Method and device for diagnosing abnormal operation of elevator

Abstract

The invention provides a diagnosis method and device for abnormal operation of an elevator. The method comprises the steps that a door lock loop conventional voltage signal, a high-frequency impedance response signal, a landing door mechanical position signal and a door lock contact micro-vibration signal are collected in real time; whether a diagnosis process is started or not is judged according to the consistency of the conventional voltage signal and the door opening and closing instruction and the voltage transient interruption condition; extracting a high-frequency impedance module value Z, an impedance angle theta, a position offset delta d, micro-vibration total energy E and principal component characteristics E1, E2 and E3; and inputting the characteristic parameters into a multi-source information fusion diagnosis model constructed based on a D-S evidence theory, respectively calculating confidence coefficients of artificial short circuit H1 and contact adhesion H2 through basic probability distribution of four evidence bodies and a D-S synthesis rule, and outputting a diagnosis result according to a decision rule. According to the method, accurate distinguishing between manual short circuit and contact adhesion of the elevator door lock loop is achieved, the diagnosis accuracy and the maintenance efficiency are remarkably improved, and potential safety hazards such as door opening and elevator walking are eliminated.

Description

Technical Field

[0001] This invention belongs to the field of elevator diagnostic technology, and particularly relates to a diagnostic method and device for elevator malfunctions. Background Technology

[0002] The elevator's landing door lock circuit is a core component of the elevator safety protection system. Its function is to allow the elevator to start operating only after the landing doors at each floor are completely closed and mechanically locked. If the landing door lock circuit is accidentally activated when the doors are not locked, the elevator may run with the doors open, leading to serious accidents such as passengers falling or being sheared.

[0003] In existing technologies, the technique of measuring the resistance of a door lock circuit using a constant current source can detect the increase in resistance caused by contact oxidation. However, this technique still cannot distinguish the type of short circuit and relies only on the single dimension of resistance, making it susceptible to environmental interference. Alternatively, a pulse signal injection technique can be used to detect short circuits, but pulse signals are still time-domain analyses and cannot fully reflect changes in the electrical characteristics of the circuit. Furthermore, existing technologies do not incorporate mechanical position verification, resulting in insufficient reliability of diagnostic results.

[0004] In view of the above problems, this technical solution designs a diagnostic method and device for elevator malfunctions. Summary of the Invention

[0005] To achieve the objectives of this invention, the following technical solution is adopted: Firstly, this application provides a diagnostic method for elevator malfunctions, specifically including: S1: Real-time acquisition of elevator operation data, which includes at least the conventional voltage signal of the door lock circuit, the high-frequency impedance response signal of the door lock circuit, the mechanical position signal of the landing door, and the micro-vibration signal of the door lock contact. S2: Determine the basic on / off state of the door lock circuit using the normal voltage signal of the door lock circuit, and compare it with the door opening and closing command issued by the elevator control system. If the basic on / off state is inconsistent with the door opening and closing command, or if there is a voltage interruption caused by abnormal vibration, it is determined that the short-circuit type diagnostic process needs to be started. S3: Digitally demodulate the acquired high-frequency impedance response signal to extract the impedance modulus |Z| and impedance angle θ of the door lock circuit at a preset high frequency; process the mechanical position signal of the landing door to calculate the offset Δd of the door lock engagement position relative to the standard locking position; perform wavelet packet decomposition on the micro-vibration signal of the door lock contact to extract the total energy E of micro-vibration reflecting the contact action; S4: Input the feature parameters |Z|, θ, Δd, E and voltage-command consistency flag Cvc into the pre-constructed multi-source information fusion diagnostic model. The multi-source information fusion diagnostic model fuses different evidence bodies based on DS evidence theory, calculates the basic probability allocation values ​​of artificial short circuit and contact adhesion respectively, and outputs the diagnostic results and confidence level according to the decision rules. S5: Based on the diagnostic results and confidence level, match the pre-set treatment strategy knowledge base to generate information including short circuit type, severity, treatment suggestions, response time, and required tools and spare parts, and display and upload it through the human-computer interaction interface; S6: The output display is based on self-learning and model updates. The actual results and feature parameters of each diagnosis are stored in the historical database. Machine learning algorithms are used regularly to optimize the parameters of the diagnostic model, so that the system can adapt to the individual characteristics of different elevators.

[0006] Furthermore, the high-frequency impedance response signal of the door lock circuit is obtained in the following way: when the elevator is in standby mode and the door lock is closed, a high-frequency sinusoidal diagnostic signal with a frequency of 80kHz to 120kHz and a voltage amplitude of 3V to 8V is actively injected into the door lock circuit by the high-frequency signal injection unit. The voltage at both ends of the circuit and the current in the circuit are collected synchronously by the high-frequency response analysis unit. The in-phase component and the quadrature component are extracted by the quadrature phase-locked amplification technology, and then the impedance magnitude |Z| and impedance angle θ are calculated.

[0007] Furthermore, the mechanical position signal of the landing door is obtained by a miniature laser displacement sensor installed at the engagement point of the door lock on each landing. The measurement accuracy of the miniature laser displacement sensor is not less than 0.05mm, and it can continuously output the absolute distance value between the door lock hook and the lock seat in real time. The standard locking position is obtained through self-learning during elevator installation and commissioning or preset by the manufacturer.

[0008] Furthermore, the micro-vibration signal of the door lock contact is acquired by a high-frequency accelerometer sensor mounted on the outer shell of the door lock contact. The high-frequency accelerometer sensor has a frequency response range of 1Hz to 50kHz and a sampling frequency of not less than 100kHz. The wavelet packet decomposition adopts the db4 wavelet basis and has 4 decomposition layers to obtain wavelet packet coefficients of 16 frequency bands. The energy of each frequency band is calculated to form a 16-dimensional feature vector. The top 3 principal components with a contribution rate greater than 85% are extracted by principal component analysis as energy features E1, E2, and E3.

