A turntable fault diagnosis method based on adaptive joint weighting

Abstract

The present invention relates to a turntable fault diagnosis method based on adaptive joint weighting. The method comprises the following steps: step 1, collecting real-time data of the turntable during operation; step 2, normalizing the six types of turntable data in step 1; step 3, calculating the weight of the normalized data in step 2 to the turntable fault; step 4, designing a turntable fault diagnosis algorithm based on adaptive joint weighting; step 5, embedding the designed adaptive joint weighted turntable fault diagnosis algorithm into the information processing module of the turntable host computer to complete the real-time monitoring of the turntable fault type. The present invention uses six types of data, namely, main power supply voltage, phase current, torque current, motor speed, motor temperature, and encoder feedback, as characteristic indicators for predicting turntable faults, and combines the tracking of turntable fault types through an adaptive joint weighting algorithm, to establish a model that can accurately track turntable fault trends and realize the prediction of turntable fault types.

Description

Technical Field

[0001] The invention relates to the field of turntables, and in particular designs a turntable fault diagnosis method based on adaptive joint weighting. Background Art

[0002] The turntable is a complex mechatronics integrated device that performs physical simulation and analog testing in the automotive, shipborne, and aerospace fields. It plays a key role in the development of various automotive, shipborne, and aerospace products. It can simulate various attitude and angular motions, reproduce its various dynamic characteristics during motion, and repeatedly test the performance of the guidance system, control system, and corresponding components of automotive, shipborne, and aerospace products to obtain sufficient test data, and redesign and improve the system based on the data to meet the performance index requirements of the overall design.

[0003] There is a complex nonlinear mapping relationship between the fault bottom event and the symptom signal of the turntable, which limits the accurate positioning of the fault by traditional diagnostic technology. The development of fault diagnosis technology has mainly gone through three stages: manual diagnosis, modern diagnosis and intelligent diagnosis. Up to now, fault diagnosis methods can be divided into methods based on analytical models, methods based on signal processing and methods based on knowledge. In recent years, data mining technology based on artificial intelligence has been widely used in the research of complex system fault detection and diagnosis, including neural networks, support vector machines, rough sets, intelligent optimization algorithms and their combination. In order to ensure the long-term safe and reliable operation of various electronic, mechanical, software and other components, it is of great significance to improve the research on fault diagnosis technology. Summary of the invention

[0004] In order to solve the above problems, the present invention proposes a turntable fault diagnosis method based on adaptive joint weighting on the basis of various data measured by the turntable. The present invention tracks the turntable fault by combining the weight of each data information index with the model state monitoring parameter and the system weighting parameter to realize the prediction of the turntable fault type. To achieve this purpose, the specific steps of the present invention are as follows, which are characterized by:

[0005] Step 1: Collect real-time data of the turntable during operation. The measured data of the turntable include: main power supply voltage, phase current, torque current, motor speed, motor temperature, encoder feedback, attach the corresponding fault type as a label to the data, and divide the data into training samples and test samples;

[0006] Step 2, normalizing the six types of turntable data in step 1 to between 0 and 1, thereby reducing the turntable fault diagnosis model;

[0007] Step 3, calculating the weight of the normalized data in step 2 for the turntable fault, and respectively calculating the weight of the main power supply voltage, phase current, torque current, motor speed, motor temperature, and encoder feedback for the fault category;

[0008] Step 4: Design a turntable fault diagnosis algorithm based on adaptive joint weighting: Adopt an adaptive weighting algorithm to dynamically adjust the weight coefficient w in the prediction stage according to the historically collected turntable measured data. j , to update the model condition monitoring parameters and system weighted parameters to achieve the prediction of turntable fault types;

[0009] Step 5, embed the designed adaptive joint weighted turntable fault diagnosis algorithm into the information processing module of the turntable host computer to complete the real-time monitoring of the turntable fault type.

[0010] Furthermore, the corresponding fault type in step 1 is expressed as follows:

[0011] The fault types corresponding to the turntable are: 24V power supply overvoltage, 24V power supply undervoltage, main power supply overvoltage, main power supply undervoltage, short circuit overcurrent fault, phase current overcurrent fault, torque current overlimit fault, overspeed fault, overtemperature fault, I2T heating fault, IT overcurrent fault, resolver fault, incremental encoder fault, absolute encoder fault, ground fault, motor locking fault, software current overlimit, motor model identification fault, parameter calculation fault, BISSC encoder fault, parameter value abnormal overlimit fault, encoder UVW fault.

