Hoisting mechanical equipment health online monitoring and remote control method and system

Abstract

The invention discloses an online health monitoring and remote control method and system for hoisting mechanical equipment, and relates to the technical field of industrial equipment health monitoring and intelligent control, and the method comprises the steps: constructing health grade judgment, credibility calculation and abnormal trend intensity analysis results into a real number BPA, carrying out the complex field expansion of the real number BPA, and carrying out the remote monitoring of the health of the hoisting mechanical equipment; obtaining a plurality of evidence sets; calculating the credibility of the plurality of evidences in the plurality of evidence set, and classifying the plurality of evidences, including high-credibility evidences and low-credibility evidences; and performing weight discount correction on the low-credibility evidence, fusing the corrected plurality of evidences by using a plurality of evidence combination rule, and generating a final health evidence value by adopting a conservative decision strategy. Therefore, a stable and credible health assessment conclusion can still be formed when various state assessment results have uncertainty or conflicts, and the conclusion is used as a remote control decision basis, so that the reliability of equipment health state judgment and the safety of a remote control process are improved.

Description

Methods and systems for online health monitoring and remote control of lifting machinery and equipment Technical Field

[0001] This invention relates to the field of industrial equipment health monitoring and intelligent control technology, and in particular to methods and systems for online health monitoring and remote control of lifting machinery. Background Technology

[0002] Because lifting machinery often endures complex dynamic loads and various environmental factors during operation, its key structures and actuators are prone to fatigue damage, structural deformation, abnormal vibration, or abnormal temperature rise during long-term operation. In recent years, with the development of the Industrial Internet, wireless communication technology, and cloud computing platforms, deploying multiple types of sensors in key parts of lifting machinery and collecting equipment operating status data in real time via the network has become an important technical approach for equipment condition monitoring.

[0003] However, existing technologies still have shortcomings in applying equipment health assessment results to remote control decision-making. Traditional methods usually determine whether equipment can be remotely operated based on a single state assessment result or a simple threshold judgment. However, the assessment results of equipment operating status often come from a variety of different analysis models or monitoring indicators. In practical applications, this information often has uncertainties, conflicts, or differences in credibility. If there is a lack of an effective information fusion mechanism, it can easily lead to unstable control decision-making basis. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] This invention provides a method and system for online health monitoring and remote control of lifting machinery and equipment. It addresses the problem that traditional methods lack an effective fusion method to handle conflicts and uncertainties among multi-source status assessment information when faced with different evidence, making it difficult to form stable and reliable equipment health judgment results to support remote control decisions.

[0006] To address the aforementioned technical problems, the present invention provides the following technical solution: Firstly, the present invention provides a method for online health monitoring and remote control of lifting machinery, comprising: deploying a sensor network at key parts of the lifting machinery; collecting multi-source status data and preprocessing and extracting feature vectors; uploading the feature vectors of the multi-source status data to a remote cloud platform via a wireless communication network; and performing health level determination, credibility calculation, and abnormal trend intensity analysis, respectively; constructing the results of health level determination, credibility calculation, and abnormal trend intensity analysis into a real number BPA; and extending the real number BPA into the complex domain to obtain a complex evidence set; calculating the reliability of the complex evidence in the complex evidence set. Reliability is assessed by classifying multiple pieces of evidence into high-reliability and low-reliability evidence. Low-reliability evidence is then weighted and discounted, and multiple evidence is fused using multiple evidence combination rules. A conservative decision-making strategy is employed to generate a final health evidence value. Control requests from remote operating terminals are received, compared with the final health evidence value, and a digital authorization token is generated. Control commands carrying the digital authorization token are then securely issued, and feedback signal data during the execution of control commands is collected in real time to generate an evidence chain. During the execution of control commands, multi-source status data is re-collected, a tiered intervention strategy is implemented, and the evidence chain covering the entire lifecycle of remote control operations is archived.

[0007] As a preferred embodiment of the online health monitoring and remote control method for lifting machinery equipment described in this invention, the step of collecting multi-source state data and performing preprocessing and feature vector extraction includes: the multi-source state data includes state data of vibration, temperature, strain, load and displacement; the feature vector extraction refers to extracting time-frequency domain feature vectors from the preprocessed vibration data and calculating statistical feature vectors from other multi-source state data besides vibration signals.

[0008] As a preferred embodiment of the online health monitoring and remote control method for lifting machinery equipment described in this invention, the steps of determining health level, calculating reliability, and analyzing abnormal trend intensity include: constructing an XGBoost classification model and training it using labeled historical multi-source state data; using the trained XGBoost classification model to predict the received feature vectors, outputting the probability distribution of the equipment belonging to each health level, and selecting the highest probability distribution value as the health level assessment result; collecting strain, temperature, and displacement data under normal operating conditions of the lifting machinery equipment, calculating the mean vector and covariance matrix of the strain, temperature, and displacement data respectively, and locating them as normal clusters; calculating the Mahalanobis distance between the real-time extracted feature vectors and the normal clusters, and mapping them to a standard interval using an exponential decay function to obtain a Mahalanobis distance similarity score; acquiring the sensor's self-test status code and real-time measurement value; if the self-test status code indicates a fault, the reliability coefficient is directly set to 0; if the self-test is normal, the normalized distance between the real-time measurement value and the rated working range boundary specified in the sensor manual is calculated and converted to a linear mapping function. The reliability coefficient of the interval is calculated; the reliability coefficient and Mahalanobis distance similarity score are synthesized using a weighted average formula to obtain a credibility assessment; historical time series data of lifting machinery and equipment in health status and historical operation periods are selected, and the Holt-Winters exponential smoothing method is used to model the historical time series data to obtain a baseline prediction model; the data points collected by the sensors recently are input into the baseline prediction model to predict the value of the next time point, and the result is compared with the actual reading of the sensors to output the residuals; a sliding window is defined, and the root mean square value of the residuals within the sliding window is calculated as the strength of the abnormal trend.

[0009] As a preferred embodiment of the online health monitoring and remote control method for lifting machinery equipment described in this invention, the following steps are included: constructing the results of health level determination, credibility calculation, and abnormal trend intensity analysis into a real number BPA, and extending the real number BPA into the complex domain to obtain a complex evidence set. This includes: mapping the health level assessment results to a real number BPA identification framework to obtain health level evidence; for credibility assessment, performing uncertainty allocation through linear mapping to obtain credibility evidence; and for abnormal trend intensity, performing mapping through exponential decay mapping to obtain trend intensity evidence. The step of extending the real number BPA into the complex domain to obtain a complex evidence set refers to defining a complex basic trust allocation function based on Euler's formula for each type of evidence and proposition, extending the three types of evidence to obtain a complex evidence set.

[0010] As a preferred embodiment of the online health monitoring and remote control method for lifting machinery equipment described in this invention, the following steps are included: calculating the credibility of complex evidence in a complex evidence set and classifying the complex evidence into high-credibility evidence and low-credibility evidence, including: for any two complex pieces of evidence, calculating the complex difference vector component by component; calculating the Jaccard distance between any two propositions and constructing a conflict coefficient matrix; calculating the complex Jousselme distance between two complex pieces of evidence based on the complex difference vector and the conflict coefficient matrix; calculating the amplitude cosine similarity between any two complex pieces of evidence, and combining the complex Jousselme distance to calculate the comprehensive similarity score between any two complex pieces of evidence, and defining the credibility of each piece of complex evidence as the arithmetic mean of its comprehensive similarity scores with all other complex pieces of evidence; taking the average credibility of all complex pieces of evidence above the median as the trust threshold, classifying the complex evidence, and defining the complex evidence with credibility greater than the trust threshold as high-credibility evidence, otherwise as low-credibility evidence.