[0009] Furthermore, the voltage-command consistency flag Cvc is defined as follows: when the elevator control system issues an opening command and the door opening sensor detects that the door has been opened more than 80%, if the door lock circuit voltage still shows as closed, then Cvc=1; when the elevator is in standby mode and the door lock is closed, if the door lock circuit voltage shows as open, then Cvc=1; otherwise, Cvc=0.

[0010] Furthermore, the multi-source information fusion diagnostic model in step S4 is constructed using DS evidence theory, with an identification framework Θ = {H1, H2, H3}, where H1 represents a man-made hardware short circuit, H2 represents a short circuit due to contact adhesion, and H3 represents a fault without a short circuit. Four evidence bodies are constructed: Evidence body E1 is based on the impedance magnitude |Z|, evidence body E2 is based on the impedance angle θ, evidence body E3 is based on the position offset Δd, and evidence body E4 is based on the combination of the total micro-vibration energy E and the voltage-command consistency Cvc. A basic probability assignment function is defined for each evidence body, and the DS synthesis rule is used for fusion. The diagnostic results are output according to the decision rule.

[0011] Furthermore, the basic probability assignment function of evidence body E1 is constructed based on the ratio of the impedance magnitude |Z| to the normal value Z0, r = |Z| / Z0: When r < 0.4, m1({H1}) = 0.9, m1({H2}) = 0.05, m1(Θ) = 0.05; When 0.4 ≤ r < 0.7, linear interpolation is used; When 0.7 ≤ r < 0.9, linear interpolation is used; When r ≥ 0.9, m1({H3}) = 0.8, m1(Θ) = 0.2; The basic probability allocation function of the evidence body E2 is constructed based on the absolute value of the impedance angle |θ|: when |θ| ≥ 15°, m2({H1}) = 0.7, m2(Θ) = 0.3; when 5° < |θ| < 15°, linear interpolation allocation is used. When |θ| ≤ 5°, m2({H3}) = 0.8, m2(Θ) = 0.2; The basic probability allocation function of the evidence body E3 is constructed based on the position offset Δd: when Δd ≥ 2.0mm, m3({H1}) = 0.6, m3({H2}) = 0.2, m3(Θ) = 0.2; When 0.5mm < Δd < 2.0mm, linear interpolation is used. When Δd ≤ 0.5mm, m3({H3}) = 0.9, m3(Θ) = 0.1; The basic probability allocation function of the evidence body E4 is constructed based on the combination of the total energy of micro-vibration E and voltage-command consistency Cvc: if Cvc=1 and E is less than 30% of the normal action energy, then m4({H2}) = 0.8, m4(Θ) = 0.2; if Cvc=0 and E is normal, then m4({H3}) = 0.8, m4(Θ) = 0.2. In other cases, fuzzy rules are assigned based on the Mahalanobis distance between the principal components E1, E2, and E3 and the center of the normal pattern.

[0012] Furthermore, the decision rule in step S4 is as follows: set a preset threshold ε = 0.6, a first difference δ1 = 0.1, and a second difference δ2 = 0.1. If after fusion m(H1) > ε and m(H1) - m(H2) > δ1 and m(H1) - m(H3) > δ2, then the diagnosis is a human-caused hardware short circuit; If after fusion m(H2) > ε and m(H2) - m(H1) > δ1 and m(H2) - m(H3) > δ2, then the diagnosis is contact adhesion short circuit; otherwise, the diagnosis is no short circuit fault or insufficient confidence.

[0013] The beneficial effects of this invention are as follows: by integrating high-frequency impedance spectrum analysis, precision mechanical position measurement and micro-vibration signal detection, a multi-source information fusion diagnostic model based on DS evidence theory is constructed, which realizes the accurate distinction between artificial short circuit and contact adhesion in elevator door lock circuit, and effectively solves the technical problem that the existing technology can only detect the continuity of the circuit but cannot identify the type of short circuit. By actively injecting high-frequency diagnostic signals to extract impedance modulus and impedance angle, and combining this with mechanical position offset to verify the authenticity of electrical anomalies, the mechanical action status of the contacts can be directly determined using micro-vibration energy characteristics, which significantly improves the accuracy and reliability of diagnosis. By fusing different evidence sources using the DS evidence theory, the uncertainty and conflict of multi-source information are effectively addressed, avoiding misjudgments caused by a single signal. The model parameters are dynamically optimized through a self-learning mechanism, enabling the system to adapt to the individual characteristics of different elevators. Ultimately, it provides maintenance personnel with clear fault types, confidence levels, and targeted handling suggestions, significantly shortening fault troubleshooting time, fundamentally eliminating safety hazards caused by short circuits in door lock circuits, and significantly improving the safety of elevator operation.

[0014] Secondly, the present invention provides a diagnostic device for elevator malfunctions, applied to the aforementioned diagnostic method for elevator malfunctions, specifically including: The data acquisition module, used to perform the data acquisition function in step S1, includes a voltage sensor, a high-frequency signal injection unit, a high-frequency response analysis unit, a miniature laser displacement sensor, a high-frequency acceleration sensor, and a door opening sensor. The data preprocessing module, electrically connected to the data acquisition module, is used to filter, amplify, convert analog to digital and synchronize various acquired signals. The feature extraction module is electrically connected to the data preprocessing module and is used to perform the feature extraction function described in step S3, extracting the impedance modulus |Z|, impedance angle θ, position offset Δd, total energy of micro-vibration E, energy characteristics E1, E2, E3 and voltage-command consistency flag Cvc; The multi-source information fusion diagnostic module is electrically connected to the feature extraction module. It has a built-in diagnostic model based on DS evidence theory and is used to perform the fusion diagnostic function described in step S4. It calculates the basic probability allocation values ​​of artificial short circuit H1 and contact adhesion H2, and outputs the diagnostic results and confidence level according to the decision rules. The treatment strategy generation module is electrically connected to the multi-source information fusion diagnosis module. It has a built-in treatment strategy knowledge base and is used to execute the treatment suggestion generation function described in step S5. It generates corresponding treatment suggestions based on the diagnosis results and confidence levels. The human-computer interaction module is electrically connected to the treatment strategy generation module. It is used to display diagnostic results, confidence levels, treatment suggestions, and to receive user commands. The communication module, electrically connected to the human-machine interaction module, is used to upload diagnostic results and related information to the elevator monitoring center and maintenance personnel's terminal.

