Method, device and storage medium for dynamically determining warning threshold

Abstract

The present application discloses a method, device, and storage medium for dynamically determining a warning threshold, relating to the field of data processing technology. The method first determines the health index of a device within a preset time window based on multi-source sensor data within the preset time window; then, a linear attenuation equation is fitted based on the health index within the preset time window, and the slope of the linear attenuation equation is used as the attenuation rate of the health index; thereafter, the warning threshold of the multi-source sensor is determined based on the attenuation rate. By introducing the health index and attenuation rate, this method can dynamically adjust the warning threshold based on the actual operating status and degradation of the device, overcoming the limitation that fixed thresholds cannot reflect dynamic changes in the device in real time, thereby improving the timeliness of equipment maintenance.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a method, device, and storage medium for dynamically determining an early warning threshold. Background Art

[0002] In the field of smart device maintenance, fixed thresholds are often used to trigger alarms. For example, unified warning thresholds are set based on factory device parameters or historical experience. However, device degradation is a complex dynamic process, influenced by multiple factors, such as the device's operating environment, operating conditions, and aging. Fixed thresholds cannot reflect the device's actual degradation status in real time under different operating stages and environments, resulting in delayed maintenance.

[0003] The above content is only used to assist in understanding the technical solution of this application and does not constitute an admission that the above content is prior art. Summary of the Invention

[0004] The main purpose of this application is to provide a method, device and storage medium for dynamically determining an early warning threshold, aiming to solve the technical problem of improving the timeliness of equipment maintenance.

[0005] To achieve the above objectives, the present application proposes a method for dynamically determining a warning threshold, the method comprising:

[0006] Determining a health index of the device within a preset time window based on multi-source sensor data within the preset time window;

[0007] Fitting a linear attenuation equation according to the health index within the preset time window, and using the slope of the linear attenuation equation as the attenuation rate of the health index;

[0008] An early warning threshold of the multi-source sensor is determined according to the attenuation rate.

[0009] In one embodiment, the step of determining the health index of the device within the preset time window based on the multi-source sensor data within the preset time window includes:

[0010] Obtaining an operation instruction received by the device within the preset time window;

[0011] The timestamp corresponding to the operation instruction is used as a time sequence mark and mapped to the multi-source sensor data to obtain a data sequence with a time sequence mark;

[0012] Determining a target operating period of the device according to a timing mark corresponding to the operation instruction, and calculating a state transition probability according to a data sequence corresponding to the target operating period using a forward-backward algorithm;

[0013] Segmenting the data sequence corresponding to the target operating time period according to the state transition probability to obtain a target data sequence;

[0014] A health index of the device within the preset time window is determined according to the target data sequence.

[0015] In one embodiment, the step of determining the health index of the device within the preset time window based on the target data sequence includes:

[0016] Determine the historical sensor data sequence corresponding to each historical failure mode based on historical failure cases;

[0017] determining a weight of the target data sequence based on a similarity between the target data sequence and the historical sensor data sequence;

[0018] The health index is determined according to the target data sequence and the weight.

[0019] In one embodiment, the step of determining the weight of the target data sequence based on the similarity between the target data sequence and the historical sensor data sequence includes:

[0020] Obtaining an initial weight of the target data sequence;

[0021] Determining a target fault mode corresponding to the target data sequence based on a similarity between the target data sequence and the historical sensor data sequence;

[0022] determining, according to the faulty sensor corresponding to the target fault mode, a compensation weight for data corresponding to the faulty sensor in the target data sequence;

[0023] The weight of the target data sequence is determined according to the initial weight of the target data sequence and the compensation weight of the data corresponding to the faulty sensor in the target data sequence.