[0011] As a preferred embodiment of the online health monitoring and remote control method for lifting machinery equipment described in this invention, the following steps are included: weight discounting correction of low-confidence evidence, fusing the corrected multiple evidence using a multiple evidence combination rule, and generating a final health evidence value using a conservative decision-making strategy. This includes: setting a discount weight based on a trust threshold, correcting the low-confidence evidence using the discount weight, and adding the corrected low-confidence evidence and high-confidence evidence back into the multiple evidence; fusing the corrected multiple evidence in two stages using a multiple Dempster combination rule to obtain the final fused evidence; the initial fusion refers to... The health level evidence and credibility evidence from the positive complex evidence are fused to obtain the initial fused evidence; secondary fusion refers to fusing the initial fused evidence and trend strength evidence to obtain the final fused evidence; the modulus of the final fused evidence is extracted as the basic probability assignment value in the real number domain and normalized; based on the normalized basic probability assignment value, the credibility and uncertainty intervals are calculated respectively, and the initial health evidence value is set; special case rules are set, and if any condition of the special case rule is met, the initial health evidence value is forcibly set to 0; the initial health evidence value is linearly mapped to the target interval to obtain the final health evidence value.

[0012] As a preferred embodiment of the online health monitoring and remote control method for lifting machinery equipment according to the present invention, the step of receiving a control request from a remote operating terminal, comparing it with a final health evidence value, and generating a digital permission token includes: receiving a control request from a remote operating terminal, and querying a preset control action-security policy mapping table based on the control action type in the control request; the control request includes a target device identifier, a requested control action type, and an operating context; the control action-security policy mapping table defines a minimum health evidence threshold value, a recommended health evidence threshold value, an action validity period, and a degradation policy for each control action type; based on the current latest health information of the equipment... The timestamp corresponding to the final health evidence value is compared with the validity period of the action to determine whether the latest final health evidence value has expired. For final health evidence values ​​that have not expired, they are compared with the minimum health evidence threshold. The abnormal trend intensity corresponding to the final health evidence values ​​that meet the minimum health evidence threshold requirements is extracted and a risk level assessment is performed, including three risk levels: high, medium, and low. Based on the risk level, the control request is modified, and a digital permission token is generated. The digital permission token includes the final health evidence value on which the control instruction is based, the expiration time, and the control request. For expired final health evidence values, a new final health evidence value is generated.

[0013] As a preferred embodiment of the online health monitoring and remote control method for lifting machinery equipment described in this invention, the following steps are included: Securely issuing control commands carrying digital authorization tokens and collecting feedback signal data during the execution of control commands in real time to generate a chain of evidence. This includes: the cloud platform encapsulating the digital authorization token into a structured secure command package, generating a digital signature using an asymmetric encryption algorithm, and appending the digital signature to the end of the secure command package; before each issuance of an operation command, performing a lightweight two-way authentication based on the TLS / DTLS protocol and negotiating to generate a temporary session key unique to this session; using the temporary session key, encrypting the secure command package with the digital signature using a symmetric encryption algorithm to form a ciphertext data stream, and issuing it to the edge gateway via the wireless network; the edge gateway receiving the encrypted command... After the command is executed, the agreed-upon temporary session key is used for decryption to obtain a secure command package with a digital signature. The digital signature is then verified using a pre-configured cloud platform public key. The verified secure command package is parsed to extract control commands. Based on the parsed control commands, a dedicated edge computing node connected to the edge gateway calls the local control logic library to generate specific device-level control signals, which are then sent to the crane's actuators. Simultaneously, the edge node collects feedback signal data from the actuators in real time. As the control commands begin execution, the edge gateway sends the execution record back to the cloud platform. The execution record includes the command reception time, signature verification result, command execution start time, and feedback signal data. The cloud platform associates and binds the execution record with the digital authorization token to form a chain of evidence, which is then stored in the cloud platform database.

[0014] As a preferred embodiment of the online health monitoring and remote control method for lifting machinery and equipment described in this invention, the step of re-collecting multi-source status data during the execution of control commands, implementing a graded intervention strategy, and archiving the evidence chain covering the entire lifecycle of the remote control operation includes: after the control command is successfully issued, re-collecting multi-source status data, calculating the final health evidence value of the current control action, comparing it with the corresponding suggested health evidence threshold value, and performing downgrading, revocation, or maintenance operations on the current control action based on the comparison result; after the control command is completed, marking the entire lifecycle data of this remote control operation as completed and archiving it to the historical log database.

[0015] Secondly, this invention provides an online health monitoring and remote control system for lifting machinery, comprising: a collection and extraction module for deploying a sensor network at key parts of the lifting machinery, collecting multi-source status data, preprocessing and extracting feature vectors, and uploading the feature vectors of the multi-source status data to a remote cloud platform via a wireless communication network for health level determination, credibility calculation, and abnormal trend intensity analysis; a construction and expansion module for constructing a real number BPA from the results of health level determination, credibility calculation, and abnormal trend intensity analysis, and extending the real number BPA into the complex domain to obtain a complex evidence set; and a calculation and classification module for calculating the credibility of the complex evidence in the complex evidence set and classifying the complex evidence. The system includes high-credibility evidence and low-credibility evidence; a correction and generation module is used to correct the low-credibility evidence by applying weight discounts, fuse the corrected multiple evidence using multiple evidence combination rules, and generate a final health evidence value using a conservative decision-making strategy; a receiving and issuing module is used to receive control requests from remote operation terminals, compare them with the final health evidence value, generate a digital authorization token, securely issue control instructions carrying the digital authorization token, and collect feedback signal data during the execution of control instructions in real time to generate an evidence chain; an intervention and archiving module is used to re-collect multi-source state data during the execution of control instructions, implement a hierarchical intervention strategy, and archive the evidence chain covering the entire lifecycle of remote control operations.

[0016] The beneficial effects of this invention are as follows: This invention constructs a real number BPA from the results of health level determination, credibility calculation, and abnormal trend intensity analysis, and extends the real number BPA into a complex domain to obtain a complex evidence set; calculates the credibility of the complex evidence in the complex evidence set, classifies the complex evidence into high-credibility evidence and low-credibility evidence; applies weight discount correction to the low-credibility evidence, uses complex evidence combination rules to fuse the corrected complex evidence, and adopts a conservative decision-making strategy to generate the final health evidence value. This ensures that even when there is uncertainty or conflict in the results of multiple state assessments, a stable and credible health assessment conclusion can still be formed and used as a basis for remote control decisions, thereby improving the reliability of equipment health status judgment and the security of the remote control process. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 is a flowchart of the online health monitoring and remote control method for lifting machinery equipment in Example 1.

[0019] Figure 2 is a schematic diagram of the online health monitoring and remote control system for lifting machinery equipment in Example 1.