[0015] Furthermore, a diagnostic device for elevator malfunctions also includes a self-learning module connected to a historical database. This module performs the self-learning and model update functions described in step S6, periodically utilizing historical diagnostic data and maintenance feedback to optimize the basic probability allocation function parameters of the evidence body using a random forest algorithm, enabling the system to adapt to the individual characteristics of different elevators.

[0016] Other features and advantages will be set forth in the following description, and the objects and other advantages of the invention are realized and obtained through the structures particularly pointed out in the description and the drawings.

[0017] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0018] The above and other features and advantages of the present invention will become more apparent from a detailed description of exemplary embodiments thereof with reference to the accompanying drawings.

[0019] Figure 1 This is a flowchart of a diagnostic method for elevator malfunctions; Figure 2 This is an architecture diagram of a diagnostic device for elevator malfunctions. Detailed Implementation

[0020] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this specification, and not all embodiments. Based on the embodiments of this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this specification.

[0021] Example 1 First aspect like Figure 1 As shown, this application provides a diagnostic method for elevator malfunctions, specifically including: S1: Real-time acquisition of elevator operation data, which includes at least the door lock circuit conventional voltage signal, door lock circuit high-frequency impedance response signal, landing door mechanical position signal, and door lock contact micro-vibration signal. Furthermore, the conventional voltage signal of the door lock circuit is acquired by a voltage sensor connected in parallel across the two ends of the door lock circuit, with an acquisition frequency of not less than 1000Hz; The high-frequency impedance response signal of the door lock circuit is obtained in the following way: when the elevator is in standby mode and the door lock is closed, a high-frequency sinusoidal diagnostic signal with a frequency of 80kHz to 120kHz and a voltage amplitude of 3V to 8V is actively injected into the door lock circuit by the high-frequency signal injection unit. The voltage at both ends of the circuit and the current in the circuit are collected synchronously by the high-frequency response analysis unit. The in-phase component and the quadrature component are extracted by the quadrature phase-locked amplification technology, and then the impedance magnitude |Z| and impedance angle θ are calculated. Quadrature lock-in amplification (QLAB) is a signal processing technique used to extract the amplitude and phase of weak signals from a background of strong noise. Its core principle is to utilize signal correlation by multiplying the signal with a reference signal of the same frequency as the measured signal and then performing a low-pass filter. This enables narrowband detection of signals at specific frequencies. Specifically, given the frequency of the measured signal, two orthogonal reference signals (sine and cosine) are constructed, multiplied by the measured signal respectively, and then the DC component is extracted using a low-pass filter, thereby calculating the amplitude and phase of the measured signal. Since this is a conventional technique, it will not be elaborated upon further here.

[0022] It should be noted that the amplitude of the high-frequency sine wave diagnostic signal is much lower than the safe voltage, and the frequency is far from the power frequency and its harmonics, so it will not interfere with the normal operation of the elevator's original safety circuit. Furthermore, the mechanical position signal of the landing door is obtained by a miniature laser displacement sensor installed at the engagement point of the door lock at each landing station. The miniature laser displacement sensor has a measurement accuracy of no less than 0.05mm and can continuously output the absolute distance value between the door lock hook and the lock seat in real time.

[0023] S2: Determine the basic on / off state of the door lock circuit using the normal voltage signal of the door lock circuit, and compare it with the door opening and closing command issued by the elevator control system. If the basic on / off state is inconsistent with the door opening and closing command, or if there is a voltage interruption caused by abnormal vibration, it is determined that the short-circuit type diagnostic process needs to be started. That is, the basic on / off state of the door lock circuit is determined by the normal voltage signal of the door lock circuit: when the normal voltage of the door lock circuit is higher than 18V, it is determined to be a closed circuit, and when it is lower than 5V, it is determined to be an open circuit. The on / off state is compared with the door opening and closing commands issued by the elevator control system, and voltage interruptions are monitored. Specifically, the voltage-command consistency flag Cvc is defined as follows: When the elevator control system issues a door opening command and the door opening sensor detects that the door has been opened more than 80%, if the door lock circuit voltage U still shows as closed, it indicates that the door has been opened but the circuit has not been disconnected. At this time, Cvc=1. When the elevator is in standby mode and the door lock is closed, if the door lock circuit voltage U shows an open circuit, it indicates that the door lock is closed but the circuit is not open, and Cvc=1 at this time; In other cases, Cvc=0.

[0024] It should also be noted that when detecting voltage interruptions caused by abnormal vibration: if the door lock circuit voltage U experiences more than 5 brief interruptions (intermittent interruption time less than 100ms) in the past 24 hours, it indicates that there may be poor contact or intermittent circuit breakage caused by abnormal vibration, and the diagnostic process needs to be initiated.

[0025] The short-circuit type diagnostic process needs to be initiated when any of the following conditions are met: Condition 1: Cvc=1; Condition 2: There are voltage interruptions caused by abnormal vibrations (i.e., the number of interruptions exceeds the threshold).

