Dredger sensor health state diagnosis method and system

Abstract

The invention provides a dredger sensor health state diagnosis method and system, and belongs to the technical field of ship equipment health management.The dredger sensor health state diagnosis method comprises the following steps that N recent continuous sampling points are selected, a data mean value and a standard deviation in a sliding window are calculated, and a threshold confidence interval is defined; if the current sampling value continuously falls into the threshold confidence interval for M times; data such as mud pump power, mud density, pressure and flow velocity are collected in real time through the data collection unit, the mean value and the standard deviation are analyzed and calculated through the signal preprocessing unit, and then the fault type is judged in combination with the dynamic threshold value adaptation module, the soil texture classification strategy module and the redundancy check module. Alarm signals such as disconnection, water inflow and border crossing are triggered according to priorities through an alarm output unit, and the problems of mismatching of static thresholds, poor soil adaptability, high false alarm rate and the like in the prior art are solved by utilizing dynamic threshold adjustment, multi-dimensional data fusion and redundancy check mechanisms.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of ship equipment health management, and particularly relates to a dredger sensor health state diagnosis method and system. BACKGROUND

[0002] In the dredging operation of a trailing suction hopper dredger, various sensors such as mud density sensors, pressure sensors, and flow sensors are core components for ensuring the safe operation of equipment and process control. However, due to the complex dredging environment, such as mud fluctuations, equipment vibrations, and soil diversity, sensors are easily affected by external interference or internal faults, leading to abnormal data. Traditional fault diagnosis methods have the following defects: 1. Static threshold dependence: fixed threshold is used to judge abnormalities, which cannot adapt to the dynamic characteristics of mud density and flow rate changing with soil quality; 2. Single-point fault sensitivity: lack of redundancy checking mechanism, a single sensor fault can easily cause false alarms or missed alarms; 3. Poor environmental adaptability: the differential influence of different soil types (clay, sand, and silt) on sensor data is not considered, resulting in high misjudgment rate; 4. Insufficient real-time performance: the sliding window parameters are fixed, making it difficult to balance noise suppression and abnormal response speed. Therefore, there is an urgent need for a self-adaptive and high-robustness sensor health state diagnosis method to improve the reliability and safety of dredging operations. SUMMARY

[0003] The present application provides an intelligent and adaptive dredger sensor health state diagnosis method and system, which solves the problems of static threshold mismatch, poor soil adaptability, and high false alarm rate in traditional technology through dynamic threshold adjustment, multi-dimensional data fusion, and redundancy checking mechanism.

[0004] To solve the above technical problems, the present application adopts the following technical solutions:

[0005] A dredger sensor health state diagnosis method, comprising the following steps:

[0006] Selecting the last N consecutive sampling points, calculating the data mean μ in the sliding window t and the standard deviation σ t , and defining a threshold confidence interval;

[0007] If the current sampling value x t falls into the threshold confidence interval for M consecutive times, it is determined that the signal is stable; otherwise, it is marked as abnormal;

[0008] Combining the historical threshold and the real-time standard deviation σ t , dynamically adjusting the sensor health state judgment threshold at the current time through an exponential weighted moving average model and forming an updated threshold confidence interval based on the sensor health state judgment threshold at the current time.

[0009] If the current sampling value x t If the current sampling value x t If the current sampling value x t If the current sampling value x

[0010] Preferably, the data mean μ t The following formula is used: The standard deviation σ t The following formula is used: Where N is the number of sampling points in the sliding window, is the i-th real-time sampling value.

[0011] Preferably, the threshold confidence interval satisfies: x t ∈ [μ t -k·σ t , μ t +k·σ t ]. Where k is the threshold confidence interval coefficient, and M is the continuous stability threshold.

[0012] Preferably, the current time sensor health state determination threshold The following formula is used: Where α is the attenuation factor, and 0 < α < 1, is the previous time sensor health state determination threshold.