[0020] Figure 3 is a detailed flowchart of the generation and fusion of multiple evidences in Example 1.

[0021] Figure 4 is a timing diagram of remote control secure delivery and evidence chain generation in Example 1. Detailed Implementation

[0022] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0023] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0024] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0025] Example 1, referring to Figures 1 to 4, is the first embodiment of the present invention. This embodiment provides a method for online health monitoring and remote control of lifting machinery, including the following steps: S1, deploying a sensor network at key parts of the lifting machinery, collecting multi-source state data, preprocessing and extracting feature vectors, and uploading the feature vectors of the multi-source state data to a remote cloud platform via a wireless communication network for health level determination, reliability calculation, and abnormal trend intensity analysis; preferably, deploying a sensor network at key parts of the lifting machinery specifically refers to deploying fiber optic strain sensors or resistors at key stress points such as the mid-span of the main beam, end beams, and outriggers. Strain gauges are used to monitor micro-strain force changes; triaxial ICP-type accelerometers are deployed in the hoisting mechanism's gearbox and the travel motor bearing housing to collect wide-bandwidth (e.g., 0-5kHz) vibration acceleration signals; digital PT100 temperature sensors or infrared thermal imagers are deployed at the motor windings, brake friction pads, and key terminals of the electrical cabinet to monitor temperature data; high-precision load cells are installed at the wire rope fixing ends or hook beams, and absolute encoders or laser rangefinders are installed in the travel mechanism to collect load and displacement status data respectively; this preprocessing specifically refers to deploying a high-precision clock source on the edge gateway and transmitting it via IEEE... The 1588 Precision Time Protocol distributes synchronization clock signals to downstream smart sensor nodes, ensuring that the clock deviation across the entire network is less than 1 millisecond. For signals requiring strict phase relationships (such as strain and vibration), the edge gateway issues a unified hardware trigger pulse, commanding relevant sensor nodes to start sampling at the same physical moment, achieving hardware-level synchronous acquisition. Sliding window adaptive amplitude limiting filtering and a real-time outlier elimination algorithm based on the 3σ criterion are applied to each sensor data stream to eliminate pulse interference. Time-frequency domain feature vectors are extracted from the preprocessed vibration data. For vibration signals within the sliding window, time-domain feature vectors are calculated, including RMS, peak value, and kurtosis. The time-domain features are converted to the frequency domain using a Fast Fourier Transform to extract the characteristic frequency amplitude and sideband energy. Statistical feature vectors, including mean, standard deviation, maximum value, and minimum value, are calculated for other multi-source state data besides vibration signals.

[0026] The wireless communication network is specifically a "5G URLLC slice" or an "industrial Wi-Fi 6" channel. It should be noted that the feature vectors of multi-source state data need to be compressed and encapsulated before being uploaded to the remote cloud platform through the wireless communication network.

[0027] Furthermore, an XGBoost classification model is constructed, specifically trained based on historical operating data and known health status labels (normal, attention, abnormal, and dangerous). Its tree structure implicitly learns the complex mapping relationship between different fault modes and multi-source features during training, forming a data-driven fault tree. The trained XGBoost classification model is used to predict the received feature vectors (e.g., through a softmax function) and outputs the probability distribution of the device belonging to each health level. The highest probability distribution value is selected as the health level assessment result. Strain, temperature, and displacement data under normal operating conditions are collected, and the mean vector and covariance matrix of the strain, temperature, and displacement data are calculated and located as normal clusters. The Mahalanobis distance between the real-time extracted feature vectors and the normal clusters is calculated, and an exponential decay function is used to map the Mahalanobis distance to a standard interval (0-1), obtaining the Mahalanobis distance similarity score. The sensor's self-test status code and real-time measurement values ​​are obtained. If the self-test status... If the status code indicates a fault, the reliability coefficient is directly set to 0. If the self-test is normal, the normalized distance between the real-time measurement value and the rated working range boundary specified in the sensor manual is calculated and converted into a reliability coefficient between 0 and 1 through a linear mapping function (the closer to the center value, the higher the coefficient). Using a weighted average formula, the reliability coefficient and Mahalanobis distance similarity score are synthesized to obtain a credibility assessment (the weights use fixed weights, pre-set based on the degree of influence of the monitoring indicators on crane safety, such as 0.5:0.5). Key historical time series data (such as structural strain, hoisting mechanism vibration acceleration, and hoisting load) of the equipment in its healthy state and historical operating period are selected, and the Holt-Winters exponential smoothing method is used to model the historical time series data to obtain a baseline prediction model. The recent real-time data points are input into the baseline prediction model to predict the parameter values ​​at the next time point, and compared with the actual readings of the sensor to output the residuals. A sliding window is defined, and the root mean square value of the residuals within the sliding window is calculated as the strength of the abnormal trend.

[0028] S2. Construct a real-valued BPA from the results of health level determination, credibility calculation, and abnormal trend intensity analysis, and extend the real-valued BPA into the complex domain to obtain a complex evidence set; preferably, constructing the health level assessment results, credibility assessment, and abnormal trend intensity into a real-valued BPA means mapping the health level assessment results to a real-valued BPA recognition framework to obtain health level evidence, specifically including: Let the probabilities of the four health levels output by the XGBoost model be Q respectively. normal (Normal), Q attention (Attention), Q abnormal (Abnormal), Q danger (Danger), satisfying Q normal +Q attention +Q abnormal+Q danger =1; Q normal With Q attention The sum serves as the initial support for proposition S (suitable operation); Q is then used as the initial support for the proposition S (suitable operation). abnormal With Q danger The sum is used as the initial support for proposition ¬S (unsuitable for operation); considering the uncertainty of the model classification itself, an uncertainty allocation coefficient δ (0 < δ < 1, typical value is 0.1) is introduced to reduce the initial support by a ratio of (1-δ), and the reduced part is allocated to the uncertain proposition Θ to obtain the real number BPA of the health level evidence: m H (S)=(Q normal +Q attention )×(1-δ); m H (¬S)=(Q abnormal +Q danger )×(1-δ); m H (Θ)=δ;where, m H (S), m H (¬S) and m H (Θ) represents the evidence H for health level in the proposition. (Suitable for operation) (Not suitable for operation) and The magnitude of the uncertainty, δ, can be set based on the classification accuracy of the model validation set. The higher the classification accuracy, the smaller the value of δ. When the model classification accuracy reaches 95% or higher, δ can be 0.05; when the classification accuracy is between 80% and 95%, δ can be 0.1; when the classification accuracy is below 80%, δ can be 0.15-0.2. The health level evidence includes suitable operation scores, unsuitable operation scores, and uncertain operation scores. For credibility assessment, uncertainty is allocated through linear mapping to obtain credibility evidence. Let the credibility assessment value be... ( This value is calculated using a weighted average formula; a linear mapping method is used to assign certainty and uncertainty to the credibility assessment, as shown in the formula: ; ; ;in, , and Respective evidence of credibility In the proposition (Suitable for operation) (Not suitable for operation) and (Uncertain) magnitude; For the allocation coefficient ( Typical value is ), used to balance the and Support for distribution The closer The more neutral one's inclination towards either positive or negative judgments, the better. The closer This indicates a greater tendency to support when the information is credible. Proposition, As a credibility assessment value, the physical meaning of this mapping method is: when credibility... High (close to) ), assigned to explicit propositions ( and The overall reliability is high, and the uncertainty is low; when the reliability is high... Low (close to 0), with most reliability assigned to uncertain propositions. This indicates that a clear judgment cannot be made due to unreliable data; for the strength of abnormal trends, evidence of trend strength is obtained by mapping using the exponential decay mapping method, formula: ; ; ;in, , and Evidence of trend strength In the proposition , and The range, It is a natural exponential function. This is a constant greater than 0, called the attenuation coefficient, which controls the strength of the abnormal trend s. The decay rate, Let be a constant between 0 and 1, representing the fundamental uncertainty, indicating that regardless of the trend, a portion of the confidence is always allocated to uncertainty; the complex domain extension of the real number BPA yields a complex evidence set, including: for each piece of evidence and proposition, a complex fundamental confidence allocation function is defined based on Euler's formula, with the following formula: ;in, As a source of evidence Multiple pieces of evidence, The modulus is represented by the amplitude. Let f be the phase factor, and let f be the focal length of 1, where f is the focal length of 1. Representative sources of evidence include H, C, and Three sources of evidence, including H, C, and These are used to refer to evidence of health level, evidence of credibility, and evidence of trend strength, respectively. The imaginary unit, For Jiao Yuan, Let the phase angle be the phase angle, where the health level evidence phase is... , and For the phase allocation of credible evidence, the credibility assessment value is used. Perform a linear mapping, as shown in the following formula: ; ; ;in, , and Respective evidence of credibility In the proposition (Suitable for operation) (Not suitable for operation) and (Uncertain) phase; Credibility assessment value Its definition and calculation method are consistent with those mentioned above, trend strength evidence phase, , and , wherein Let be the trend phase modulation function, where The modulation amplitude coefficient limits the maximum phase offset to no more than 45 degrees to avoid overmodulation. For symbolic functions, ,in The change in the abnormal trend intensity value s represents the difference between the current time and the previous time.

