supply device does not exist in the cloud server's internal device list, it first registers the elevator power supply device, and then performs the above operations of extracting the signature, generating the key sequence, and sending it.

[0079] After receiving the key sequence, the elevator energy supply device calculates the key value (including authorization time and random signature) using the same encryption algorithm as the cloud server, and compares the random signature in the key value with its own stored unique random signature. If the comparison does not match (i.e., Figure 6 If the key in the data is incorrect, authorization fails, the elevator power supply device stops executing the S110-S150 data monitoring method, and the cloud service will not receive data from the edge. If the comparison is consistent (i.e., ...), the authorization fails. Figure 6 If the key in the data is correct, the authorization is valid. The elevator power supply device updates the authorization time and executes the data monitoring methods of S110-S150 to continuously collect and send data to the cloud server. If the authorization time limit exceeds 1 hour, the authorization flag is cleared and a heartbeat packet is sent to the cloud server again.

[0080] After receiving a new heartbeat packet, the cloud server re-executes the operations of checking the existence of the elevator power supply device, extracting the feature code, generating a key sequence, and sending it to re-authorize. This process is repeated to achieve dynamic authorization.

[0081] like Figure 1 As shown, this embodiment of the invention also provides a data monitoring and analysis system for an elevator energy supply device, including: The elevator energy feeding device is used to collect its own power generation, power generation current, and temperature in real time; according to multiple preset current ranges and power generation durations, it records in real time the number of times the power generation current lasts for each range and the maximum current value within a certain period of time, forming a multi-dimensional current event matrix; it monitors in real time the number of times the power generation current exceeds the current safety threshold, as the first count; it monitors in real time the number of times the temperature exceeds the temperature safety threshold, as the second count; it identifies the first count exceeding the first count threshold as an overcurrent event, and the second count exceeding the second count threshold as an overtemperature event; it records the number of overcurrent events in the first period of time and the number of overtemperature events in the second period of time, forming a multi-level protection matrix, wherein the first period of time is shorter than the second period of time; based on the power generation and power generation current, it calculates the cumulative power generation and power generation within a certain period of time in real time; and it sends the multi-dimensional current event matrix, the multi-level protection matrix, and the cumulative power generation and power generation to the cloud server in real time. The cloud server is used to analyze the operating data of the elevator power supply device using a trained neural network model, and to identify the operating status of the elevator and the elevator power supply device. The operating data includes: a multi-dimensional current event matrix and a multi-level protection matrix for the same time period, as well as the cumulative power generation and power output. The operating status includes: normal status and various abnormal statuses. The normal status means that both the elevator and the elevator power supply device are operating normally. The various abnormal statuses include elevator abnormality, elevator power supply device abnormality, and abnormal coordination between the elevator and the elevator power supply device.

[0082] Furthermore, in another specific embodiment, the data monitoring and analysis system for the aforementioned elevator energy supply device is also used to execute the following "one device, one code" dynamic security authorization method. This method improves upon the encryption method based on fixed feature codes, realizing random feature codes and dynamic authorization. Optionally, a unique random feature code is written into the elevator energy supply device at the factory, one code per device. When the cloud server authorizes (authorizes the elevator energy supply device to upload operating data), it first remotely reads the feature code, then combines it with the current authorization time information, uses an encryption algorithm to generate a dynamic key, and sends it to the elevator energy supply device. The elevator energy supply device uses the same algorithm and its own feature code for verification. After successful verification, data monitoring and uploading are performed. Authorization needs to be updated every hour. If the timeout or no authorization is obtained, data monitoring and uploading will stop (DTU communication is maintained). The specific process is as follows: Figure 6 As shown, this method binds the device (elevator super-powered device), time, and authorization commands, effectively preventing message attacks and device spoofing.

[0083] Reference Figure 6 The aforementioned dynamic security authorization method specifically includes the following steps: After the elevator power supply device is powered on, it sends a heartbeat packet to the cloud server. The heartbeat packet contains a unique random signature of the elevator power supply device (i.e., Figure 6 (The random number in the packet). After receiving the heartbeat packet, the cloud server first checks if the elevator power supply device exists in the cloud server's internal device list. If it exists, it extracts the unique random signature from the heartbeat packet, encrypts the signature and the current time information (i.e., the authorized time) according to the encryption algorithm, generates a key sequence, and sends the key sequence to the elevator power supply device. If the elevator power supply device does not exist in the cloud server's internal device list, it first registers the elevator power supply device, and then performs the above operations of extracting the signature, generating the key sequence, and sending it.

