Electromechanical equipment failure alarm rescue system

Abstract

The present application relates to the technical field of electromechanical equipment alarm and rescue, and particularly relates to an electromechanical equipment fault alarm and rescue system, which is arranged in parallel with a main control system of the electromechanical equipment at an electromechanical equipment control terminal, and switches the main control system and an emergency rescue system of the electromechanical equipment through a triggering module, wherein the emergency rescue system of the electromechanical equipment comprises: a state prediction module, which inputs collected electromechanical equipment operation data into a pre-constructed state prediction model to generate equipment state information; and the triggering module, which switches the main control system and the emergency rescue system of the electromechanical equipment. The present application can suspend the main control system of the electromechanical equipment when a fault occurs in the electromechanical equipment, and all current and previous preset internal and external instructions of the main control system of the electromechanical equipment are invalid, so as to avoid irrational operation behaviors of users in panic and secondary safety accidents.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of alarm and rescue of electromechanical equipment, and particularly relates to an electromechanical equipment fault alarm and rescue system. BACKGROUND

[0002] A kind of electromechanical equipment fault alarm system is disclosed in CN114594335A, including start control module, data monitoring module, data analysis module, data processing module, alarm module, short message module, PLC extension load module, display module, oil pressure temperature sensor module, lifting speed sensor module, voltage current module, vibration sensor module and communication processing module, the input end of the data monitoring module is connected to the output end of the start control module, the input end of the data analysis module is connected to the output end of the data monitoring module, the input end of the data processing module is connected to the output end of the data analysis module, the input end of the alarm module is connected to the output end of the data processing module, PLC extension load module is electrically connected to the outside of the data processing module, display module is fixedly installed on the signal output end of the data processing module.

[0003] As the above application, the existing electromechanical equipment operation system is a separate control system, which is provided with an alarm triggering device, and sends an alarm signal outward, when the equipment fails, the control system still maintains the running state, at this time, the user in panic may produce irrational operation behavior, such as emergency stop operation, or pressing operation button randomly, such disordered instruction input not only cannot effectively deal with the sudden situation, but may cause logical confusion of the control system, leading to secondary safety accidents, secondly, the existing fault handling mechanism has the problem of data disconnection, the system only has the basic remote alarm function, lacks the synchronous acquisition identification transmission capability of key fault parameters and fault types, so that the rescue personnel cannot obtain fault data before arriving at the scene, and such information gap directly leads to repeated fault diagnosis of on-site rescue, which seriously restricts the emergency rescue efficiency. SUMMARY

[0004] In order to solve the above problems, the application provides an electromechanical equipment fault alarm and rescue system.

[0005] The application adopts the following technical scheme, an electromechanical equipment fault alarm and rescue system is arranged in parallel with the main control system of the electromechanical equipment in the electromechanical equipment control terminal, and the main control system of the electromechanical equipment and the emergency rescue system of the electromechanical equipment are switched through a triggering module, the emergency rescue system of the electromechanical equipment comprises:

[0006] A data acquisition module acquires electromechanical equipment operation data in real time.

[0007] A state prediction module inputs the collected electromechanical equipment operation data into a pre-constructed state prediction model to generate equipment state information.

[0008] A triggering module receives the equipment state information or a manual button instruction to switch the electromechanical equipment main control system and the electromechanical equipment emergency rescue system.

[0009] A brake control module acquires the electromechanical equipment motion state and generates a speed reduction instruction based on the electromechanical equipment motion state.

[0010] A data processing module acquires electromechanical equipment operation data, generates a first fault feature parameter, and establishes an operation feature vector; acquires car state data, generates a second fault feature parameter, and establishes a state feature vector; and acquires electromechanical equipment maintenance history statistical features, generates a third fault feature parameter, and establishes a history feature vector.

[0011] A feature splicing module splices the operation feature vector, the state feature vector, and the history feature vector into a feature long vector, performs dimension reduction processing on the feature long vector, eliminates redundant information, and obtains a low-dimensional vector Y.

[0012] A fault prediction module inputs the low-dimensional vector Y into a pre-constructed fault type prediction model to generate an electromechanical equipment fault type, and transmits the low-dimensional vector Y and the electromechanical equipment fault type to a terminal PC to issue an alarm.

[0013] As a further description of the above technical solution: the electromechanical equipment operation data includes operation parameters and electrical parameters.