[0029] It should be noted that the phase angle in the above multiple pieces of evidence The assignment of values ​​is not arbitrary, but a systematic engineering design based on the information representation requirements of evidence theory and the geometric characteristics of the complex field. Fundamentally, it uses the geometric position of the phase to encode the semantic information and dynamic characteristics of different evidence sources in the decision space.

[0030] The specific design is based on the following: the health level evidence comes from the probability output of the XGBoost classification model, which is essentially a static classification result of the current state of the device. Therefore, its phase design follows the binary opposition principle: the phase of proposition S (suitable for operation) is set to θ. H (S)=0 corresponds to the positive real axis direction in the complex plane, symbolizing a "clear and certain" decision-making tendency; the phase of proposition ¬S (unsuitable for operation) is set to θ. H (¬S)=π, corresponding to the negative real axis direction in the complex plane, symbolizes a decision-making tendency of "explicit negation"; the 180° phase difference between 0 and π represents completely opposite decision semantics in the complex domain, facilitating the rapid identification of fundamental conflicts between evidence through phase comparison during subsequent fusion; the phase of the uncertain proposition Θ is set as θ. H (Θ)=0 because uncertain information itself does not carry a clear decision-making tendency, so it is placed in the positive real axis direction as a neutral benchmark; the credibility assessment integrates the sensor self-test status and Mahalanobis distance similarity, reflecting the degree of confidence in the current data quality.

[0031] Its phase design introduces a reliability modulation factor (1) The purpose is to encode the metadata of "confidence level" as a phase offset: When credibility When it is higher (close to 1), (1) ) close to 0, θ C (S) approaching 0 indicates that the "fitness for operation" judgment of highly credible evidence is consistent with the positive direction of health level evidence; when credibility... At lower levels, (1 As θ increases C (S) shifts in the negative direction, symbolizing a decrease in confidence in the "suitable operation" judgment due to unreliable data; Similarly, low confidence causes the phase of "unsuitable for operation" to shift from π in the positive direction, symmetrical to the shift direction of S, maintaining the geometric symmetry of the decision space; the phase shift is limited to the range of [-π / 2, π / 2], ensuring that confidence modulation does not lead to a fundamental reversal of the decision semantics (i.e., it does not cause the phase of S to cross the quadrant and enter the region of ¬S); θ C (Θ)=0: The phase of the uncertain proposition remains at 0, serving as a stable reference point during fusion.

[0032] The intensity of the abnormal trend reflects the direction and rate of dynamic evolution of the equipment status over time. In terms of phase design, a trend phase modulation function Φ(s, Δs) is introduced to encode the trend information as a dynamic phase adjustment amount: θ T (S)=0: When trend evidence supports "suitable for action", maintain alignment with the affirmative direction; θ T (¬S) = π + Φ(s, Δs): Based on the basic negation direction π, a trend modulation amount Φ(s, Δs) is superimposed so that the phase "unsuitable for operation" can reflect the deterioration or mitigation of the trend; Φ(s, Δs) = sgn(Δs)·π / 4·tanh(s); where sgn(Δs) encodes the trend direction (Δs>0 indicates deterioration, Δs<0 indicates mitigation); tanh(s) maps the abnormal trend intensity s (original range is [0, +∞)) to the interval (0, 1) to ensure that the phase modulation amount is bounded; π / 4 (45°) is the maximum phase offset, which, after engineering verification, can achieve the best balance between fully characterizing the trend influence and avoiding over-modulation leading to semantic confusion in decision-making. The 45° offset is sufficient to be effectively identified by subsequent similarity calculations, and will not cause the phase of ¬S to cross into the quadrant region of S; when the trend is stable (Δs=0), Φ=0, and the phase reverts to the baseline negation direction π; for uncertain propositions Θ, the phase always remains θ. T(Θ)=0; Through the above systematic phase design, the amplitude mE(ψ) of complex evidence continues to play the traditional role of confidence level, while the phase θ E (ψ) then takes on the coding function of "reliability properties".

[0033] It includes multi-dimensional semantic information such as evidence source, credibility level, and dynamic trend; while the complex evidence model, which uses amplitude to represent the strength of credibility and phase to represent the credibility attribute, enables subsequent complex evidence similarity calculation, conflict identification, and fusion decision-making to be carried out on a richer information dimension, thereby significantly improving the accuracy of multi-source evidence processing.

[0034] S3. Calculate the credibility of the complex evidence in the complex evidence set, and classify the complex evidence into high-credibility evidence and low-credibility evidence; preferably, for any two complex pieces of evidence, calculate the complex difference vector component by component; calculate the Jaccard distance between any two propositions and construct a conflict coefficient matrix; based on the complex difference vector and the conflict coefficient matrix, calculate the complex Jousselme distance between two pieces of complex evidence, using the formula: ;in, Multiple evidence With multiple evidence The complex Jousselme distance between them The operator for taking the real part of a complex number. Multiple evidence With multiple evidence The conjugate transpose of the complex difference vector between them. The sign for conjugate transpose. The conflict coefficient matrix is ​​used to calculate the amplitude cosine similarity between any two complex pieces of evidence. Combined with the complex Jousselme distance, the comprehensive similarity score between any two complex pieces of evidence is calculated using the following formula: ;in, Multiple evidence With multiple evidence The overall similarity between them For example, if the attenuation coefficient is greater than 0, then... This attenuation coefficient is used to control the strength of the effect of distance on similarity; it can be set through expert experience. Multiple evidence With multiple evidence The amplitude cosine similarity between them and Multiple evidences and The magnitude vector, where , Similarly, the credibility of each piece of complex evidence is defined as the arithmetic mean of its overall similarity score with all other complex evidence. The average credibility score of all complex evidence above the median is taken as the trust threshold. Complex evidence is then classified, with the complex evidence whose credibility score is greater than the trust threshold defined as high-credibility evidence, and the rest as low-credibility evidence.