[0013] Preferably, the update threshold confidence interval satisfies:

[0014] Preferably, the dredger sensor health state diagnosis method further includes the steps of soil classification identification and sensor fault judgment, including the following steps:

[0015] If the detected signal continuously exceeds the upper limit of the threshold confidence interval, trigger the water entry suspicion;

[0016] If the soil classification identification is clay at this time, it is directly confirmed as a water entry fault;

[0017] If the soil classification identification is sand, the corrected threshold is obtained according to the sand adaptation formula to obtain the sand update threshold confidence interval, and then the fault is judged. If the signal still exceeds the upper limit of the sand update threshold confidence interval at this time, it is confirmed as a water entry fault;

[0018] If the detected signal continuously falls below the lower limit of the threshold confidence interval, trigger the broken line fault suspicion;

[0019] If the soil classification is identified as sand at this time, it is directly confirmed as a broken line fault;

[0020] If the soil classification is identified as clay, the corrected threshold value is obtained according to the clay adaptation formula, the clay renewal threshold confidence interval is obtained, and then fault judgment is performed. If the signal is still lower than the lower limit of the clay renewal threshold confidence interval at this time, it is confirmed as a broken line fault.

[0021] Preferably, the sand renewal threshold confidence interval includes the following steps:

[0022] Determine the sensor health state judgment threshold value at the current time under sand conditions wherein, is the sensor health state judgment threshold value at the current time, and γ is a correction coefficient. For sand, γ = 0.2 is taken;

[0023] Based on the sensor health state judgment threshold value at the current time under sand conditions, the sand renewal threshold confidence interval is obtained: [μ t ±k·v' carvause ].

[0024] Preferably, the clay renewal threshold confidence interval includes the following steps:

[0025] Determine the sensor health state judgment threshold value at the current time under clay conditions wherein, is the sensor health state judgment threshold value at the current time, and γ is a correction coefficient. For clay, γ = 0.1 is taken;

[0026] Based on the sensor health state judgment threshold value at the current time under clay conditions, the clay renewal threshold confidence interval is obtained: [μ t ±k·v″ carvause ].

[0027] A dredger sensor health state diagnosis system according to the foregoing, comprising:

[0028] A data acquisition unit, the data acquisition unit is used for real-time acquisition of mud pump working parameters;

[0029] A signal preprocessing unit, the signal preprocessing unit calculates the mean and standard deviation based on sliding window variance analysis;

[0030] A diagnosis decision unit, the diagnosis decision unit is used for determining the sensor fault type; and

[0031] An alarm output unit, the alarm output unit is used for outputting an alarm signal.

[0032] Preferably, the diagnostic decision unit comprises a dynamic threshold adaptation module, a soil classification strategy module and a redundancy check module.

[0033] From the above technical solutions, the present application has the following beneficial effects:

[0034] 1. In the present application, the data acquisition unit is used to collect data such as mud pump power, mud density, pressure, flow rate, etc. in real time, the signal preprocessing unit is used to analyze and calculate the mean value and standard deviation, and then the dynamic threshold adaptation module, the soil classification strategy module and the redundancy check module are used to determine the fault type, and then the alarm output unit is used to trigger the alarm signals such as broken line, water ingress, boundary crossing, etc. according to the priority.

[0035] 2. In the present application, the mud density sensor, the pressure sensor, the flow sensor and other sensors are used to monitor the stability of the sensor signals in real time, the exponential weighted moving average model is used to dynamically adjust the sensor health state determination threshold at the current time, to adapt to the fluctuation characteristics of the mud parameters with the change of soil, and the diagnostic decision unit is used to distinguish the fault types such as broken line, water ingress, boundary crossing, etc.

[0036] 3. In the present application, the soil classification is used to optimize the diagnostic parameters, to reduce the influence of environmental interference, and the double sensor redundancy check is used to improve the reliability of the health state determination. BRIEF DESCRIPTION OF DRAWINGS

[0037] Figure 1 A dredger sensor health state diagnosis method block diagram is provided for the present application.

[0038] Figure 2 A structure block diagram of a dredger sensor health state diagnosis system is provided for the present application.

[0039] Figure 3 A dredger sensor health state diagnosis method determination flowchart is provided for the present application.

[0040] In the figure: 10, data acquisition unit; 20, signal preprocessing unit; 30, diagnostic decision unit; 310, dynamic threshold adaptation module; 320, soil classification strategy module; 330, redundancy check module; 340, alarm decision module; 40, alarm output unit. DETAILED DESCRIPTION

[0041] The preferred embodiment of the present application will be described in detail below with reference to the accompanying drawings.

[0042] To achieve the above-mentioned purpose, the embodiments of the present application adopt the following technical solutions: refer to Figure 1 , Figure 3The application provides a dredger sensor health state diagnosis method M10, and the dredger sensor health state diagnosis method M10 of the embodiment comprises the following steps: S100, S200, 300 and S400,

[0043] Further, S100: select the latest continuous N sampling points, calculate the data mean μ t and the standard deviation σ t in the sliding window, and define a threshold confidence interval.