[0014] The operation parameters include the speed and acceleration of the electromechanical equipment during operation, and the operation parameters use a MEMS accelerometer to collect acceleration data of the electromechanical equipment during startup, acceleration, constant speed, and deceleration, and obtain speed and displacement information through integral processing.

[0015] The electrical parameters include voltage data and temperature data of the electromechanical equipment motor and brake, and a voltage divider or an isolation amplifier is used to acquire the voltage data of the electromechanical equipment motor and brake, and an RTD sensor is used to monitor the temperature data of the electromechanical equipment motor and brake.

[0016] As a further description of the above technical solution: the method for acquiring the electromechanical equipment motion state and generating a speed reduction instruction based on the electromechanical equipment motion state includes:

[0017] When the equipment is still running, real-time position information is acquired through an encoder and a magnetic sensor in the equipment to determine the position and motion direction of the equipment.

[0018] A "safe stop" target position is preset based on the moving direction of the equipment, and a remaining distance d between the current equipment position and the target position is calculated 剩 , and a brake distance ds of the equipment is obtained according to the current speed and the maximum achievable deceleration of the equipment;

[0019] If d remains < ds+Δ, Δ is a preset safety margin, the distance of the electromechanical equipment to the next target position is obtained, the correction braking distance is generated, and the speed reduction instruction is generated based on the correction braking distance and the running speed of the electromechanical equipment;

[0020] If d 剩 ≥ ds+Δ, the speed reduction instruction is generated based on d 剩 and the running speed of the electromechanical equipment.

[0021] As a further description of the above technical solution: the method for obtaining the running data of the electromechanical equipment and generating the first fault characteristic parameter and establishing the running characteristic vector comprises:

[0022] The speed and acceleration during the running of the electromechanical equipment, the voltage data, temperature data, noise data and vibration signal of the motor and brake of the electromechanical equipment are obtained, low-pass and band-pass filters are used to remove high-frequency noise and extract feature data;

[0023] The corresponding speed average, acceleration root mean square value, voltage root mean square value, temperature root mean square value, noise data average and vibration signal root mean square value are obtained through average value calculation and root mean square value calculation;

[0024] The running characteristic vector is established based on the feature data, and the running characteristic vector is: {speed average, acceleration root mean square value, voltage root mean square value, temperature root mean square value, noise data average, vibration signal root mean square value}.

[0025] As a further description of the above technical solution: the method for obtaining the car state data, generating the second fault characteristic parameter and establishing the state characteristic vector comprises:

[0026] The vibration data of the car is obtained through acceleration sensor collection, and the noise data during the running of the electromechanical equipment is monitored through sound sensor collection, and low-pass and band-pass filters are used to remove high-frequency noise;

[0027] All data are aligned according to the same sampling frequency through a unified time reference;

[0028] The corresponding vibration average, vibration root mean square value, noise average and noise root mean square value are obtained through average value calculation and root mean square value calculation, and the state characteristic vector is established, and the state characteristic vector is: {vibration average, vibration root mean square value, noise average, noise root mean square value}.

[0029] As a further description of the above technical solution: the method for obtaining the statistical characteristics of the electromechanical equipment maintenance history, generating the third fault characteristic parameter, and establishing the historical characteristic vector comprises;

[0030] The statistical characteristics of the electromechanical equipment are obtained from the maintenance history;

[0031] The statistical characteristics include the fault frequency, the average time interval between consecutive maintenance events, and the total number of statistical maintenance events, and the proportion of each type of fault is counted and calculated;

[0032] The maintenance history statistical characteristics are combined into a vector to establish a historical characteristic vector, and the historical characteristic vector is: {fault frequency, average time interval, total number of maintenance events, proportion of each type of fault}.

[0033] As a further description of the above technical solution: the method for combining the running characteristic vector, the state characteristic vector and the historical characteristic vector into a feature long vector comprises:

[0034] The Z-score standardization is used for each element in the running characteristic vector, the state characteristic vector and the historical characteristic vector to eliminate the dimension influence, each time window of the running characteristic vector, the state characteristic vector and the historical characteristic vector corresponds to a row of data, all features are arranged in sequence together, and a feature long vector is spliced.