[0035] S4. Apply weight discounting to low-credibility evidence, fuse the corrected multiple evidence using the multiple evidence combination rule, and generate the final health evidence value using a conservative decision-making strategy. Preferably, only low-credibility evidence is corrected to avoid unnecessary correction of high-credibility evidence and to effectively suppress the negative impact of low-credibility evidence. Set the discount weights using the formula: ;in, Multiple evidence Discount weight, Multiple evidence The credibility of the evidence (i.e., the arithmetic mean of the overall similarity scores of the evidence to all other evidence). The dynamic trust factor is defined as follows: In the formula, A collection of highly credible evidence. This is the arithmetic mean of the credibility of all evidence in the highly credible evidence set. This represents the median credibility of all complex evidence. The physical meaning of this discount weight is: for highly credible evidence with a credibility level greater than or equal to the dynamic trust factor, its original weight is maintained. For low-credibility evidence with a credibility level lower than the dynamic trust factor, the credibility level is reduced proportionally to the dynamic trust factor; the lower the credibility level, the smaller the discount weight. The low-credibility evidence is then corrected using the discount weight, as shown in the formula: ;in, For the corrected multiple evidence , For the newly added uncertain part, From the clear proposition The discounted portion of the reliability score, stemming from the low credibility of the evidence, can no longer support any explicit proposition and is therefore reassigned to the "uncertain" item. , This indicates that its phase is 0, meaning that this newly added uncertainty is purely cognitive uncertainty and does not carry dynamic characteristic information; the corrected low-confidence evidence and high-confidence evidence are added back to the complex evidence; the corrected complex evidence is fused twice using the complex Dempster combination rule to obtain the final fused evidence; the first fusion refers to fusing the health level evidence and confidence evidence in the corrected complex evidence to obtain the initial fused evidence; the second fusion refers to fusing the initial fused evidence and the trend strength evidence to obtain the final fused evidence; where, the complex Dempster combination rule, for the first fusion, has the formula: ; ; ;in, The initial fusion evidence for Jiao Yuan A is obtained by fusing health level evidence and credibility evidence. and Represent , and Jiao Yuan, The complex conflict coefficient, and These are the revised health grade evidence and credibility evidence for Jiao Yuan, with scores of B and C, respectively. and The magnitudes of the revised health grade evidence and credibility evidence for foci B and C, respectively. and These are the phases of the corrected health grade evidence and credibility evidence, respectively, with focal lengths of B and C; (The above is incomplete and requires further context.) The intersection operation of sets follows these rules: ; ; ; The above The specific empty set represents a conflict; the complex Dempster combination rule for secondary fusion is consistent with the first fusion described above; it should be noted that the application of the above complex Dempster combination rule requires clarification of the following operational basis: regarding the representation of focal elements, the focal elements in this system include single-element focal elements. , , and possible composite focal elements such as (Indicates suitability or uncertainty) (Indicating unsuitability or uncertainty), when the BPA assignment of evidence involves compound focal elements, the standard definition of Dempster-Shafer theory should be followed, i.e., a compound focal element represents support for any one of the propositions, and the intersection operation rule for compound focal elements is: for any two focal elements... and Their intersection Defined as simultaneously belonging to and The set of propositions, with the following specific operational rules: If and All are single-element focal elements, calculated according to established rules; if It is a composite coke element. It is a single-element focal element if and only if ,but ,otherwise ;like and Since they are all composite coke elements, then and That is, taking the intersection of two sets of propositions; any focal element and The intersection is equal to the focal element itself; regarding the normalization process after phase combination, since the phase angle may exceed the limit after additive combination... The interval needs to be normalized to... Within the scope, the normalization method is as follows: ,in It is a four-quadrant arctangent function, ensuring that the phase angle is always within the range of the four quadrants. Within the interval; regarding the interpretation of the combined results, the modulus of the fused complex evidence represents the normalized comprehensive support for the proposition, and the dynamic credibility characteristics and quantification deviation of the fused phase angle-encoded evidence; the dynamic credibility characteristics of the fused phase angle-encoded evidence refer to the definition that a fused phase angle equal to zero corresponds to high-credibility steady-state evidence (such as a calibration reference source). If the fused phase angle is greater than zero, it indicates a trend of deterioration in the implicit performance or a decrease in credibility, and less than zero indicates a trend of improvement in performance or an increase in credibility; the fused phase angle reflects the degree of quantification deviation, meaning that the absolute value of the fused phase angle reflects the significance of the deviation from the reference. The larger the value, the more significant the deviation of the evidence from the ideal steady state; after the second fusion, the conflict value of the second fusion is calculated using the formula: ;in, For conflicting values, Evidence for the corrected trend strength. As a representative The focal elements are determined; the magnitude (amplitude) of the final fused evidence is extracted as the basic probability allocation value in the real domain and normalized (the magnitude of the complex evidence of each focal element is normalized so that its sum is 1, while keeping their respective phase angles unchanged); based on the normalized basic probability allocation, the confidence and uncertainty intervals are calculated respectively, and the initial health evidence value is set, using the formula: ; ;in: This is the initial health evidence value; To the proposition (Suitable for operation) reliability, after normalization Basic probability distribution of propositions express; To the proposition The similarity is calculated by taking the first term of all terms in the formula as a function of similarity to the real number. The sum of the normalized probability assignment values ​​of the focal elements in the non-empty intersection, with the second term being the phase modulation term; This is the reliability weighting coefficient. This represents the uncertainty penalty coefficient. The phase weighting coefficient has a range of values. Typical value is , For the final fusion of evidence Phase value of the proposition; This represents the maximum permissible phase offset. The uncertainty interval reflects the uncertainty regarding... The degree of uncertainty in propositional cognition, To identify any focal element in the frame.

[0036] and The determination method is as follows: Reliability weight coefficient This reflects the system's emphasis on explicit supporting evidence.

[0037] The value can be calibrated using historical operating data of the equipment: collect historical data of the equipment in known health states, and calculate the value for each state. The value was determined through regression analysis. The mapping relationship between the actual health status of the equipment makes satisfy It can best fit the actual health status of the equipment.

[0038] Typically, when the device is in perfect health, It should be close to 1, at which point we hope... Approaching the upper limit, therefore Can be set to When the equipment is in a critical state, if you want right Sensitive to changes It can be adjusted appropriately.

[0039] Uncertainty penalty coefficient It reflects the degree to which the system avoids cognitive uncertainty. The value can be determined according to the safety level requirements of the equipment: for high-risk operation scenarios (such as hoisting molten metal), a higher value should be taken. To strengthen the penalty for uncertainty; for general operational scenarios, a moderate value can be taken ( For low-risk scenarios, a lower value can be used. ).