[0012] Furthermore, the data normalization process in step 2 is expressed as follows:

[0013] The main power supply voltage, phase current, torque current, motor speed, motor temperature, and encoder feedback data of the turntable measured data in the turntable sample are normalized by the following formula

[0014]

[0015] In the formula, x' i is the measured data of the turntable, x min is the minimum value of the turntable measured data, x max is the maximum value of the turntable measured data, x i is the normalized turntable measured data, and i is the sampling time of the data.

[0016] Furthermore, in step 3, the process of normalizing the six types of turntable data in step 2 can be expressed as follows:

[0017] Step 3.1 Calculate the distance d between the turntable data and the turntable fault classification label:

[0018]

[0019] in is the test feature sample, is the feature sample of the training set, and n is the number of sampling points;

[0020] Step 3.2 Weight the n normalized turntable measured data and sum them to obtain the jth data weight w j , and its calculation formula is as follows

[0021] As shown in formula 3;

[0022]

[0023] Where d represents the distance between the normalized turntable measured data and the turntable fault type, and ε is the weight factor.

[0024] Furthermore, the process of turntable fault diagnosis based on adaptive joint weighting in step 4 can be expressed as follows:

[0025] Step 4.1 Initialize a set of turntable measured data x'0 j and 6 system weighting parameters θ0 j , where j = 1, ..., 6, and the prediction start time is determined to be k, that is, θ0 j =θ k j 、x'0 j =x' k j ;

[0026] Step 4.2 Update the system weighted parameter point θ at the next moment according to the weighted mean and covariance of the system weighted parameter point k+1 j :

[0027] Step 4.2.1 Calculate the weighted average of the system weighted parameters and covariance

[0028]

[0029]

[0030] Step 4.2.2 Weighted average of parameter points and system weighted parameter points Find the mean of the beta distribution at time k

[0031]

[0032] Where h is a constant of 0.3;

[0033] Step 4.2.3 Obtain the parameters of the Beta distribution and Constructing the Beta distribution function

[0034]

[0035]

[0036] Step 4.2.4: Beta distribution function constructed in step 5.2.3 Sampling is performed in the system to obtain the updated system weighted parameter θ k+1 j ;

[0037] Step 4.3 Update the model state monitoring parameters according to the updated system weighted parameter points

[0038]

[0039] In the formula, f -1 () is the mapping function;

[0040] Step 4.4: According to step 1.4, the system weighted parameters and the model state monitoring parameters at the previous moment are used to iteratively update the system weighted parameters and the model state monitoring parameters;

[0041] Step 4.5 Predict turntable fault type based on turntable data

[0042] The turntable fault type predicted at time k0+1 is expressed as follows:

[0043]

[0044] In the formula, y k0+1 is the fault type label of the turntable at time k0+1, g -1 () is the anti-normalization function.

[0045] The present invention provides a turntable fault diagnosis method based on adaptive joint weighting, which has the following beneficial effects:

[0046] 1. The present invention uses six kinds of data, namely, main power supply voltage, phase current, torque current, motor speed, motor temperature and encoder feedback, as characteristic indicators for predicting turntable faults, and can accurately predict turntable faults.

[0047] 2. The present invention combines the tracking of turntable fault types through an adaptive joint weighted algorithm to establish a model that can accurately track turntable fault trends and realize the prediction of turntable fault types.

[0048] 3. The algorithm of the present invention has low complexity and strong real-time performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 Flowchart of the present invention;

[0050] Figure 2 Data communication block diagram of the present invention. DETAILED DESCRIPTION

[0051] The present invention proposes a turntable fault diagnosis method based on adaptive joint weighting, which aims to complete the prediction of the fault type of the electric turntable and improve the accuracy of the prediction of the turntable fault type. Figure 1 The present invention is further described below in conjunction with the accompanying drawings and specific embodiments:

[0052] Step 1: Collect the real-time data of the turntable during operation through RTX communication. The turntable communication block diagram is as follows: Figure 2 As shown, the measured data of the turntable includes: main power supply voltage, phase current, torque current, motor speed, motor temperature, encoder feedback, and the corresponding fault type is attached to the data as a label, and the data is divided into training samples and test samples;