[0035] As a further description of the above technical solution: the method for performing dimension reduction processing on the feature long vector to eliminate redundant information comprises:

[0036] The feature long vector is combined into a data matrix X with a size of nxd, n is the sample number, d is the feature dimension, each row of the data matrix X represents a time window, and each column represents a feature, the mean value μ is calculated for each column, then each data is subtracted from the mean value, the data is centralized at 0, and normalization is performed thereon;

[0037] The covariance matrix C of the data matrix X is calculated, and the calculation method of the covariance matrix is that each feature vector column of the data matrix X is subtracted from the mean value μ to obtain a centralized matrix Xnorm, and then the transpose thereof is obtained The original matrix is multiplied, and finally a scalar The calculation formula is: Wherein, n is the sample number, is the transpose matrix; Xnorm=X-μ, which is the value of each feature vector in the update matrix subtracted from the mean value μ.

[0038] The eigenvalue λi and the corresponding eigenvector wi of the covariance matrix are obtained by eigenvalue decomposition.

[0039] A preset cumulative contribution rate threshold GX is used. Based on this threshold, k features are selected to construct a dimensionality reduction transformation matrix P. The formula for calculating the cumulative contribution rate threshold GX is as follows: In the formula, d is the total number of features, and λi is the feature value;

[0040] The original data is projected onto a new low-dimensional space to obtain the dimension-reduced feature matrix Y, which is calculated as Y = Xnorm·P.

[0041] As a further description of the above technical solution: the training method of the state prediction model includes: obtaining the historical operating parameters of all electromechanical equipment in the area to which the electromechanical equipment belongs through the property database, wherein the historical operating parameters of the electromechanical equipment include historical operating data and corresponding equipment status information; using the historical operating parameters of the electromechanical equipment as training data for the state prediction model, using the historical operating data as input, and using the corresponding equipment status information as output to form training data pairs; building the state prediction model, and training the state prediction model based on the training data.

[0042] As a further description of the above technical solution: the training method of the fault type prediction model includes:

[0043] Historical low-dimensional vectors Y and fault types are extracted from the property's historical database. Based on these data, a fault type prediction model is constructed using a generative adversarial network. The fault types of electromechanical equipment are then obtained based on the acquired low-dimensional vectors Y and the trained fault type prediction model.

[0044] Beneficial effects:

[0045] In the above technical solution, the present invention provides an electromechanical equipment fault alarm and rescue system. By setting up the fault rescue system in parallel, the main control system of the electromechanical equipment can be suspended when the electromechanical equipment fails. All current and previously preset internal and external commands of the main control system of the electromechanical equipment are invalidated, avoiding irrational operation behavior that users may produce in a panic, such as emergency stop operation or random pressing of operation buttons. Such disordered command input not only cannot effectively deal with the emergency situation, but may also cause the control system logic to be confused, resulting in secondary safety accidents such as abnormal movement of electromechanical equipment or overload of mechanical structure.

[0046] Furthermore, by preprocessing and extracting features from data from various components of electromechanical equipment (such as motion data, electrical status, and maintenance history), a feature vector reflecting the real-time health status of the electromechanical equipment can be constructed. Multi-dimensional data fusion can capture subtle abnormal changes in equipment operation, predict potential faults in advance, and ensure accurate and timely early warning information, thereby greatly reducing the risk of accidents. By comprehensively utilizing the operating data and historical maintenance data of electromechanical equipment, and through layer-by-layer processing and fusion, a long feature vector reflecting the operating health status of electromechanical equipment can be constructed, thereby improving the accuracy and timeliness of fault early warning and emergency response. The acquired low-dimensional vector Y and the fault type of electromechanical equipment are transmitted to the terminal PC to issue an alarm, thereby enabling rescue personnel to understand the fault parameters and fault types in real time and improving rescue efficiency. Attached Figure Description

[0047] The present invention will be further explained below with reference to the accompanying drawings and embodiments:

[0048] Figure 1 This invention provides a module connection diagram for an electromechanical equipment fault alarm and rescue system.

[0049] Figure 2 This is a flowchart of a method for generating a deceleration command based on the motion state of electromechanical equipment, provided in an embodiment of the present invention.

[0050] Figure 3 A flowchart of a method for establishing historical feature vectors provided in an embodiment of the present invention;

[0051] Figure 4 This is a flowchart illustrating a method for dimensionality reduction of long feature vectors and removal of redundant information, provided in an embodiment of the present invention. Detailed Implementation

[0052] To make the technical means, creative features, objectives, and effects of this invention readily understandable, the invention is further described below with reference to specific illustrations. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0053] Example 1

[0054] Please see Figure 1 The present invention provides a technical solution: an electromechanical equipment fault alarm and rescue system, which is set in parallel with the main control system of the electromechanical equipment in the electromechanical equipment control terminal, and switches between the main control system of the electromechanical equipment and the emergency rescue system of the electromechanical equipment through a trigger module;

[0055] It should be noted that the main control system for the electromechanical equipment is responsible for the normal operation of the electromechanical equipment, including floor selection, speed control, door opening and closing, and overload alarm operation. The emergency rescue system for the electromechanical equipment has a higher priority than the main control system to ensure that it can take over immediately when a fault occurs.