[0026] S3: Process the raw signals acquired in S1 and extract the feature parameters used for diagnosis; Specifically, this includes: digitally demodulating the acquired high-frequency impedance response signal to extract the characteristic parameters of the door lock circuit at a preset high frequency, namely the impedance modulus |Z| and impedance angle θ; processing the mechanical position signal of the landing door to calculate the characteristic parameter offset Δd of the door lock engagement position relative to the standard locking position; and performing wavelet packet decomposition on the micro-vibration signal of the door lock contact to extract the characteristic parameter total energy E of micro-vibration reflecting the contact action. Furthermore, the extraction process of the impedance magnitude |Z| and impedance angle θ is as follows: The high-frequency voltage and current signals in the acquired high-frequency impedance response signal are digitally demodulated, i.e., digital bandpass filtered with a center frequency of 100kHz and a bandwidth of 20kHz. Then, based on quadrature lock-in amplification technology, they are multiplied by the in-phase reference signal cos(ωt) and the quadrature reference signal sin(ωt), respectively. After low-pass filtering, the in-phase component I and the quadrature component Q are obtained, and then the following calculations are performed: Impedance magnitude |Z| = sqrt(I) 2 + Q 2 ) / ,in, The effective value of the injected current is obtained by measuring the effective value of the current in the injection circuit through a high-frequency response analysis unit. Impedance angle θ = arctan(Q / I).

[0027] Under normal circumstances, the door lock circuit has a |Z| of approximately 10Ω and a θ close to 0° at 100kHz. When the circuit is short-circuited, |Z| will decrease significantly, and θ may shift due to the parasitic inductance of the shorting wire or the sticky contacts.

[0028] It should be noted that the standard locking position The information is acquired through self-learning during elevator installation and commissioning or preset by the manufacturer. Furthermore, the process for extracting the offset Δd is as follows: Δd = | |, The lock engagement position is indicated by Δd, which reflects the depth of the mechanical engagement of the lock. Normally, it should be less than 0.5mm. If Δd exceeds 2.0mm, it means that the door is not truly locked.

[0029] The extraction process of the total micro-vibration energy E is as follows: The micro-vibration signal of the door lock contact is acquired by a high-frequency accelerometer sensor mounted on the outer shell of the door lock contact. The frequency response range of the high-frequency accelerometer sensor is 1Hz to 50kHz, and the sampling frequency is not less than 100kHz. Wavelet packet decomposition uses the db4 wavelet basis, with 4 decomposition layers, obtaining wavelet packet coefficients for 16 frequency bands. The energy of each frequency band is calculated to form a 16-dimensional feature vector. Then, principal component analysis is used to extract the top 3 principal components with a contribution rate greater than 85% as energy features E1, E2, and E3. At the same time, the total micro-vibration energy E of the entire signal during the contact operation is calculated as an auxiliary feature. During normal contact operation, the energy in a specific frequency band (e.g., 5kHz to 15kHz) is relatively high; if the contact is stuck, this energy will be lost during the door opening process. It should be noted that the micro-vibration characteristics include total energy E and principal component characteristics E1, E2, and E3. E is used to determine whether the contact has activated (total energy level), while E1, E2, and E3 are used to determine whether the vibration mode is normal.

[0030] S4: Input the feature parameters |Z|, θ, Δd, E and voltage-command consistency flag Cvc into the pre-built multi-source information fusion diagnostic model. The multi-source information fusion diagnostic model fuses different evidence bodies based on DS evidence theory, calculates the basic probability allocation values ​​of artificial short circuit and contact adhesion respectively, and outputs the diagnostic results and confidence level according to the decision rules. Preferably, the multi-source information fusion diagnostic model is constructed using the DS evidence theory, with an identification framework Θ = {H1, H2, H3}, where H1 represents a man-made hardware short circuit, H2 represents a short circuit due to contact adhesion, and H3 represents a fault without a short circuit. The method for constructing evidence is as follows: Evidence body E1 is based on the impedance magnitude |Z|, evidence body E2 is based on the impedance angle θ, evidence body E3 is based on the position offset Δd, and evidence body E4 is based on the combination of specific frequency band energy characteristics E and voltage-command consistency Cvc. Then, a basic probability allocation function is defined for each evidence body, and DS synthesis rules are used for fusion. The diagnostic results are output according to the decision rules. Evidence body E1: Based on the ratio of the impedance magnitude |Z| to the normal value Z0, r = |Z| / Z0. Define the basic probability assignment function m1: When r < 0.4, m1({H1}) = 0.9, m1({H2}) = 0.05, m1(Θ) = 0.05; When 0.4 ≤ r < 0.7, the problem may be due to contact adhesion or intentional short circuit. Linear interpolation is used, assuming... , , =0.2; Calculate the total weight S= + + That is, m1({H1}) = m1({H2}) = m1(Θ)= ; When 0.7 ≤ r < 0.9, it may indicate contact adhesion or normal operation. Linear interpolation is used, assuming... , , =0.1; Calculate the total weight S= + + That is, m1({H2}) = m1({H3}) = m1(Θ)= ; When r ≥ 0.9, m1({H3}) = 0.8, m1(Θ) = 0.2.

[0031] It should be noted that the values ​​of 0.4, 0.7, and 0.9 are based on the following: Actual measurements show that when Z drops below 30% of its normal value, the impedance of the shorting wire is extremely low when manually short-circuited; therefore, 0.4 is taken as the lower limit of the strong confidence interval. When contacts stick together, there is still some contact resistance due to the metal connection formed by the melting of the contacts. Actual measurements show that Z is between 50% and 80% of its normal value; therefore, 0.4-0.7 is taken as the main suspected range for contact sticking. When contacts are oxidized or slightly contaminated, the resistance increases slightly but is not completely failed; actual measurements show that Z is between 70% and 90% of its normal value. In a normal door lock circuit, Z fluctuation is usually within ±10%; therefore, 0.9 is taken as the lower limit of the normal range. And to retain a small amount of uncertainty: even if r is extremely low, there are theoretically still very rare cases that could be misjudged. Therefore, H2 = 0.05 and the total set Θ = 0.05 are assigned to cover these small possibilities and unknown factors. This conforms to the principle of DS evidence theory in dealing with uncertainty, avoiding the complete exclusion of other hypotheses; then, combined with summation normalization, 0.9 + 0.05 + 0.05 = 1, therefore m1(Θ) = 0.05.