[0084] After receiving the key sequence, the elevator energy supply device calculates the key value (including authorization time and random signature) using the same encryption algorithm as the cloud server, and compares the random signature in the key value with its own stored unique random signature. If the comparison does not match (i.e., Figure 6 If the key in the data is incorrect, authorization fails, the elevator power supply device stops executing the S110-S150 data monitoring method, and the cloud service will not receive data from the edge. If the comparison is consistent (i.e., ...), the authorization fails. Figure 6 If the key in the data is correct, the authorization is valid. The elevator power supply device updates the authorization time and executes the data monitoring methods of S110-S150 to continuously collect and send data to the cloud server. If the authorization time limit exceeds 1 hour, the authorization flag is cleared and a heartbeat packet is sent to the cloud server again.

[0085] After receiving a new heartbeat packet, the cloud server re-executes the operations of checking the existence of the elevator power supply device, extracting the feature code, generating a key sequence, and sending it to re-authorize. This process is repeated to achieve dynamic authorization.

[0086] It should be noted that the system in this embodiment is based on the same inventive concept as the data monitoring method and data analysis method described above. Any limitations in the embodiments of the data monitoring method and data analysis method described above are applicable to this embodiment, and will not be repeated here.

[0087] In summary, this invention provides a data monitoring and analysis method and system for elevator power supply devices, integrating high-precision sampling, intelligent data aggregation, reliable and safe control, edge computing capabilities, and intelligent backend service capabilities. This method treats the elevator power supply device as an edge device and a cloud server as a cloud device. At the edge, through customized data structures and algorithms, it achieves refined statistics of electrical parameters, multi-level early warning, and protection. The multi-dimensional current event matrix provides refined multi-dimensional recording of current, accurately characterizing load features; the multi-level protection matrix effectively distinguishes between short-term disturbances and real risks, recording more realistic overcurrent and overheat protection data. These data structures and algorithms more effectively reflect key information about device operation, reducing the amount of data uploaded from the edge to the backend, improving real-time performance, alleviating bandwidth pressure, and providing a more accurate data foundation for backend data analysis, thus improving the backend's ability to analyze and identify the device's operating status. Furthermore, this embodiment also enhances the security, reliability, and intelligence level of data monitoring through a time-bound dynamic encryption authorization mechanism.

[0088] Meanwhile, this embodiment performs long-term data analysis in the cloud to learn the patterns in the operational data and analyze and judge the operating status of elevators and elevator power supply devices. Specifically, considering the characteristics of numerous types of abnormal equipment states and a small sample size, this embodiment proposes a state recognition network model including a basic feature extraction layer, a normal identification branch, and an abnormal type identification branch. The basic feature extraction layer extracts the basic electrical features of the operational data; the normal identification branch extracts the common features of normal samples and the common features of abnormal samples, thereby learning the essential differences between normal and abnormal samples and effectively distinguishing between normal and abnormal states; then, the abnormal type identification branch removes the common features of abnormal samples and quickly learns the differentiated type features of abnormal samples, thereby quickly distinguishing different abnormal types. This model structure and training method improve the ability to identify the operating status of equipment (including elevators and elevator power supply devices), enabling rapid and effective differentiation between normal states and various abnormal states, providing effective data support for equipment monitoring and maintenance.

[0089] It should be noted that the equipment operation data and other user data involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of the relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and corresponding operation portals are provided for users to choose to authorize or refuse.

[0090] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the technical solutions of the embodiments of the present invention.