[0056] The electromechanical equipment emergency rescue system includes:

[0057] The data acquisition module collects real-time operating data of electromechanical equipment;

[0058] Specifically, the electromechanical equipment operating data includes operating parameters and electrical parameters;

[0059] The operating parameters include the speed and acceleration of the electromechanical equipment during operation. These operating parameters are obtained by using a MEMS accelerometer to collect acceleration data of the electromechanical equipment during startup, acceleration, constant speed, and deceleration. The speed and displacement information are obtained through integration processing. The MEMS accelerometer is installed in the car or hoistway and does not affect the normal operation of the electromechanical equipment. It should be noted that abnormal acceleration curves or speed fluctuations may indicate problems with mechanical components (such as the traction system and braking system).

[0060] The electrical parameters include voltage and temperature data for the motor and brake of the electromechanical equipment. Voltage data for the motor and brake is obtained using a voltage divider or isolation amplifier, and temperature data for the motor and brake is monitored using an RTD sensor. It should be noted that if the voltage or power fluctuates abnormally within a certain period of time, it may indicate a fault in the electrical components, and the temperature of key components (such as motors and brakes) may be abnormally high, possibly related to overload, poor heat dissipation, or internal short circuit.

[0061] It should be noted that after the electromechanical equipment operation data is collected, it is transmitted in real time to the status prediction module through the signal conditioning module and communication interfaces such as RS485 and EtherCAT for status prediction.

[0062] The status prediction module inputs the collected electromechanical equipment operation data into the pre-built status prediction model to generate equipment status information, which includes normal status information and abnormal status information.

[0063] Specifically, the training method for the state prediction model includes: obtaining the historical operating parameters of all electromechanical equipment in the area to which the electromechanical equipment belongs through the property database. The historical operating parameters include historical operating data and corresponding equipment status information. The historical operating parameters are used as training data for the state prediction model, with the historical operating data as input and the corresponding equipment status information as output, forming a training data pair. The state prediction model is then built and trained based on the training data.

[0064] The trigger module receives equipment status information or manual button commands. When it receives abnormal status information or manual button commands, it generates trigger commands to control the electromechanical equipment to enter the rescue state and switch between the main control system and the emergency rescue system of the electromechanical equipment.

[0065] It should be noted that when the electromechanical equipment switches to the electromechanical equipment emergency rescue system, the main control system of the electromechanical equipment is suspended, and all current and previously preset internal and external commands of the main control system of the electromechanical equipment become invalid.

[0066] In this embodiment, by setting up a fault rescue system in parallel, the main control system of the electromechanical equipment can be suspended when the equipment malfunctions. All current and previously preset internal and external commands of the main control system of the electromechanical equipment become invalid, thus avoiding irrational operation behaviors that users may engage in during panic, such as emergency stop operations or random button presses. Such disordered command input not only fails to effectively deal with emergencies, but may also cause logical chaos in the control system, leading to secondary safety accidents such as abnormal movement of the electromechanical equipment or overload of the mechanical structure.

[0067] Example: When the electromechanical equipment is an elevator;

[0068] The electromechanical equipment fault alarm and rescue system is set up in parallel with the elevator main control system in the elevator control terminal. The elevator main control system and the elevator emergency rescue system are switched through the trigger module.

[0069] It should be noted that the elevator main control system is responsible for the normal operation of the elevator, including floor selection, speed control, door opening and closing, and elevator overload alarm operation. The elevator emergency rescue system has a higher priority than the elevator main control system to ensure that it can take over immediately in the event of a failure.

[0070] The electromechanical equipment fault alarm and rescue system includes:

[0071] The data acquisition module collects elevator operation data in real time.

[0072] The collected elevator operation data is input into a pre-built fault prediction model to generate elevator status information, which includes normal status information and abnormal status information.