[0032] Evidence body E2: Based on the absolute value of the impedance angle |θ|, define the basic probability assignment function m2: When |θ| ≥ 15°, m2({H1}) = 0.7, m2(Θ) = 0.3; When 5° < |θ| < 15°, it may be due to artificial short-circuiting or normal linear interpolation assignment. , , =0.2; Calculate the total weight S= + + That is, m2({H1}) = m2({H3}) = m2(Θ) = ; When |θ| ≤ 5°, m2({H3}) = 0.8, m2(Θ) = 0.2.

[0033] It should be noted that when a short circuit is intentionally applied, the shorting wire introduces parasitic inductance (approximately 0.5-2μH), resulting in an inductive reactance of approximately 0.3-1.2Ω at 100kHz, causing a significant shift in the impedance angle. Actual measurements show that an impedance angle of θ between 15° and 30° strongly indicates intentional short circuit. When contacts are stuck together, the metal connection formed by the melting of the contacts still retains some resistance characteristics, and the inductive effect is not significant; an impedance angle between 5° and 15° indicates possible contact sticking. A normal door lock circuit exhibits purely resistive characteristics; an angle within ±5° indicates it is basically normal.

[0034] Based on statistical data experiments, when |θ| ≥ 15°, H1 is given a support of 0.7, and the remaining 0.3 is allocated to the entire set Θ; when |θ| ≤ 5°, H3 is given a support of 0.8, and the remaining 0.2 is allocated to the entire set Θ.

[0035] Evidence body E3: Based on the position offset Δd, define the basic probability allocation function m3: When Δd ≥ 2.0mm, the mechanical part is seriously out of place. Based on the sample analysis, the values ​​assigned are: m3({H1}) = 0.6, m3({H2}) = 0.2, m3(Θ) = 0.2. When 0.5mm < Δd < 2.0mm, a slight mechanical anomaly is identified. Linear interpolation is used, and the following is assumed: *0.6, , *0.9; Calculate the total probability P= + + That is, m3({H1}) = m3({H2})= m3({H3})= m3(Θ) = 1 - P; When Δd ≤ 0.5mm, the machine is normal, m3({H3}) = 0.9, m3(Θ) = 0.1.

[0036] It should be noted that Δd ≥ 2.0mm, according to GB 7588-2003 "Safety Code for Elevator Manufacturing and Installation", the door lock engagement depth should not be less than 7mm. Through actual measurement, the positional fluctuation during normal engagement is usually within ±0.5mm. When Δd exceeds 2.0mm, it indicates a serious lack of mechanical engagement of the door lock, and the door is not truly locked. 0.5mm < Δd < 2.0mm indicates slight insufficient engagement due to wear of the door lock rollers, loose door blades, etc. In this case, the door may be barely locked, but there is a risk. Δd ≤ 0.5mm represents minor fluctuations within the normal engagement range.

[0037] Evidence body E4: Based on the combination of specific frequency band energy characteristics E and voltage-command consistency Cvc, a basic probability allocation function m4 is defined: If Cvc=1 and E is less than 30% of the normal motion energy, after analysis of samples collected in the laboratory and on-site, the values ​​assigned are: m4({H2}) = 0.8, m4(Θ) = 0.2; If Cvc=0 and E is normal, then m4({H3}) = 0.8, m4(Θ) = 0.2; In other cases, fuzzy rules are used for assignment based on the distances of principal components E1, E2, and E3 to the normal pattern. Let E1, E2, and E3 be the three principal component features obtained through principal component analysis, and calculate their Mahalanobis distance D to the center of the normal pattern (obtained through training on historical data). Define membership degree: =exp(- That is, m4({H3}) = 0.6* m4(Θ) = 1 - 0.6* m4({H1})=0, m4({H2})=0. This rule indicates that the further the deviation from the normal pattern, the greater the uncertainty.

[0038] It should be noted that the setting of E being lower than 30% of the normal operating energy is based on the fact that the energy of the micro-vibration signal collected by the high-frequency accelerometer is relatively stable during normal contact operation. By statistically analyzing 50 normal contact operations, 30% of the average energy is taken as the threshold for determining "energy deficiency"; the setting of Cvc=1 is based on the fact that after the door opening command is issued, the door has opened but the circuit is still conducting, which directly proves that the circuit is abnormal.

[0039] Evidence synthesis: The four evidence bodies are fused using the DS synthesis rule to obtain the basic probability distribution m = m1 + m2 + m3 + m4 after fusion.

[0040] The synthesis rules are as follows: m(A) = (1 / K) * Σ_{B∩C∩D∩E=A} m1(B) *m2(C) * m3(D)* m4(E), where K is the normalization factor.

[0041] Where A: the subset whose probability needs to be calculated after synthesis, is any subset of Θ; C, D, E: represent the focal elements with non-zero probabilities in m1, m2, m3, and m4 (i.e., the non-zero subsets involved in the basic probability assignment). The normalization factor K is the total probability of all non-empty intersections: K = m1(B)*m2(C)*m3(D)*m4(E); The intersection operation B∩C represents the common part of the focal elements. It only contributes to the normalization factor K when the intersection is non-empty. When the intersection equals a specific A, its multiplication is added to the numerator of m(A).

[0042] Decision output: Set a preset threshold ε = 0.6, first difference δ1 = 0.1, second difference δ2 = 0.1. The decision rule is: If m(H1) > ε and m(H1) - m(H2) > δ1 and m(H1) - m(H3) > δ2, then the diagnosis is "human-caused hardware short circuit", and the confidence level m(H1) is output. If m(H2) > ε and m(H2) - m(H1) > δ1 and m(H2) - m(H3) > δ2, then the diagnosis is "contact adhesion short circuit", and the confidence level m(H2) is output. Otherwise, diagnose it as "no short circuit fault" or insufficient confidence level, and continue monitoring.