Claims

1. A data monitoring method for an elevator energy supply device, characterized in that, The data monitoring method, applied to elevator energy supply devices, includes: S110: Real-time acquisition of its own power generation, power generation current and temperature; S120. Based on multiple preset current ranges and multiple power generation durations, record in real time the number of times the power generation current lasts for each duration in each range and the maximum current value within a certain period of time, forming a multi-dimensional current event matrix. S130. Real-time monitoring of the number of times the generator current exceeds the current safety threshold, as the first count; real-time monitoring of the number of times the temperature exceeds the temperature safety threshold, as the second count; and taking the first count reaching the corresponding threshold as an overcurrent event, and the second count reaching the corresponding threshold as an overtemperature event; recording the number of overcurrent events in the first time period and the number of overtemperature events in the second time period to form a multi-level protection matrix, wherein the first time period is shorter than the second time period; S140. Calculate the cumulative power generation and power output within a certain period of time based on the power generation and power output current. S150. The multidimensional current event matrix, multi-level protection matrix, cumulative power generation and power output are sent to the cloud server in real time for the cloud server to analyze the operating status of the elevator and elevator power supply device.

2. The data monitoring method according to claim 1, characterized in that, Prior to S120, it also included: Multiple power generation durations are preset based on the number of floors the elevator passes through in a single power generation cycle of the elevator power supply device. Multiple current ranges are preset based on the power generation current of the elevator energy supply device under different loads and floors.

3. The data monitoring method according to claim 1, characterized in that, The current safety threshold includes a current warning threshold and a current protection threshold, wherein the current protection threshold is greater than the current warning threshold; Accordingly, S130 includes: In response to the first detection that the power generation current exceeds the current warning threshold, overcurrent event and overtemperature event monitoring are initiated. Real-time monitoring of the number of times the current exceeds the current warning threshold and the number of times the power generation current exceeds the current protection threshold. ; whenever Reaching the threshold Trigger an overcurrent warning event; whenever Reaching the threshold This triggers an overcurrent protection event; among which, ; During the first time period starting from the start time, record the number of overcurrent warning events and overcurrent protection events, and write them into the multi-level protection matrix; During the second time period starting from the aforementioned start time, the number of over-temperature events is recorded and written into the multi-level protection matrix.

4. The data monitoring method according to claim 1, characterized in that, Also includes: In response to the lack of data monitoring authorization or the authorization exceeding the time limit, a random feature code is fed back to the cloud server for the cloud server to generate a key sequence; In response to the key sequence of the cloud server, a key value is calculated using an encryption algorithm, and the key value is compared with the random feature code; If the comparison is inconsistent, authorization fails, and data monitoring operations on S110-S150 are stopped; If the comparison matches, the authorization is successful. Perform data monitoring operations S110-S150 and monitor whether the authorization has exceeded the time limit.

5. A data analysis method for an elevator energy feeding device, characterized in that, Applied to cloud servers, the data analysis method includes: By using a trained neural network model, the operating data of the elevator energy supply device is analyzed to identify the operating status of the elevator and the elevator energy supply device. The operational data includes: a multi-dimensional current event matrix and a multi-level protection matrix for the same time period, as well as cumulative power generation and power output; the multi-dimensional current event matrix includes the number of times the power generation current lasts for each power generation duration in each current range and the maximum current value; the multi-level protection matrix includes the number of overcurrent events in the first time period and the number of overtemperature events in the second time period, wherein the first time period is shorter than the second time period, and the number of times the power generation current exceeds the current safety threshold is considered an overcurrent event, and the number of times the temperature exceeds the temperature safety threshold is considered an overtemperature event. The operating status includes: normal status and various abnormal statuses; the normal status means that both the elevator and the elevator power supply device are normal, and the various abnormal statuses include elevator malfunction, elevator power supply device malfunction, and abnormal coordination between the elevator and the elevator power supply device.

6. The data analysis method according to claim 5, characterized in that, The neural network model includes a basic feature extraction layer, a normal identification branch, and an abnormal type identification branch; Accordingly, the analysis of the operating data of the elevator energy supply device using the trained neural network model to identify the operating status of the elevator and the elevator energy supply device includes: Input the operating data of the elevator energy supply device into the basic feature extraction layer to extract basic electrical features; The basic electrical features are input into the normal recognition branch to extract the first depth features, and based on the first depth features, the elevator and elevator power supply device are identified as being in a normal or abnormal state. If an abnormal state is detected, the basic electrical features are input into the abnormal state type identification branch to extract the second deep features, and the abnormal state type is identified based on the second deep features. In the second deep feature, common features of abnormal states are removed.