[0073] The trigger module receives elevator status information or manual button commands. When it receives abnormal status information or manual button commands, it generates a trigger command to control the elevator to enter the rescue state and switch between the elevator main control system and the elevator emergency rescue system.

[0074] Example 2

[0075] Please see Figures 1-2 Based on the above embodiments, this invention adds a braking control module;

[0076] The braking control module acquires the motion status of the electromechanical equipment and generates a deceleration command based on the motion status.

[0077] The method for acquiring the motion state of electromechanical equipment and generating a deceleration command based on the motion state of the electromechanical equipment includes:

[0078] While the device is running, real-time position information is obtained through the encoder and magnetic sensor inside the device to determine the device's position and direction of movement.

[0079] Based on the direction of the equipment's movement, a target position for "safe docking" is preset, and the remaining distance d between the current position of the equipment and the target position is calculated. 剩 The braking distance ds of the device is obtained based on the current speed and the maximum achievable deceleration of the device; where, a is a positive value representing the deceleration and the magnitude of the braking force, while v is the operating speed of the electromechanical equipment.

[0080] If d 剩 <ds+Δ, where Δ is the preset safety margin; this indicates that the braking distance from the electromechanical equipment to the target position is insufficient. The distance from the electromechanical equipment to the next target position is obtained, a corrected braking distance is generated, and a deceleration command is generated based on the corrected braking distance and the operating speed of the electromechanical equipment.

[0081] If d 剩 If ≥ds+Δ, it means that the braking distance of the electromechanical equipment to the target is sufficient, based on d 剩 Generate a speed reduction command based on the operating speed of the electromechanical equipment.

[0082] It should be noted that the introduction of Δ (safety margin) is mainly to compensate for uncertainties and delays in the control and execution process, ensuring that the equipment can stop safely. In actual operation, due to sensor noise, data acquisition delay, and control signal execution lag, the calculated ds may underestimate the actual braking distance required. Increasing Δ can compensate for these errors, ensuring that the equipment still has enough distance to come to a complete stop after braking is initiated.

[0083] The method for generating the deceleration command is to use a PID (proportional-integral-derivative) controller, based on d 剩 The difference e between ds and ds is used to calculate and generate a deceleration command, which enables the equipment to decelerate smoothly and stop accurately at the target position.

[0084] The formula for generating deceleration commands based on a PID controller is as follows:

[0085]

[0086] In the formula, u represents the generated deceleration (or slowdown) command, which controls the actuator to apply corresponding braking force to the equipment; kp is the proportional coefficient, used to amplify e; ki is the integral coefficient, used to amplify the cumulative effect of the error over time; kd is the differential coefficient, used to amplify the rate of change of the error, reflecting the trend of the error; and ∫edt represents the integral of e over a period of time. Let e ​​be the derivative of e, which is the rate of change of the error over time.

[0087] In this embodiment, by comparing the remaining distance with the theoretical stopping distance plus the safety margin Δ in real time, the system can start braking before the equipment is fully close to the target position, ensuring sufficient distance for smooth deceleration and avoiding the equipment "rushing out" of the predetermined safety area due to braking delay. By adopting closed-loop PID control and other methods, the braking process is made smoother, reducing the impact force on mechanical parts and structures, thereby extending the service life of the equipment and reducing the additional wear and failure risk caused by emergency stops.

[0088] Example 3

[0089] Please see Figures 1-4 This embodiment adds a data processing module, a feature splicing module, and a fault prediction module to the above embodiments.

[0090] Data processing module: acquires the operating data of electromechanical equipment, generates the first fault feature parameter, and establishes the operating feature vector; acquires the car status data, generates the second fault feature parameter, and establishes the status feature vector; acquires the maintenance history statistical features of the electromechanical equipment, generates the third fault feature parameter, and establishes the historical feature vector.

[0091] Methods for acquiring operational data of electromechanical equipment, generating first fault characteristic parameters, and establishing operational feature vectors include:

[0092] The system acquires data on the speed and acceleration of the electromechanical equipment during operation, as well as voltage, temperature, noise, and vibration signals of the motors and brakes. Low-pass and band-pass filters are used to remove high-frequency noise and extract feature data.

[0093] The corresponding average velocity, root mean square acceleration, root mean square voltage, root mean square temperature, average noise data, and root mean square vibration signal are obtained through average value calculation and root mean square value calculation.

[0094] An operational feature vector is established based on feature data, wherein the operational feature vector is: {average velocity, root mean square acceleration, root mean square voltage, root mean square temperature, average noise data, and root mean square vibration signal}.