[0043] It should be noted that ε = 0.6 is set based on the DS evidence theory, where a credible decision threshold of 0.5-0.7 is typically used. 0.6 is a compromise value that comprehensively considers sensitivity and specificity. At the same time, through ROC curve analysis, the point with the largest Youden index is selected as the threshold. After verification with 100 sets of historical data, the diagnostic accuracy reaches 92.5% when ε = 0.6, ensuring that the confidence of the diagnostic results is high enough. The setting of δ1 = δ2 = 0.1 is based on the requirement that the probability of the dominant hypothesis is at least 0.1 higher than that of other hypotheses, avoiding ambiguous diagnoses and ensuring the discriminative power of the decision.

[0044] S5: Based on the diagnostic results and confidence level, match the pre-set treatment strategy knowledge base to generate specific treatment suggestions. The treatment suggestions include information such as short circuit type, severity, treatment suggestion text, response time, and required tools and spare parts, and are displayed and uploaded through a human-computer interaction interface; The details are as follows: If the diagnosis is "human-caused short circuit" with a confidence level greater than 0.8, it is classified as an emergency. The recommended action is: "Emergency stop! A suspected short circuit has been detected in the door lock circuit. The mechanical position indicator shows the door is not locked, but the electrical circuit is abnormally conductive. Immediately check the door lock wiring on this floor for any short circuits. At the same time, reassure the passengers inside the elevator car." The recommended response time is "immediate". Required tools include a multimeter, screwdriver, and jumper wire finder. There are no special requirements for spare parts.

[0045] If the diagnosis is "human-caused short circuit" and the confidence level is between 0.6 and 0.8, it is classified as an alarm level. The recommended action is: "Stop the elevator and check! There may be a human-caused short circuit in the door lock circuit; please focus your investigation on this area." The recommended response time is "within 2 hours". Required tools include a multimeter and a screwdriver. There are no special requirements for spare parts.

[0046] If the diagnosis is "contact adhesion" with a confidence level greater than 0.8, it is classified as an emergency. The recommended action is: "Emergency stop! Door lock contacts may be stuck. The circuit remains conductive after the door opening command is issued, and the contact micro-vibration signal is missing. Please immediately check if the door lock contacts are melted and replace them if necessary." The recommended response time is "immediate". Required tools include a multimeter and a screwdriver. The required spare part is a door lock switch.

[0047] If the diagnosis is "contact adhesion" and the confidence level is between 0.6 and 0.8, it is classified as an alarm level. The recommended action is: "Stop the elevator and check! There may be a risk of contact adhesion at the door lock contacts. Please check the contact status." The recommended response time is "within 4 hours". Required tools include a multimeter and a screwdriver. The required spare part is a door lock switch.

[0048] If the diagnosis is "no short-circuit fault" or the confidence level is insufficient, it is considered normal. The recommended action is: "No abnormality, continue monitoring." The recommended response time is "-", and the required tools and spare parts are all "-". S6: The output display is based on self-learning and model update. The actual results and feature parameters of each diagnosis are stored in the historical database. Machine learning algorithms are used regularly to optimize the parameters of the diagnostic model so that the system can adapt to the individual characteristics of different elevators. Specifically, the diagnostic results, treatment suggestions, and severity levels are displayed through a human-computer interaction interface and uploaded to the elevator monitoring center and maintenance personnel's mobile app via 4G / 5G networks. Then, historical diagnostic records are statistically analyzed monthly. The actual fault types reported by maintenance personnel are used as labels, and a random forest algorithm is used to train a classification model to optimize the basic probability assignment function parameters of the evidence body, such as adjusting thresholds and membership function shapes. The updated parameters are stored in the model library for subsequent diagnostic use, enabling the system to adapt to the unique characteristics of different elevators.

[0049] It should also be noted that the basic probability allocation function in step S4 can be dynamically adjusted based on the elevator's historical operating data. For example, for newly installed elevators, the threshold can be tightened appropriately; for older elevators, the threshold can be relaxed appropriately to avoid frequent false alarms.

[0050] Example 2 like Figure 2 As shown, in a second aspect, the present invention provides a diagnostic device for elevator malfunctions, applied to the aforementioned diagnostic method for elevator malfunctions, specifically including: A data acquisition module is used to perform the data acquisition function in step S1; the data acquisition module includes: A voltage sensor, connected in parallel across the door lock circuit, is used to collect the normal voltage signal of the door lock circuit; The high-frequency signal injection unit is used to actively inject a high-frequency sine wave diagnostic signal with a frequency of 80kHz to 120kHz and a voltage amplitude of 3V to 8V into the door lock circuit when the elevator is in standby mode and the door lock is closed. The high-frequency response analysis unit, connected to the high-frequency signal injection unit, is used to synchronously acquire the voltage and current responses in the circuit, and calculate the impedance magnitude |Z| and impedance angle θ through orthogonal lock-in amplification technology; Miniature laser displacement sensors are installed at the engagement point of each landing door lock to collect mechanical position signals of the landing door, with a measurement accuracy of not less than 0.05mm; A high-frequency accelerometer is mounted on the door lock contact housing to collect micro-vibration signals of the door lock contacts. The frequency response range is 1Hz to 50kHz, and the sampling frequency is not less than 100kHz. A door opening sensor is used to acquire door opening signals.

[0051] The data preprocessing module, electrically connected to the data acquisition module, is used to filter, amplify, convert analog to digital, and synchronize various acquired signals.

[0052] The feature extraction module, electrically connected to the data preprocessing module, is used to perform the feature extraction function described in step S3, including: Extract the impedance magnitude |Z| and impedance angle θ from the high-frequency impedance response signal; Extract the position offset Δd from the mechanical position signal of the landing door; The total energy E of micro-vibration is extracted from the micro-vibration signal, and the energy features E1, E2, and E3 are extracted by wavelet packet decomposition and principal component analysis. A voltage-command consistency flag Cvc is generated by comparing the door lock circuit voltage with the door opening / closing command.