[0040] Adjustment The value makes The output can conservatively reflect the actual risks and avoid misjudgments caused by uncertainty.

[0041] In practical applications, for specific models of lifting machinery and typical operating conditions, semi-physical simulation or field tests can be used to evaluate... and Perform calibration and then embed it into the system; set special case rules to force the initial health evidence value to 0 if any of the following conditions are met: If the support level is too low, the initial health evidence value will be forcibly set to 0. This is the reliability threshold, a parameter based on engineering experience and set according to safety requirements; for example, a value of 0.1. This means the probability of explicitly supporting safe operation is less than 10%; if If the uncertainty is too high, then the initial health evidence value is forcibly set to 0. This is an uncertainty threshold, a safety decision boundary set based on evidence theory, such as a value of 0.8. This means that over 80% of the probability quality is assigned to the uncertain term, indicating that the system's understanding of the device's state is very vague. This could be due to problems such as severely conflicting sensor data, contradictory results from multiple sub-state assessments, or the device being in an abnormal but undefined fault mode. , indicating the maximum of the conflict value, where The conflict threshold is set based on a balance between numerical stability and safety boundaries; for example, a value of 0.1. mean A conflict value close to 1 indicates severe conflict in the evidence; at this point, the normalization factor... It tends to infinity, which can lead to numerical instability and the Zadeh paradox. Therefore, we force the initial health evidence value to be 0. We then linearly map the initial health evidence value to the target interval ([0, 100]) to obtain the final health evidence value.

[0042] S5. Receive a control request from a remote operating terminal, compare it with the final health evidence value, generate a digital authorization token, securely issue control instructions carrying the digital authorization token, and collect feedback signal data during the execution of the control instructions in real time to generate an evidence chain; Receive a control request from a remote operating terminal, and query a preset control action-safety policy mapping table based on the control action type in the control request; The control request includes a target device identifier, the requested control action type (e.g., "speed limit to 30%", "execute emergency stop", etc.), and the operation context (e.g., job task ID and operator ID); The control action-safety policy mapping table defines a minimum health evidence threshold value, a recommended health evidence threshold value, an action validity period, and a degradation strategy for each control action type; The minimum health evidence threshold value is specifically the minimum stable quantification value of the device required to execute the action type; The recommended health evidence threshold value is specifically the recommended value for executing the action type under ideal conditions; Action The validity period, specifically, is the maximum permitted duration of the action from the time of health evidence assessment (e.g., a speed limit command is valid for 10 minutes, and a shutdown command is valid until manual reset); the degradation strategy, specifically, is the more conservative action the system should automatically take when conditions are not met (e.g., a "load limit" request can be downgraded to "alarm and lock"); the timestamp generated based on the device's latest final health evidence value is compared with the action validity period to determine if the latest final health evidence value has expired; for unexpired final health evidence values, they are compared with the minimum health evidence threshold value, and the abnormal trend intensity corresponding to the final health evidence values ​​that meet the minimum health evidence threshold value requirement (final health evidence values ​​greater than the minimum health evidence threshold value) is extracted for risk level assessment, including high, medium, and low risk levels, and the control request is modified based on the risk level to generate a digital permission token; the risk level assessment specifically involves setting a risk coefficient based on the abnormal trend intensity and the current health evidence value, using the formula: ;in The risk coefficient is a dimensionless comprehensive risk indicator. The abnormal trend strength value is calculated from the root mean square of the residuals within the sliding window, reflecting the degree to which the equipment status deviates from the historical baseline; The current health evidence value is calculated, with a value range of [0, 100], reflecting the overall health level of the equipment; Numerical stability constant ( This is used to avoid numerical instability caused by the denominator approaching 0; The calibration coefficient is used to map the risk coefficient to a standardized range that facilitates setting thresholds.

[0043] In other words, the risk coefficient and the strength of the abnormal trend Proportional to (the stronger the abnormal trend, the higher the risk), and related to the current health evidence value. Inversely proportional (the lower the equipment health, the higher the risk).

[0044] Calibration coefficient The method for determining it is as follows: The determination needs to be based on the equipment's historical operating data and calibrated using statistical analysis methods.

[0045] The specific steps are as follows: collect data on the equipment during a continuous period (e.g., 30 days) while it is operating normally. and Data, calculation The statistical distribution of the ratio; calculate the mean of the ratio. and standard deviation Set standardized targets to determine the risk coefficient under normal operating conditions. The target interval falls within a convenient range, such as [0, 5] or [0, 10]. In this embodiment, the upper limit of the target interval is set to 5. Calculate... value: That is, to ensure more than 95% under normal operating conditions. The value should not exceed 5; for different models of lifting machinery or different working conditions, repeat the above steps to obtain targeted results. The value is pre-configured in the system.

[0046] Regarding risk level classification, based on the calculated... The value, combined with the statistical quantiles of the equipment's historical risk coefficient data, sets two dynamic thresholds. and Take the 70th percentile of historical data value; Take the 90th percentile of historical data Value; The risk level is divided into: low risk ), medium risk ( ) and high risk ( ).

[0047] The aforementioned thresholds can be recalculated and updated periodically (e.g., monthly) based on the accumulated operating time of the equipment to adapt to long-term changes such as equipment aging. Two dynamic thresholds are preset based on the statistical quantiles (e.g., 70th and 90th percentiles) of the equipment's historical risk coefficient data. After comparing the risk coefficients, the intensity of abnormal trends is divided into three risk levels: high, medium, and low. Control commands at the low risk level do not require modification. For control commands at the medium risk level, the intensity parameters (e.g., speed, acceleration, load) in the control command are multiplied by 0.9. If the scaled parameter is lower than the minimum health evidence threshold for the action, a degradation strategy is automatically triggered, and the dynamic threshold is adjusted accordingly. The validity period of control commands is shortened to 50%. For high-risk control commands, the intensity parameters in the control command are multiplied by 0.7. If the scaled parameter is lower than the minimum health evidence threshold for the action, a downgrade strategy is automatically triggered, and the validity period of the action is shortened to 20%, with additional safety measures. Specifically, before the command is issued, a second pop-up confirmation is required on the remote console. Only after confirmation can a digital authorization token be generated, and the on-site safety officer is simultaneously notified for monitoring. The digital authorization token includes the final health evidence value on which it is based, the expiration date, and the control request. For expired final health evidence values, a new final health evidence value is generated.

[0048] Furthermore, the cloud platform encapsulates the digital authorization token into a structured security instruction package and generates a digital signature (using the cloud platform's private key) using an asymmetric encryption algorithm (such as RSA), appending the digital signature to the end of the security instruction package. Before each operation instruction is issued, a lightweight two-way authentication (such as based on a device certificate) is performed based on the TLS / DTLS protocol, and a temporary session key unique to this session is negotiated and generated. Using the temporary session key, the security instruction package with the digital signature is encrypted using a symmetric encryption algorithm (such as AES-256), forming a ciphertext data stream, which is then transmitted to the edge gateway via the wireless network. The edge gateway receives the encrypted instruction. Then, the negotiated temporary session key is used to decrypt the data to obtain a security instruction packet with a digital signature, and the digital signature is verified using a pre-set cloud platform public key. The verified security instruction packet is then parsed to extract control commands. Based on the parsed control commands, the dedicated edge computing node (such as an industrial PLC or industrial control computer) connected to the edge gateway calls the local control logic library to generate specific device-level control signals (such as analog output, relay on / off, CAN bus messages) and sends them to the crane's actuators (such as frequency converters and brake controllers). At the same time, the edge node will collect feedback signal data from the actuators in real time (such as actual speed, current, and contactor status).