[0044] Specifically, the latest continuous N sampling points (for example, N=15) are selected, the data mean μ t and the standard deviation σ t in the sliding window are calculated, and

[0045]

[0046] In formula (1) and formula (2),

[0047] μ t : the data mean at the current moment t;

[0048] N: the number of sampling points in the sliding window;

[0049] The ith real-time sampling value (the ith data point at the moment t);

[0050] σ t : the data standard deviation at the current moment t.

[0051] Further, S200: if the current sampling value x t falls into the threshold confidence interval for M consecutive times, it is determined that the signal is stable; otherwise, it is marked as abnormal (such as disconnection or water ingress).

[0052] Specifically, the threshold confidence interval satisfies:

[0053] x t ∈ [μ t -k·σ t , μ t +k·σ t ](3)

[0054] In formula (3),

[0055] k: the threshold confidence interval coefficient (usually 3, corresponding to a 99.7% confidence level);

[0056] M: the continuous stability threshold, to prevent transient fluctuation misjudgment.

[0057] Based on the foregoing threshold confidence interval and σ tfurther, execute the next step, S300: combine historical threshold and real-time standard deviation σ t by the exponential weighted moving average model dynamic adjustment of the current time sensor health status determination threshold

[0058] Specifically, the current time sensor health status determination threshold satisfies the following formula:

[0059]

[0060] In formula (4),

[0061] The current time sensor health status determination threshold;

[0062] The previous time sensor health status determination threshold (historical data weight), the initial value is 0;

[0063] α: decay factor (0 < α < 1), control the influence of historical threshold;

[0064] σ t : real-time standard deviation of the current time sensor data (reflecting the degree of data fluctuation).

[0065] It should be noted that, By balancing the instantaneous fluctuation and historical trend through α parameter:

[0066] Low α value: real-time fluctuation weight high → confidence interval rapid response change → abnormal detection sensitive;

[0067] High α value: historical data weight high → confidence interval change gently → strong noise resistance.

[0068] Through the calculation results of (4) formula, the update threshold confidence interval is formed, and the update threshold confidence interval is In this way, the next step S400 is performed: when the sensor real-time detection value x t Fall into the update threshold confidence interval for many times (M times, defined before), then determine that the sensor is working normally; when the sensor real-time detection value x t Continue to be lower than the lower limit of the update threshold confidence interval, it is determined that the line break fault occurs; when the sensor real-time detection value x t Continue to be higher than the upper limit of the update threshold confidence interval, it is determined that the water inlet fault occurs.

[0069] In addition, it should be noted that the physical properties of different soil (such as clay, sand, silt, etc.) are significantly different, which directly affects the distribution of sensor data and the running state of the equipment. According to the real-time identification of soil type (clay, sand, etc.) according to the parameters of mud density, flow rate, sand content, etc., and adjusting the key parameters, therefore, the dredger sensor health state diagnosis method also includes the steps of soil classification identification and sensor fault judgment, including the following steps:

[0070] If the signal is detected to continue beyond the upper limit of the threshold confidence interval, the water entry suspicion is triggered;

[0071] If the soil classification identification is clay at this time, it is directly confirmed as a water entry fault;

[0072] If the soil classification identification is sand, the corrected threshold value is obtained according to the sand adaptation formula to obtain the sand updated threshold confidence interval, and then the fault is judged. If the signal still exceeds the upper limit of the sand updated threshold confidence interval at this time, it is confirmed as a water entry fault;

[0073] If the signal is detected to continue below the lower limit of the threshold confidence interval, the broken line fault suspicion is triggered;

[0074] If the soil classification identification is sand at this time, it is directly confirmed as a broken line fault;

[0075] If the soil classification identification is clay, the corrected threshold value is obtained according to the clay adaptation formula to obtain the clay updated threshold confidence interval, and then the fault is judged. If the signal is still below the lower limit of the clay updated threshold confidence interval at this time, it is confirmed as a broken line fault.

[0076] Further, the sand updated threshold confidence interval includes the following steps:

[0077] In the sand adaptation case, the sensor vibration frequency is increased due to the friction of sand particles, and the threshold value is floated γ, therefore, the current time sensor health state judgment threshold value v' of sand working condition curvause Satisfies the following formula:

[0078]

[0079] In formula (5),

[0080] The current time sensor health state judgment threshold value;

[0081] γ: correction coefficient, take γ = 0.2 (threshold value floating 20%), inhibit false alarm caused by particle friction.