[0053] The multi-source information fusion diagnostic module is electrically connected to the feature extraction module. It has a built-in diagnostic model based on DS evidence theory and is used to perform the fusion diagnostic function described in step S4. That is, based on the input feature parameters |Z|, θ, Δd, E, Cvc and E1, E2, E3, it calculates the basic probability allocation values ​​of artificial short circuit H1 and contact adhesion H2 according to the predefined basic probability allocation function of evidence body and DS synthesis rules, and outputs the diagnostic results and confidence level according to the decision rules.

[0054] The treatment strategy generation module is electrically connected to the multi-source information fusion diagnosis module. It has a built-in treatment strategy knowledge base and is used to perform the treatment suggestion generation function described in step S5. It generates corresponding treatment suggestions based on the diagnosis results and confidence levels, including short circuit type, severity level, treatment text, response time, required tools and spare parts.

[0055] The human-computer interaction module is electrically connected to the treatment strategy generation module. It is used to display diagnostic results, confidence levels, treatment suggestions, and other information, and to receive user commands.

[0056] The communication module, electrically connected to the human-machine interaction module, is used to upload diagnostic results and related information to the elevator monitoring center and maintenance personnel terminal via 4G / 5G network.

[0057] The self-learning module connects to the historical database and is used to perform the self-learning and model update functions described in step S6. It periodically utilizes historical diagnostic data and maintenance feedback, and uses the random forest algorithm to optimize the basic probability allocation function parameters of the evidence body, so that the system can adapt to the individual characteristics of different elevators.

[0058] It should be noted that all the above modules are integrated into the same main controller, working together to achieve accurate diagnosis and generate handling suggestions for elevator door lock circuit short circuits.

[0059] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments of apparatus, devices, and non-volatile computer storage media are basically similar to the method embodiments, so the descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0060] The foregoing has described specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0061] The above description is merely one or more embodiments of this specification and is not intended to limit this specification. Various modifications and variations can be made to the one or more embodiments of this specification by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of one or more embodiments of this specification should be included within the scope of the claims of this specification.

Claims

1. A diagnostic method for elevator malfunctions, characterized in that, Specifically, it includes: S1: Real-time acquisition of elevator operation data, which includes at least the conventional voltage signal of the door lock circuit, the high-frequency impedance response signal of the door lock circuit, the mechanical position signal of the landing door, and the micro-vibration signal of the door lock contact. S2: Determine the basic on / off state of the door lock circuit using the normal voltage signal of the door lock circuit, and compare it with the door opening and closing command issued by the elevator control system. If the basic on / off state is inconsistent with the door opening and closing command, or if there is a voltage interruption caused by abnormal vibration, it is determined that the short-circuit type diagnostic process needs to be started. S3: Digitally demodulate the acquired high-frequency impedance response signal to extract the impedance modulus |Z| and impedance angle θ of the door lock circuit at a preset high frequency; process the mechanical position signal of the landing door to calculate the offset Δd of the door lock engagement position relative to the standard locking position; perform wavelet packet decomposition on the micro-vibration signal of the door lock contact to extract the total energy E of micro-vibration reflecting the contact action; S4: Input the feature parameters |Z|, θ, Δd, E and voltage-command consistency flag Cvc into the pre-constructed multi-source information fusion diagnostic model. The multi-source information fusion diagnostic model fuses different evidence bodies based on DS evidence theory, calculates the basic probability allocation values ​​of artificial short circuit and contact adhesion respectively, and outputs the diagnostic results and confidence level according to the decision rules. S5: Based on the diagnostic results and confidence level, match the pre-set treatment strategy knowledge base to generate information including short circuit type, severity, treatment suggestions, response time, and required tools and spare parts, and display and upload it through the human-computer interaction interface; S6: The output display is based on self-learning and model updates. The actual results and feature parameters of each diagnosis are stored in the historical database. Machine learning algorithms are used regularly to optimize the parameters of the diagnostic model, so that the system can adapt to the individual characteristics of different elevators.

2. The diagnostic method for elevator malfunctions according to claim 1, characterized in that, The high-frequency impedance response signal of the door lock circuit is obtained in the following way: when the elevator is in standby mode and the door lock is closed, a high-frequency sinusoidal diagnostic signal with a frequency of 80kHz to 120kHz and a voltage amplitude of 3V to 8V is actively injected into the door lock circuit by the high-frequency signal injection unit. The voltage at both ends of the circuit and the current in the circuit are collected synchronously by the high-frequency response analysis unit. The in-phase component and the quadrature component are extracted by the quadrature phase-locked amplification technology, and then the impedance magnitude |Z| and impedance angle θ are calculated.

3. The diagnostic method for elevator malfunctions according to claim 1, characterized in that, The mechanical position signal of the landing door is obtained by a miniature laser displacement sensor installed at the engagement point of the door lock on each landing. The miniature laser displacement sensor has a measurement accuracy of not less than 0.05mm and can continuously output the absolute distance value between the door lock hook and the lock seat in real time. The standard locking position is obtained through self-learning during elevator installation and commissioning or preset by the manufacturer.

4. The diagnostic method for elevator malfunctions according to claim 1, characterized in that, The micro-vibration signal of the door lock contact is acquired by a high-frequency accelerometer sensor mounted on the outer shell of the door lock contact. The high-frequency accelerometer sensor has a frequency response range of 1Hz to 50kHz and a sampling frequency of not less than 100kHz. The wavelet packet decomposition adopts the db4 wavelet basis and has 4 decomposition layers to obtain wavelet packet coefficients of 16 frequency bands. The energy of each frequency band is calculated to form a 16-dimensional feature vector. The top 3 principal components with a contribution rate greater than 85% are extracted by principal component analysis as energy features E1, E2, and E3.