7. The data analysis method according to claim 6, characterized in that, Before analyzing the operating data of the elevator energy supply device using the trained neural network model to identify the operating status of the elevator and the elevator energy supply device, the following steps are also included: Pre-collect operational data of the elevator energy supply device under various operating conditions to form a sample set; Each sample is passed through the basic feature extraction layer and the normal recognition branch in sequence. By constraining the minimization of the first deep feature difference of normal samples, the convergence of the first deep features of abnormal samples, and the maximization of the first deep feature difference between normal samples and abnormal samples, the basic feature extraction layer and the normal recognition branch are trained so that the normal recognition branch can distinguish between normal and abnormal states. The center of the first depth feature of each abnormal sample is taken as the common feature of the abnormal state; Each abnormal sample is sequentially passed through a pre-trained basic feature extraction layer and an anomaly type identification branch. By constraining the second deep features of each abnormal sample to be orthogonal to the common features, the difference in the second deep features of samples of the same anomaly type is minimized, and the difference in the second deep features between samples of different anomaly types is maximized. The anomaly type identification branch is trained so that it can distinguish between different anomaly types.

8. The data analysis method according to claim 6, characterized in that, Before analyzing the operating data of the elevator energy supply device using the trained neural network model to identify the operating status of the elevator and the elevator energy supply device, the following steps are also included: Pre-collect operational data of the elevator energy supply device under various operating conditions to form a sample set; Each sample is input into the basic feature extraction layer to extract the basic electrical features of each sample. Input the basic electrical characteristics of each normal sample into the normal identification branch, and input the basic electrical characteristics of each abnormal sample into both the normal identification branch and the abnormal identification branch. By constraining the minimization of the first-depth feature differences of normal samples, the convergence of the first-depth features of abnormal samples, the maximization of the first-depth feature differences between normal and abnormal samples, the minimization of the second-depth feature differences of samples of the same abnormal type, the maximization of the second-depth feature differences between samples of different abnormal types, and the orthogonality of the second-depth features of each abnormal sample with the center of the first-depth features of all abnormal samples, the parameters of the basic feature extraction layer, the normal identification branch, and the abnormal type identification branch are updated so that the normal identification branch can distinguish between normal and abnormal states, and the abnormal type identification branch can distinguish between different abnormal types.

9. The data analysis method according to claim 5, characterized in that, Also includes: Random feature code in response to feedback from the elevator power supply device; The random feature code is encrypted using an encryption algorithm to generate a key sequence, which is then sent for comparison and verification by the elevator energy supply device, thus completing the data monitoring authorization.

10. A data monitoring and analysis system for an elevator energy supply device, characterized in that, include: Elevator energy feeding device is used to collect its own power generation, power generation current and temperature in real time; Based on multiple preset current ranges and multiple power generation durations, the number of times the power generation current lasts for each duration in each range and the maximum current value are recorded in real time within a certain period of time, forming a multi-dimensional current event matrix; the number of times the power generation current exceeds the current safety threshold is monitored in real time and used as the first count. The number of times the temperature exceeds the safe temperature threshold is recorded in real time and counted as the second count; The first time the number exceeds the threshold is considered an overcurrent event, and the second time the number exceeds the threshold is considered an overtemperature event. The number of overcurrent events in the first time period and the number of overtemperature events in the second time period are recorded to form a multi-level protection matrix, wherein the first time period is shorter than the second time period; based on the power generation and power generation current, the cumulative power generation and power generation within a certain time period are calculated in real time; the multi-dimensional current event matrix, the multi-level protection matrix, and the cumulative power generation and power generation are sent to the cloud server in real time; The cloud server is used to analyze the operating data of the elevator power supply device using a trained neural network model, and to identify the operating status of the elevator and the elevator power supply device. The operating data includes: a multi-dimensional current event matrix and a multi-level protection matrix for the same time period, as well as the cumulative power generation and power output. The operating status includes: normal status and various abnormal statuses. The normal status means that both the elevator and the elevator power supply device are operating normally. The various abnormal statuses include elevator abnormality, elevator power supply device abnormality, and abnormal coordination between the elevator and the elevator power supply device.

Images