[0053] Furthermore, the corresponding fault type in step 1 is expressed as follows:

[0054] The fault types corresponding to the turntable are: 24V power supply overvoltage, 24V power supply undervoltage, main power supply overvoltage, main power supply undervoltage, short circuit overcurrent fault, phase current overcurrent fault, torque current overlimit fault, overspeed fault, overtemperature fault, I2T heating fault, IT overcurrent fault, resolver fault, incremental encoder fault, absolute encoder fault, ground fault, motor locking fault, software current overlimit, motor model identification fault, parameter calculation fault, BISSC encoder fault, parameter value abnormal overlimit fault, encoder UVW fault.

[0055] Step 2, normalizing the six types of turntable data in step 1 to between 0 and 1, thereby reducing the turntable fault diagnosis model;

[0056] Furthermore, the data normalization process in step 2 is expressed as follows:

[0057] The main power supply voltage, phase current, torque current, motor speed, motor temperature, and encoder feedback data of the turntable measured data in the turntable sample are normalized by the following formula

[0058]

[0059] In the formula, x' i is the measured data of the turntable, x minis the minimum value of the turntable measured data, x max is the maximum value of the turntable measured data, x i is the normalized turntable measured data, and i is the sampling time of the data.

[0060] Step 3, calculating the weight of the normalized data in step 2 for the turntable fault, and respectively calculating the weight of the main power supply voltage, phase current, torque current, motor speed, motor temperature, and encoder feedback for the fault category;

[0061] Furthermore, in step 3, the process of normalizing the six types of turntable data in step 2 can be expressed as follows:

[0062] Step 3.1 Calculate the distance d between the turntable data and the turntable fault classification label:

[0063]

[0064] in is the test feature sample, is the feature sample of the training set, and n is the number of sampling points;

[0065] Step 3.2 Weight the n normalized turntable measured data and sum them to obtain the jth data weight w j , and its calculation formula is as follows

[0066] As shown in formula 3;

[0067]

[0068] Where d represents the distance between the normalized turntable measured data and the turntable fault type, and ε is the weight factor.

[0069] Step 4: Design a turntable fault diagnosis algorithm based on adaptive joint weighting: Adopt an adaptive weighting algorithm to dynamically adjust the weight coefficient w in the prediction stage according to the historically collected turntable measured data. j , to update the model condition monitoring parameters and system weighted parameters to achieve the prediction of turntable fault types;

[0070] Furthermore, the process of turntable fault diagnosis based on adaptive joint weighting in step 4 can be expressed as follows:

[0071] Step 4.1 Initialize a set of turntable measured data x'0 j and 6 system weighting parameters θ0 j , where j = 1, ..., 6, and the prediction start time is determined to be k, that is, θ0 j =θ k j 、x'0 j =x' k j ;

[0072] Step 4.2 Update the system weighted parameter point θ at the next moment according to the weighted mean and covariance of the system weighted parameter point k+1 j :

[0073] Step 4.2.1 Calculate the weighted average of the system weighted parameters and covariance

[0074]

[0075]

[0076] Step 4.2.2 Weighted average of parameter points and the system weighted parameter point θ k j Find the mean of the beta distribution at time k

[0077]

[0078] Where h is a constant of 0.3;

[0079] Step 4.2.3 Obtain the parameters of the Beta distribution and Constructing the Beta distribution function

[0080]

[0081]

[0082] Step 4.2.4: Beta distribution function constructed in step 5.2.3 Sampling is performed in the system to obtain the updated system weighted parameter θ k+1 j ;

[0083] Step 4.3 Update the model state monitoring parameters according to the updated system weighted parameter points

[0084]

[0085] In the formula, f -1 () is the mapping function;

[0086] Step 4.4: According to step 1.4, the system weighted parameters and the model state monitoring parameters at the previous moment are used to iteratively update the system weighted parameters and the model state monitoring parameters;

[0087] Step 4.5 Predict turntable fault type based on turntable data

[0088] The turntable fault type predicted at time k0+1 is expressed as follows:

[0089]

[0090] In the formula, y k0+1 is the fault type label of the turntable at time k0+1, g -1 () is the anti-normalization function.

[0091] Step 5, embed the designed adaptive joint weighted turntable fault diagnosis algorithm into the information processing module of the turntable host computer to complete the real-time monitoring of the turntable fault type.