5. The diagnostic method for elevator malfunctions according to claim 1, characterized in that, The voltage-command consistency flag Cvc is defined as follows: when the elevator control system issues an opening command and the door opening sensor detects that the door is opened more than 80%, if the door lock circuit voltage still shows as closed, then Cvc=1; when the elevator is in standby mode and the door lock is closed, if the door lock circuit voltage shows as open, then Cvc=1; otherwise, Cvc=0.

6. The diagnostic method for elevator malfunctions according to claim 1, characterized in that, The multi-source information fusion diagnostic model in step S4 is constructed using DS evidence theory, with an identification framework Θ = {H1, H2, H3}, where H1 represents a man-made hardware short circuit, H2 represents a short circuit due to contact adhesion, and H3 represents a fault without a short circuit. Four evidence bodies are constructed: Evidence body E1 is based on the impedance magnitude |Z|, evidence body E2 is based on the impedance angle θ, evidence body E3 is based on the position offset Δd, and evidence body E4 is based on the combination of the total micro-vibration energy E and the voltage-command consistency Cvc. A basic probability assignment function is defined for each evidence body, and the DS synthesis rule is used for fusion. The diagnostic results are output according to the decision rule.

7. The diagnostic method for elevator malfunctions according to claim 6, characterized in that, The basic probability allocation function of the evidence body E1 is constructed based on the ratio of the impedance magnitude |Z| to the normal value Z0, r = |Z| / Z0: When r < 0.4, m1({H1}) = 0.9, m1({H2}) = 0.05, m1(Θ) = 0.05; When 0.4 ≤ r < 0.7, linear interpolation is used; When 0.7 ≤ r < 0.9, linear interpolation is used; When r ≥ 0.9, m1({H3}) = 0.8, m1(Θ) = 0.2; The basic probability allocation function of the evidence body E2 is constructed based on the absolute value of the impedance angle |θ|: when |θ| ≥ 15°, m2({H1}) = 0.7, m2(Θ) = 0.3; when 5° < |θ| < 15°, linear interpolation allocation is used. When |θ| ≤ 5°, m2({H3}) = 0.8, m2(Θ) = 0.2; The basic probability allocation function of the evidence body E3 is constructed based on the position offset Δd: when Δd ≥ 2.0mm, m3({H1}) = 0.6, m3({H2}) = 0.2, m3(Θ) = 0.2; When 0.5mm < Δd < 2.0mm, linear interpolation is used. When Δd ≤ 0.5mm, m3({H3}) = 0.9, m3(Θ) = 0.1; The basic probability allocation function of the evidence body E4 is constructed based on the combination of the total energy of micro-vibration E and voltage-command consistency Cvc: if Cvc=1 and E is less than 30% of the normal action energy, then m4({H2}) = 0.8, m4(Θ) = 0.2; if Cvc=0 and E is normal, then m4({H3}) = 0.8, m4(Θ) = 0.

2. In other cases, fuzzy rules are assigned based on the Mahalanobis distance between the principal components E1, E2, and E3 and the center of the normal pattern.

8. The diagnostic method for elevator malfunctions according to claim 1, characterized in that, The decision rule in step S4 is as follows: set a preset threshold ε = 0.6, a first difference δ1 = 0.1, and a second difference δ2 = 0.

1. If after fusion m(H1) > ε and m(H1) - m(H2) > δ1 and m(H1) - m(H3) > δ2, then the diagnosis is a human-caused hardware short circuit; If after fusion m(H2) > ε and m(H2) - m(H1) > δ1 and m(H2) - m(H3) > δ2, then the diagnosis is contact adhesion short circuit; otherwise, the diagnosis is no short circuit fault or insufficient confidence.

9. A diagnostic device for elevator malfunctions, applied to the diagnostic method for elevator malfunctions as described in any one of claims 1 to 8, characterized in that, Specifically, it includes: The data acquisition module, used to perform the data acquisition function in step S1, includes a voltage sensor, a high-frequency signal injection unit, a high-frequency response analysis unit, a miniature laser displacement sensor, a high-frequency acceleration sensor, and a door opening sensor. The data preprocessing module, electrically connected to the data acquisition module, is used to filter, amplify, convert analog to digital and synchronize various acquired signals. The feature extraction module is electrically connected to the data preprocessing module and is used to perform the feature extraction function described in step S3, extracting the impedance modulus |Z|, impedance angle θ, position offset Δd, total energy of micro-vibration E, energy characteristics E1, E2, E3 and voltage-command consistency flag Cvc; The multi-source information fusion diagnostic module is electrically connected to the feature extraction module. It has a built-in diagnostic model based on DS evidence theory and is used to perform the fusion diagnostic function described in step S4. It calculates the basic probability allocation values ​​of artificial short circuit H1 and contact adhesion H2, and outputs the diagnostic results and confidence level according to the decision rules. The treatment strategy generation module is electrically connected to the multi-source information fusion diagnosis module. It has a built-in treatment strategy knowledge base and is used to execute the treatment suggestion generation function described in step S5. It generates corresponding treatment suggestions based on the diagnosis results and confidence levels. The human-computer interaction module is electrically connected to the treatment strategy generation module. It is used to display diagnostic results, confidence levels, treatment suggestions, and to receive user commands. The communication module, electrically connected to the human-machine interaction module, is used to upload diagnostic results and related information to the elevator monitoring center and maintenance personnel's terminal.

10. The diagnostic device for elevator malfunction according to claim 9, characterized in that, It also includes a self-learning module, which is connected to a historical database to perform the self-learning and model update functions described in step S6. It periodically uses historical diagnostic data and maintenance feedback to optimize the basic probability allocation function parameters of the evidence body using a random forest algorithm, so that the system can adapt to the individual characteristics of different elevators.

Images