[0082] Subsequently, based on the current time sensor health state judgment threshold value of sand working condition, the sand updated threshold confidence interval is obtained: [μ t± k · v' carvause ].

[0083] Similarly, the obtained clay renewal threshold confidence interval includes the following steps:

[0084] In the clay adaptation case, due to the poor flowability of clay, the signal fluctuation is small, and the threshold is reduced γ, therefore, the current time sensor health state judgment threshold v" of the clay working condition curvause satisfies the following formula:

[0085]

[0086] In formula (6)

[0087] the current time sensor health state judgment threshold;

[0088] γ: correction coefficient, take γ = 0.1 (threshold 10% down), adapt to low fluctuation characteristics.

[0089] Subsequently, based on the current time sensor health state judgment threshold of the clay working condition, the clay renewal threshold confidence interval [μ t ± k · v" carvause ] is obtained.

[0090] In this way, after obtaining the judgment threshold through formula (4), if it is detected that the signal continuously exceeds the upper limit of the confidence interval, the water inflow suspicion is triggered; if the soil classification identification is clay at this time, the water inflow fault is directly confirmed; if the soil classification is sand, the modified threshold is obtained according to the soil adaptation strategy of formula (5), and then the sand renewal threshold confidence interval [μ t ± k · v' carvause ] is obtained again, and then the fault is judged, and if the signal still exceeds the upper limit of the confidence interval, the water inflow fault is confirmed.

[0091] If it is detected that the signal continuously falls below the lower limit of the confidence interval, the broken line fault suspicion is triggered; if the soil identification is clay at this time, the threshold is corrected according to formula (6), and the clay renewal threshold confidence interval [μ t ± k · v" carvause ] is obtained, and then the fault is judged, and if the signal still falls below the lower limit of the confidence interval, the broken line fault is confirmed; if the soil identification is sand, the broken line fault is directly confirmed.

[0092] In addition, in some embodiments, redundant sensors are deployed for key parameters (such as mud pump power, pressure, etc.), that is, for the same measurement signal, two sensors are configured, for convenience of distinction, the two sensors can be named as sensor A and sensor B, when used, if the sensor A data continuously exceeds the threshold confidence interval If the sensor B data is stable, the sensor A is marked as "faulty state", and the data isolation is performed, and the data input diagnosis process is suspended.

[0093] Referring to Figure 2 In some embodiments, the present application also provides a dredger sensor health state diagnosis system, comprising a data acquisition unit 10, a signal preprocessing unit 20, a diagnosis decision unit 30 and an alarm output unit 40. Specifically, the data acquisition unit 10 is used to acquire the working parameters of the mud pump in real time, the signal preprocessing unit 20 is used to calculate the mean value and standard deviation based on the sliding window variance analysis, the diagnosis decision unit 30 is used to determine the sensor fault type, and the alarm output unit 40 is used to output the alarm signal.

[0094] In addition, the dredger sensor health state diagnosis system further comprises an alarm decision module 340, and the dynamic threshold adaptation module 320, the soil classification strategy module 330 and the redundancy verification module 330 are electrically connected with the alarm decision module 340. In this way, the diagnosis decision result can be transmitted to the alarm decision module 340 and then transmitted to the alarm output unit 40 for output through the alarm decision module 340.

[0095] Further, the data acquisition unit is a mud pump power sensor, a mud density sensor, a pressure sensor, etc., which is used to acquire the sensor data of the mud pump power, the mud density, the pressure, the flow rate, etc. in real time.

[0096] Further, the diagnosis decision unit 30 comprises a dynamic threshold adaptation module 310, a soil classification strategy module 320 and a redundancy verification module 330. In use, the data acquisition unit is used to acquire the data of the mud pump power, the mud density, the pressure, the flow rate, etc. in real time, and the signal preprocessing unit is used to analyze and calculate the mean value μ t and the standard deviation σ t Further, the diagnosis decision unit 30 comprises a dynamic threshold adaptation module 310, a soil classification strategy module 320 and a redundancy verification module 330. In use, the data acquisition unit is used to acquire the data of the mud pump power, the mud density, the pressure, the flow rate, etc. in real time, and the signal preprocessing unit is used to analyze and calculate the mean value μ

[0097] The above-described embodiments are merely preferred embodiments of the present application, and are not intended to limit the scope of the present application. Various modifications and improvements to the technical solutions of the present application made by those skilled in the art without departing from the design spirit of the present application shall fall within the protection scope of the claims of the present application.