[0092] The above description is only a preferred embodiment of the present invention and does not constitute any other form of limitation to the present invention. Any modification or equivalent change made based on the technical essence of the present invention still falls within the scope of protection required by the present invention.

Claims

1. A turntable fault diagnosis method based on adaptive joint weighting, the specific steps are as follows, characterized in that: Step 1: Collect real-time data of the turntable during operation. The measured data of the turntable include: main power supply voltage, phase current, torque current, motor speed, motor temperature, encoder feedback, attach the corresponding fault type as a label to the data, and divide the data into training samples and test samples; Step 2, normalizing the six types of turntable data in step 1 to between 0 and 1, thereby reducing the turntable fault diagnosis model; Step 3, calculating the weight of the normalized data in step 2 for the turntable fault, and respectively calculating the weight of the main power supply voltage, phase current, torque current, motor speed, motor temperature, and encoder feedback for the fault category; Step 4: Design a turntable fault diagnosis algorithm based on adaptive joint weighting: Adopt an adaptive weighting algorithm to dynamically adjust the weight coefficient w in the prediction stage according to the historically collected turntable measured data. j , to update the model condition monitoring parameters and system weighted parameters to achieve the prediction of turntable fault types; Step 5: embed the designed adaptive joint weighted turntable fault diagnosis algorithm into the information processing module of the turntable host computer to complete the real-time monitoring of the turntable fault type; the corresponding fault type in step 1 is expressed as follows: The fault types corresponding to the turntable: 24V power supply overvoltage, 24V power supply undervoltage, main power supply overvoltage, main power supply undervoltage, short circuit overcurrent fault, phase current overcurrent fault, torque current overlimit fault, overspeed fault, overtemperature fault, I2T Overheating fault, IT overcurrent fault, resolver fault, incremental encoder fault, absolute encoder fault, ground fault, motor locking fault, software current over limit, motor model identification fault, parameter calculation fault, BISSC encoder fault, parameter value abnormal over limit fault, encoder UVW fault; the data normalization processing in step 2 is expressed as follows: The main power supply voltage, phase current, torque current, motor speed, motor temperature, and encoder feedback data of the turntable measured data in the turntable sample are normalized by the following formula In the formula, x' i is the measured data of the turntable, x min is the minimum value of the turntable measured data, x max is the maximum value of the turntable measured data, x i is the normalized turntable measured data, i is the sampling time of the data; in step 3, The process of normalizing the six types of turntable data can be expressed as follows: Step 3.1 Calculate the distance d between the turntable data and the turntable fault classification label: in is the test feature sample, is the feature sample of the training set, and n is the number of sampling points; Step 3.2 Weight the n normalized turntable measured data and sum them to obtain the jth data weight w j , the calculation formula is shown in formula 3; Where d represents the distance between the normalized turntable measured data and the turntable fault type, and ε is the weight factor; the turntable fault diagnosis process based on adaptive joint weighting in step 4 can be expressed as follows: Step 4.1 Initialize a set of turntable measured data x'0 j and 6 system weighting parameters θ0 j , where j = 1, ..., 6, and the prediction start time is determined to be k, that is, θ0 j =θ k j 、x'0 j =x' k j ; Step 4.2 Update the system weighted parameter point θ at the next moment according to the weighted mean and covariance of the system weighted parameter point k+1 j :Step 4.2.1 Calculate the weighted average of the system weighted parameters and covariance Step 4.2.2 Weighted average of parameter points and the system weighted parameter point θ k j Find the mean of the beta distribution at time k Where h is a constant of 0.3; Step 4.2.3 Obtain the parameters of the Beta distribution and Constructing the Beta distribution function Step 4.2.4: Beta distribution function constructed in step 5.2.3 Sampling is performed in the system to obtain the updated system weighted parameter θ k+1 j ; Step 4.3 Update the model state monitoring parameters according to the updated system weighted parameter points In the formula, f -1 () is the mapping function; Step 4.4: According to step 1.4, the system weighted parameters and the model state monitoring parameters at the previous moment are used to iteratively update the system weighted parameters and the model state monitoring parameters; Step 4.5 Predict turntable fault type based on turntable data The turntable fault type predicted at time k0+1 is expressed as follows: In the formula, y k0+1 is the fault type label of the turntable at time k0+1, g -1 () is the anti-normalization function.

Images