[0049] As the control command begins execution, the edge gateway sends the execution record back to the cloud platform. The execution record includes the command reception time, signature verification result, command execution start time, and feedback signal data. The cloud platform associates and binds the execution record with the digital authorization token to form a chain of evidence, and stores this chain of evidence in the cloud platform database.

[0050] S6. During the execution of control commands, multi-source status data is re-collected, a tiered intervention strategy is implemented, and the evidence chain covering the entire lifecycle of remote control operations is archived. After the control command is successfully issued, multi-source status data is re-collected, the final health evidence value of the current control action is calculated, compared with the corresponding recommended health evidence threshold value, and based on the comparison result, a downgrade, revocation, or maintenance operation is performed on the current control action. If the final health evidence value of the current control action is lower than the recommended health evidence threshold value of the current control action, but higher than the threshold value of a lower-level action (such as downgrading from "speed limit" to "stop"), then... The system automatically triggers a permission downgrade strategy. If the final health evidence value of the current control action is lower than the minimum health evidence threshold value for all actions, the system immediately and forcibly revoks all remote control permissions and triggers a local audible and visual alarm, completely returning control of the equipment to the on-site safety mechanism. If the final health evidence value of the current control action is higher than the recommended health evidence threshold value for the current control action and the validity period has not expired, the current control permissions are maintained. After the control command is executed, the entire lifecycle data of this remote control operation (decision, issuance, execution, and intervention) is marked as completed and archived to the historical log database.

[0051] This embodiment also provides an online health monitoring and remote control system for lifting machinery, including: a collection and extraction module, used to deploy a sensor network at key parts of the lifting machinery, collect multi-source status data, preprocess and extract feature vectors, and upload the feature vectors of the multi-source status data to a remote cloud platform via a wireless communication network, performing health level determination, credibility calculation, and abnormal trend intensity analysis respectively; a construction and expansion module, used to construct a real number BPA from the results of health level determination, credibility calculation, and abnormal trend intensity analysis, and to extend the real number BPA into the complex domain to obtain a complex evidence set; and a calculation and classification module, used to calculate the credibility of the complex evidence in the complex evidence set, classify the complex evidence, and include... It includes high-credibility evidence and low-credibility evidence; a correction and generation module is used to correct the weight discount of low-credibility evidence, fuse the corrected multiple evidence using multiple evidence combination rules, and generate the final health evidence value using a conservative decision-making strategy; a receiving and issuing module is used to receive control requests from remote operation terminals, compare them with the final health evidence value, generate a digital authorization token, securely issue control instructions carrying the digital authorization token, and collect feedback signal data in real time during the execution of control instructions to generate an evidence chain; an intervention and archiving module is used to re-collect multi-source state data during the execution of control instructions, implement a hierarchical intervention strategy, and archive the evidence chain covering the entire lifecycle of remote control operations.

[0052] This embodiment also provides a computer device applicable to the online health monitoring and remote control method for lifting machinery, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the online health monitoring and remote control method for lifting machinery as proposed in the above embodiment.

[0053] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0054] This embodiment also provides a storage medium storing a computer program. When executed by a processor, the program implements the method for online health monitoring and remote control of lifting machinery as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0055] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for online health monitoring and remote control of lifting machinery and equipment, characterized in that, include: A sensor network is deployed at key parts of the lifting machinery to collect multi-source status data. This data is preprocessed and feature vectors are extracted. The feature vectors from the multi-source status data are then uploaded to a remote cloud platform via a wireless communication network for health level determination, credibility calculation, and abnormal trend intensity analysis. The results of these analyses are constructed as a real-valued BPA, and the real-valued BPA is extended to the complex domain to obtain a complex evidence set. The credibility of the complex evidence in the complex evidence set is calculated, and the complex evidence is classified into high-credibility evidence and low-credibility evidence. We apply weight discounts to low-credibility evidence, fuse the corrected multiple evidence using multiple evidence combination rules, and generate the final health evidence value using a conservative decision-making strategy. It receives control requests from remote operation terminals, compares them with the final health evidence value, generates a digital authorization token, securely issues control instructions carrying the digital authorization token, and collects feedback signal data in real time during the execution of control instructions to generate a chain of evidence. During the execution of control commands, multi-source status data is re-collected, a tiered intervention strategy is implemented, and an evidence chain covering the entire lifecycle of remote control operations is archived.

2. The method for online health monitoring and remote control of lifting machinery and equipment as described in claim 1, characterized in that, The process of collecting multi-source state data and performing preprocessing and feature vector extraction includes: the multi-source state data includes state data of vibration, temperature, strain, load and displacement; the feature vector extraction refers to extracting time-frequency domain feature vectors from the preprocessed vibration data and calculating statistical feature vectors from other multi-source state data besides the vibration signal.

3. The method for online health monitoring and remote control of lifting machinery and equipment as described in claim 2, characterized in that, The steps of determining health level, calculating reliability, and analyzing abnormal trend intensity include: constructing an XGBoost classification model and training it using labeled historical multi-source state data; using the trained XGBoost classification model to predict the received feature vectors, outputting the probability distribution of the equipment belonging to each health level, and selecting the highest probability distribution value as the health level assessment result; collecting strain, temperature, and displacement data under normal operating conditions of the lifting machinery, calculating the mean vector and covariance matrix of the strain, temperature, and displacement data respectively, and locating them as normal clusters; calculating the Mahalanobis distance between the real-time extracted feature vectors and the normal clusters, and mapping them to a standard interval using an exponential decay function to obtain a Mahalanobis distance similarity score; acquiring the sensor's self-test status code and real-time measurement value; if the self-test status code indicates a fault, the reliability coefficient is directly set to 0; if the self-test is normal, the normalized distance between the real-time measurement value and the rated working range boundary specified in the sensor manual is calculated and converted to a linear mapping function. The reliability coefficient of the interval is calculated; the reliability coefficient and Mahalanobis distance similarity score are synthesized using a weighted average formula to obtain a credibility assessment; historical time series data collected from the lifting machinery and equipment in its health state and historical operation period are selected, and the Holt-Winters exponential smoothing method is used to model the historical time series data to obtain a baseline prediction model; the data points collected by the recent sensors are input into the baseline prediction model to predict the value at the next time point, and the value is compared with the actual reading of the sensors to output the residuals; a sliding window is defined, and the root mean square value of the residuals within the sliding window is calculated as the strength of the abnormal trend.