[0095] The methods for acquiring car status data, generating second fault characteristic parameters, and establishing a status feature vector include:

[0096] Vibration data of the car is acquired by collecting data through an accelerometer; noise data of the electromechanical equipment during operation is collected by a sound sensor, and low-pass and band-pass filters are used to remove high-frequency noise.

[0097] It should be noted that vibration energy or noise data exceeding the normal baseline may indicate mechanical wear, guide rail misalignment, or loose components.

[0098] All sensors align all data at the same sampling frequency using a unified time reference.

[0099] The corresponding vibration average value, vibration root mean square value, noise average value, and noise root mean square value are obtained by calculating the average value and root mean square value, and a state feature vector is established. The state feature vector is: {vibration average value, vibration root mean square value, noise average value, and noise root mean square value}.

[0100] like Figure 3 As shown, the methods for obtaining the historical statistical characteristics of the maintenance of the electromechanical equipment, generating the third fault characteristic parameter, and establishing the historical feature vector include:

[0101] Obtain statistical characteristics of electromechanical equipment from maintenance history;

[0102] The statistical features include the frequency of failures and the average time interval between consecutive maintenance events, as well as the total number of maintenance events, and the counting and calculation of the proportion of various types of failures.

[0103] A historical feature vector is established by composing the statistical features of maintenance history into a vector. The historical feature vector is: {fault frequency, average time interval, total number of maintenance events, and percentage of each type of fault}.

[0104] The feature concatenation module concatenates the running feature vector, the state feature vector, and the historical feature vector into a long feature vector. The long feature vector is then subjected to dimensionality reduction processing to remove redundant information, resulting in a low-dimensional vector Y.

[0105] Methods for concatenating the running feature vector, state feature vector, and historical feature vector into a single long feature vector include:

[0106] Z-score standardization is applied to each element in the running feature vector, state feature vector, and historical feature vector to eliminate the influence of dimensions. Each time window of the running feature vector, state feature vector, and historical feature vector corresponds to a row of data. All features are arranged in order and concatenated into a long feature vector, which is represented as: {average velocity, root mean square acceleration, root mean square voltage, root mean square temperature, average noise data, root mean square vibration signal, average vibration, root mean square vibration, average noise, root mean square noise, fault frequency, average time interval, total number of maintenance events, and percentage of each type of fault}.

[0107] Specifically, by preprocessing and extracting features from data from various components of electromechanical equipment (such as motion data, electrical status, and maintenance history), a feature vector reflecting the real-time health status of the electromechanical equipment can be constructed. Multi-dimensional data fusion can capture subtle abnormal changes in equipment operation, predict potential faults in advance, and ensure accurate and timely early warning information, thereby greatly reducing the risk of accidents. By comprehensively utilizing the operating data and historical maintenance data of electromechanical equipment, and through layer-by-layer processing and fusion, a long feature vector reflecting the operating health status of electromechanical equipment can be constructed, thereby improving the accuracy and timeliness of fault early warning and emergency response.

[0108] Methods for dimensionality reduction and redundant information removal of the feature long vector include:

[0109] The feature vectors are combined into a data matrix X (size n×d, where n is the number of samples and d is the feature dimension). Each row of the data matrix X represents a time window and each column represents a feature. The mean μ of each column is calculated, and then each data point is subtracted from the mean to make the data converge to 0 and normalize it.

[0110] Calculate the covariance matrix C of the data matrix X; the method for calculating the covariance matrix is ​​as follows: subtract the mean μ from each eigenvector column of the data matrix X to obtain the centered matrix Xnorm, and then transpose it to obtain... Multiply by the original matrix, and finally multiply by the scalar. The calculation formula is as follows: Where n is the number of samples, Xnorm is the transpose of Xnorm; Xnorm = X - μ is the centering matrix.

[0111] The covariance matrix is ​​decomposed into eigenvalues ​​to obtain eigenvalues ​​λi and corresponding eigenvectors wi;

[0112] Eigenvalue decomposition formula: Cwi=λiwi; It should be noted that the eigenvalue wi represents the amount of information contained in each eigenvector;

[0113] A preset cumulative contribution rate threshold GX is used. Based on this threshold, k features are selected to construct a dimensionality reduction transformation matrix P. The formula for calculating the cumulative contribution rate threshold GX is as follows: In the formula, d is the total number of features, and λi is the feature value;

[0114] The original data is projected onto a new low-dimensional space to obtain the dimension-reduced feature matrix Y, which is calculated as Y = Xnorm·P.