Claims

1. A method of diagnosing the health state of a dredger sensor, characterized by, The method comprises the following steps: Select the last N consecutive sampling points, calculate the data mean μ in the sliding window t And standard deviation σ t And define the threshold confidence interval; If the current sample value x t If the current sample value x t If the current sample value x t If the current sample value x t If the current sample value x t If the current sample value x t If the current sample value x t Combine historical threshold value And real-time standard deviation σ t The current time sensor health state judgment threshold value is dynamically adjusted by an exponential weighted moving average model And based on the current time sensor health state judgment threshold value Form an updated threshold confidence interval; If the current sampling value x t If the current sampling value x t If the current sampling value x t If the current sampling value x 2. The dredge sensor health diagnosis method of claim 1, wherein, The data mean μ t The following formula is used: The standard deviation σ t The following formula is used: Where N is the number of sampling points in the sliding window, The i-th real-time sampling value.

3. The dredge sensor health diagnosis method of claim 2, wherein, The threshold confidence interval satisfies: [μ t -k·σ t ,μ t +k·σ t ]; wherein k is a threshold confidence interval coefficient, and M is a continuous stability frequency threshold value.

4. The dredge sensor health diagnosis method of claim 3, wherein, the current time point sensor health state determination threshold value using the following equation: wherein a is an attenuation factor, and 0 is a previous time point sensor health state determination threshold value.

5. The dredger sensor health diagnosis method according to claim 4, characterized in that, The updated threshold confidence interval satisfies:

6. The dredge sensor health diagnosis method of claim 1, wherein, The dredger sensor health state diagnosis method further comprises a soil classification identification and sensor fault judgment step, which comprises the following steps: If the signal is detected to be continuously above the upper limit of the threshold confidence interval, water entry suspicion is triggered; If the soil classification identification is clay at this time, water entry fault is directly confirmed; If the soil classification identification is sand, a corrected threshold is obtained according to a sand adaptation formula to obtain a sand updated threshold confidence interval, and then fault judgment is performed; if the signal is still above the upper limit of the sand updated threshold confidence interval at this time, water entry fault is confirmed; If the signal is detected to be continuously below the lower limit of the threshold confidence interval, broken wire fault suspicion is triggered; If the soil classification identification is sand at this time, broken wire fault is directly confirmed; If the soil classification identification is clay, a corrected threshold is obtained according to a clay adaptation formula to obtain a clay updated threshold confidence interval, and then fault judgment is performed; if the signal is still below the lower limit of the clay updated threshold confidence interval at this time, broken wire fault is confirmed.

7. The dredger sensor health diagnosis method according to claim 6, characterized in that, The sand updated threshold confidence interval comprises the following steps: Determination of sensor health state judgment threshold value at current time under sandy soil working condition wherein, is the sensor health state judgment threshold value at current time, and γ is a correction coefficient, for sandy soil, γ=0.2 is taken; Based on the sand soil working condition current time sensor health state determination threshold, the sand soil update threshold confidence interval is obtained: [μ t ±k·v' carvause ].

8. The dredger sensor health diagnosis method according to claim 7, characterized in that, The clay updated threshold confidence interval comprises the following steps: Determination of sensor health state judgment threshold value at current time under clay working condition wherein, is the sensor health state judgment threshold value at current time, and γ is a correction coefficient, for clay, γ=0.1 is taken. Based on the sensor health state determination threshold at the current time under the clay working condition, a clay updating threshold confidence interval is obtained: [μ t ±k·v″ carvause ].

9. A dredger sensor health diagnosis system according to any one of claims 1-8, characterized in that, It comprises: A data acquisition unit (10) for acquiring mud pump working parameters in real time; A signal preprocessing unit (20) for calculating mean and standard deviation based on sliding window variance analysis; A diagnosis decision unit (30) for determining sensor fault type; and An alarm output unit (40) for outputting an alarm signal.

10. The dredge sensor health diagnostic system of claim 9, wherein, The diagnosis decision unit (30) comprises a dynamic threshold adaptation module (310), a soil classification strategy module (320) and a redundancy verification module (330).

Images