4. The method for online health monitoring and remote control of lifting machinery and equipment as described in claim 3, characterized in that, The process of constructing a real-valued BPA (Basic Trust Assignment) from the health level assessment, credibility calculation, and abnormal trend intensity analysis results, and then extending the real-valued BPA into the complex domain to obtain a complex evidence set, includes: mapping the health level assessment results to the real-valued BPA identification framework to obtain health level evidence; for credibility assessment, allocating uncertainty through linear mapping to obtain credibility evidence; and for abnormal trend intensity, mapping through exponential decay mapping to obtain trend intensity evidence. The extension of the real-valued BPA into the complex domain to obtain the complex evidence set refers to defining a complex basic trust assignment function based on Euler's formula for each type of evidence and proposition, extending the three types of evidence to obtain a complex evidence set.

5. The method for online health monitoring and remote control of lifting machinery and equipment as described in claim 4, characterized in that, The credibility of complex evidence in a set of complex evidence is calculated, and the complex evidence is classified into high-credibility evidence and low-credibility evidence. This includes: for any two complex pieces of evidence, calculating the complex difference vector component by component; calculating the Jaccard distance between any two propositions and constructing a conflict coefficient matrix; calculating the complex Jousselme distance between two pieces of evidence based on the complex difference vector and the conflict coefficient matrix; calculating the magnitude cosine similarity between any two pieces of evidence, and combining this with the complex Jousselme distance to calculate the comprehensive similarity score between any two pieces of evidence. The credibility of each piece of evidence is defined as the arithmetic mean of its comprehensive similarity scores with all other pieces of evidence; the average credibility score above the median among all pieces of evidence is taken as the trust threshold, and the complex evidence is classified accordingly. Complex evidence with a credibility score greater than the trust threshold is defined as high-credibility evidence, otherwise it is considered low-credibility evidence.

6. The method for online health monitoring and remote control of lifting machinery and equipment as described in claim 5, characterized in that, The process of weight discounting and correcting low-credibility evidence, fusing the corrected complex evidence using a complex evidence combination rule, and generating a final health evidence value using a conservative decision-making strategy includes: setting discount weights based on a trust threshold, correcting low-credibility evidence using the discount weights, and adding the corrected low-credibility and high-credibility evidence back into the complex evidence; fusing the corrected complex evidence twice using a complex Dempster combination rule to obtain the final fused evidence; initial fusion, which involves fusing health level evidence and credibility evidence from the corrected complex evidence to obtain initial fused evidence; secondary fusion, which involves fusing the initial fused evidence and trend strength evidence to obtain the final fused evidence; extracting the modulus of the final fused evidence as the basic probability allocation value in the real domain and normalizing it; calculating the credibility and uncertainty intervals based on the normalized basic probability allocation value, and setting the initial health evidence value; setting a special case rule, whereby the initial health evidence value is forcibly set to 0 if any condition of the special case rule is met; and linearly mapping the initial health evidence value to the target interval to obtain the final health evidence value.

7. The method for online health monitoring and remote control of lifting machinery and equipment as described in claim 6, characterized in that, The process of receiving a control request from a remote operating terminal, comparing it with a final health evidence value, and generating a digital permission token includes: receiving a control request from a remote operating terminal; querying a preset control action-security policy mapping table based on the control action type in the control request; the control request includes a target device identifier, the requested control action type, and an operation context; the control action-security policy mapping table defines a minimum health evidence threshold, a recommended health evidence threshold, an action validity period, and a degradation policy for each control action type; comparing the timestamp generated based on the device's latest final health evidence value with the action validity period to determine if the latest final health evidence value has expired; for unexpired final health evidence values, comparing them with the minimum health evidence threshold, extracting the abnormal trend intensity corresponding to the final health evidence values ​​that meet the minimum health evidence threshold requirements, performing a risk level assessment, including high, medium, and low risk levels, and modifying the control request based on the risk level to generate a digital permission token; the digital permission token includes the final health evidence value on which the control instruction is based, the expiration time, and the control request; for expired final health evidence values, regenerating a new final health evidence value.

8. The method for online health monitoring and remote control of lifting machinery and equipment as described in claim 7, characterized in that, The process of securely issuing control commands carrying digital authorization tokens and collecting feedback signal data during the execution of control commands in real time to generate a chain of evidence includes: the cloud platform encapsulating the digital authorization token into a structured secure command package, generating a digital signature using an asymmetric encryption algorithm, and appending the digital signature to the end of the secure command package; before each issuance of an operation command, performing a lightweight two-way authentication based on the TLS / DTLS protocol and negotiating to generate a temporary session key unique to this session; using the temporary session key, encrypting the secure command package with the digital signature using a symmetric encryption algorithm to form a ciphertext data stream, and issuing it to the edge gateway via the wireless network; after receiving the encrypted command, the edge gateway decrypts it using the negotiated temporary session key. The system obtains a secure instruction packet with a digital signature and verifies the digital signature using a pre-set cloud platform public key. The verified secure instruction packet is then parsed to extract control commands. Based on the parsed control commands, a dedicated edge computing node connected to the edge gateway calls the local control logic library to generate specific device-level control signals, which are then sent to the crane's actuators. Simultaneously, the edge node collects feedback signal data from the actuators in real time. As the control commands begin execution, the edge gateway sends an execution record back to the cloud platform. This execution record includes the instruction reception time, signature verification result, instruction execution start time, and feedback signal data. The cloud platform associates and binds the execution record with a digital authorization token to form a chain of evidence, which is then stored in the cloud platform database.

9. The method for online health monitoring and remote control of lifting machinery and equipment as described in claim 5, characterized in that, During the execution of control commands, multi-source state data is re-collected, a tiered intervention strategy is implemented, and an evidence chain covering the entire lifecycle of the remote control operation is archived. This includes: after the control command is successfully issued, multi-source state data is re-collected, the final health evidence value of the current control action is calculated, compared with the corresponding recommended health evidence threshold value, and based on the comparison result, downgrade, revocation, or maintenance operations are performed on the current control action; after the control command is completed, the entire lifecycle data of this remote control operation is marked as completed and archived to the historical log database.

10. A health online monitoring and remote control system for lifting machinery, based on the health online monitoring and remote control method for lifting machinery according to any one of claims 1 to 9, characterized in that, include: The collection and extraction module is used to deploy a sensor network in key parts of the lifting machinery, collect multi-source status data, preprocess and extract feature vectors, and upload the feature vectors of the multi-source status data to a remote cloud platform through a wireless communication network for health level determination, credibility calculation and abnormal trend intensity analysis. An extension module is constructed to convert the results of health level determination, confidence calculation, and abnormal trend intensity analysis into real number BPAs, and to extend the real number BPAs into the complex domain to obtain a complex evidence set. The classification module calculates the credibility of multiple pieces of evidence in a set of multiple pieces of evidence and classifies the multiple pieces of evidence into high-credibility evidence and low-credibility evidence. The correction generation module is used to correct the weight discount of low-credibility evidence, fuse the corrected multiple evidence using the multiple evidence combination rule, and generate the final health evidence value using a conservative decision-making strategy. The receiving and sending module is used to receive control requests from remote operation terminals, compare them with the final health evidence value, generate a digital authorization token, securely send control instructions carrying the digital authorization token, and collect feedback signal data in real time during the execution of control instructions to generate a chain of evidence. The intervention archiving module is used to re-collect multi-source status data during the execution of control commands, implement hierarchical intervention strategies, and archive the evidence chain covering the entire lifecycle of remote control operations.

Images