[0115] It should be noted that the low-dimensional vector Y for each time window can be regarded as a representation of the current operating health status of the electromechanical equipment.

[0116] Specifically, by compressing high-dimensional feature vectors into low-dimensional representations, redundant information is removed while the main variation information is retained, thereby reducing computational complexity and accelerating the response speed of the fault prediction module. The dimensionality-reduced data is easier to process in real time, providing accurate and rapid support for fault early warning and improving the overall system security.

[0117] The fault prediction module inputs the low-dimensional vector Y into the pre-built fault type prediction model to generate the fault type of electromechanical equipment. The obtained low-dimensional vector Y and the fault type of electromechanical equipment are then transmitted to the terminal PC to issue an alarm.

[0118] Specifically, the training methods for fault type prediction models include:

[0119] Historical low-dimensional vectors Y and fault types are extracted from the property's historical database. Based on these data, a fault type prediction model is constructed using a generative adversarial network. The fault types of electromechanical equipment are then obtained based on the acquired low-dimensional vectors Y and the trained fault type prediction model.

[0120] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.

Claims

1. An electromechanical device failure alarm rescue system, characterized by, It is parallelly arranged with the main control system of the electromechanical equipment in the electromechanical equipment control terminal, and switches the main control system of the electromechanical equipment and the emergency rescue system of the electromechanical equipment through the trigger module, the emergency rescue system of the electromechanical equipment comprises: A data acquisition module acquires real-time electromechanical equipment operation data; A state prediction module inputs the collected electromechanical equipment operation data into a pre-constructed state prediction model to generate equipment state information; A trigger module receives equipment state information or manual button instructions to switch the main control system of the electromechanical equipment and the emergency rescue system of the electromechanical equipment; A brake control module acquires the motion state of the electromechanical equipment and generates a speed reduction instruction based on the motion state of the electromechanical equipment; The method for acquiring the motion state of the electromechanical equipment and generating the speed reduction instruction based on the motion state of the electromechanical equipment comprises: When the equipment is still running, real-time position information is acquired through an encoder and a magnetic sensor in the equipment to determine the position and motion direction of the equipment; Based on the moving direction of the device, a target position of "safe stop" is preset, and a remaining distance d between the current device position and the target position is calculated 剩， The braking distance d of the device is obtained according to the current speed and the maximum achievable deceleration of the device s ; If d 剩 < d s + Δ, Δ is a preset safety margin, the distance of the electromechanical device to the next target position is acquired, a correction braking distance is generated, and a speed reduction instruction is generated based on the correction braking distance and the running speed of the electromechanical device; If d 剩 ≥ d s + Δ, a speed reduction command is generated based on d 剩 and the electromechanical device operating speed. A data processing module acquires electromechanical equipment operation data, generates first fault feature parameters, and establishes an operation feature vector; acquires car state data, generates second fault feature parameters, and establishes a state feature vector; and acquires electromechanical equipment maintenance history statistical features, generates third fault feature parameters, and establishes a history feature vector; The feature splicing module splices the running feature vector, the state feature vector and the historical feature vector into a feature long vector, performs dimension reduction processing on the feature long vector, eliminates redundant information, and obtains a low-dimensional vector ; The fault prediction module will use low-dimensional vectors Inputting the data into a pre-built fault type prediction model generates fault types for electromechanical equipment, and the obtained low-dimensional vectors are then processed. The fault type of the electromechanical equipment is transmitted to the terminal PC to issue an alarm.

2. A system according to claim 1, wherein the system is configured to determine the location of the mechanical or electrical device by triangulation. The electromechanical equipment operation data comprises operation parameters and electrical parameters; The operation parameters comprise the speed and acceleration of the electromechanical equipment during operation, and the operation parameters acquire acceleration data of the electromechanical equipment during startup, acceleration, constant speed, and deceleration through a MEMS accelerometer, and obtain speed and displacement information through integral processing; The electrical parameters comprise voltage data and temperature data of the electromechanical equipment motor and brake, and the voltage data of the electromechanical equipment motor and brake is acquired through a voltage divider or an isolation amplifier, and the temperature data of the electromechanical equipment motor and brake is monitored through an RTD sensor.

3. An electromechanical device failure alarm rescue system according to claim 1, wherein The method for acquiring the electromechanical equipment operation data, generating the first fault feature parameters, and establishing the operation feature vector comprises: The speed and acceleration of the electromechanical equipment during operation, the voltage data and temperature data of the electromechanical equipment motor and brake, noise data, and vibration signals are acquired, high-frequency noise is removed through a low-pass and band-pass filter, and feature data is extracted; The corresponding speed average, acceleration root mean square, voltage root mean square, temperature root mean square, noise data average, and vibration signal root mean square are acquired through average value calculation and root mean square value calculation; The operation feature vector is established based on the feature data, and the operation feature vector is: {speed average, acceleration root mean square, voltage root mean square, temperature root mean square, noise data average, vibration signal root mean square}.

4. An electromechanical device failure alarm rescue system according to claim 1, wherein The method for acquiring the car state data, generating the second fault feature parameters, and establishing the state feature vector comprises: Vibration data of the car is acquired through an acceleration sensor, and noise data during operation of the electromechanical equipment is monitored through a sound sensor, and high-frequency noise is removed through a low-pass and band-pass filter; All data is aligned at the same sampling frequency through a unified time reference by each sensor; The vibration average value, the vibration root mean square value, the noise average value and the noise root mean square value are obtained by average value calculation and root mean square value calculation, and a state feature vector is established, and the state feature vector is: {vibration average value, vibration root mean square value, noise average value, noise root mean square value}.

5. An electromechanical device failure alarm rescue system according to claim 1, wherein The method for obtaining the statistical characteristics of the electromechanical equipment from the maintenance history, generating the third fault feature parameter, and establishing the historical feature vector comprises: The statistical characteristics of the electromechanical equipment are obtained from the maintenance history. The statistical characteristics comprise a fault frequency and an average time interval between consecutive maintenance events, and a total number of maintenance events, and each type of fault is counted and a proportion is calculated; The maintenance history statistical characteristics are combined into a vector to establish a historical feature vector, and the historical feature vector is: {fault frequency, average time interval, total number of maintenance events, proportion of each type of fault}.

6. An electromechanical device failure alarm rescue system according to claim 1, wherein The method for splicing the operation feature vector, the state feature vector and the historical feature vector into a feature long vector comprises: Z-score standardization is used on each element in the operation feature vector, the state feature vector and the historical feature vector to eliminate the influence of the dimension, each time window of the operation feature vector, the state feature vector and the historical feature vector corresponds to a row of data, all features are arranged in sequence together, and a feature long vector is spliced.

7. An electromechanical device failure alarm rescue system according to claim 6, wherein The method for performing dimension reduction processing on the feature long vector to eliminate redundant information comprises: The feature long vectors are composed into a data matrix X with the size of × , is the number of samples, is the feature dimension, each row of the data matrix X represents a time window, and each column represents a feature. The mean value μ is calculated for each column, and then each data is subtracted from the mean value to make the data centralized at 0 and normalized. The covariance matrix C of the data matrix X is calculated by subtracting the mean μ of each eigenvector column of the data matrix X to obtain a centralised matrix , which is transposed to obtain , multiplied by the original matrix, and finally multiplied by the scalar ; Eigenvalues of the covariance matrix are obtained by eigenvalue decomposition and corresponding eigenvectors ; The preset accumulated contribution rate threshold GX is selected based on the accumulated contribution rate threshold of the features, and a dimension reduction transformation matrix is constructed ; Projecting the original data to a new low-dimensional space to obtain a feature matrix after dimension reduction .

8. An electromechanical device failure alarm rescue system according to claim 1, wherein The training method of the state prediction model comprises: obtaining electromechanical equipment historical operation parameters of all electromechanical equipment in the area to which the electromechanical equipment belongs through a property database, the electromechanical equipment historical operation parameters comprising historical operation data and corresponding equipment state information, taking the electromechanical equipment historical operation parameters as training data of the state prediction model, taking the historical operation data as input and taking the corresponding equipment state information as output to form a training data pair, and building the state prediction model and training the state prediction model according to the training data.

9. An electromechanical device failure alarm rescue system according to claim 1, wherein The training method of the fault type prediction model comprises: Extracting a historical low-dimensional vector from a property history database and the fault type, based on the historical low-dimensional vector and the fault type, constructing a fault type prediction model using an adversarial generative network; according to the obtained low-dimensional vector and the trained fault type prediction model, obtaining the fault type of the electromechanical equipment.

Images