[0024] In one embodiment, after the step of fitting a linear attenuation equation according to the health index within the preset time window and using the slope of the linear attenuation equation as the attenuation rate of the health index, the method further includes:

[0025] Creating a health index sequence according to the health index within the preset time window;

[0026] performing packet decomposition on the health index sequence to obtain a low-frequency component corresponding to the health index sequence;

[0027] Determining a Hurst exponent corresponding to the low-frequency component, and determining an adjustment amount for the attenuation rate based on the Hurst exponent, wherein the Hurst exponent represents the likelihood that a changing trend of the health index sequence will continue in the future;

[0028] The sum of the attenuation rate and the adjustment amount of the attenuation rate is used as the new attenuation rate.

[0029] In one embodiment, the step of determining the warning threshold of the multi-source sensor according to the attenuation rate includes:

[0030] determining an initial threshold value of the multi-source sensor according to the health index;

[0031] determining a threshold adjustment amount of the multi-source sensor according to the attenuation rate;

[0032] An early warning threshold of the multi-source sensor is determined according to the initial threshold and the threshold adjustment amount.

[0033] In one embodiment, after the step of determining the warning threshold of the multi-source sensor according to the attenuation rate, the method further includes:

[0034] Using the health index and the target failure mode as decision variables, and using maintenance resources as constraint variables, wherein the maintenance resources include at least one of the number of equipment spare parts, the maintenance time window, and the matching degree of maintenance personnel;

[0035] Inputting the decision variables and the constraint variables into a preset work order generation model, generating at least one reference maintenance plan through the preset work order generation model, and calculating the maintenance cost corresponding to the reference maintenance plan;

[0036] A target maintenance plan is determined in the reference maintenance plan according to the maintenance cost, and a maintenance work order is generated according to the target maintenance plan.

[0037] In one embodiment, after the steps of determining a target maintenance plan from the reference maintenance plan based on the maintenance cost and generating a maintenance work order based on the target maintenance plan, the method further includes:

[0038] Obtaining the health index and maintenance time of the equipment after maintenance;

[0039] Calculating a reward function value corresponding to the target maintenance plan according to the health index after maintenance and the maintenance time;

[0040] If the reward function value is less than a preset reward value, re-determining the decay rate of the device according to the health index after maintenance;

[0041] The warning threshold of the multi-source sensor is re-determined according to the re-determined attenuation rate.

[0042] In addition, to achieve the above-mentioned purpose, the present application also proposes a device for dynamically determining a warning threshold, which includes: a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the computer program is configured to implement the steps of the method for dynamically determining the warning threshold as described above.

[0043] In addition, to achieve the above-mentioned purpose, the present application also proposes a storage medium, which is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by the processor, the steps of the method for dynamically determining the warning threshold as described above are implemented.

[0044] This application provides a method for dynamically determining warning thresholds. The method first determines the health index of a device within a preset time window based on multi-source sensor data within that window. A linear attenuation equation is then fitted to the health index within the preset time window, with the slope of the linear attenuation equation serving as the attenuation rate of the health index. Finally, the warning threshold for the multi-source sensor is determined based on the attenuation rate. By introducing the health index and attenuation rate, this method dynamically adjusts the warning threshold based on the device's actual operating status and degradation. This overcomes the limitation of fixed thresholds, which cannot reflect dynamic changes in the device in real time, and improves the timeliness of device maintenance. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0046] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0047] Figure 1 A flowchart illustrating a first embodiment of a method for dynamically determining a warning threshold value according to the present application;

[0048] Figure 2 A flowchart illustrating a second embodiment of a method for dynamically determining an early warning threshold value according to this application;

[0049] Figure 3 A schematic diagram of the method flow for the real-time example 2 of the method for dynamically determining the warning threshold value of this application;

[0050] Figure 4 A flowchart of the third embodiment of the method for dynamically determining the warning threshold value of this application is provided;

[0051] Figure 5 Schematic diagram of a brief flow chart of a method for dynamically determining a warning threshold value in an embodiment of the present application;

[0052] Figure 6 Schematic diagram of the device structure of the hardware operating environment involved in the method for dynamically determining the warning threshold in the embodiment of the present application.

[0053] The purpose, features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0054] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application and are not intended to limit the present application.

[0055] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0056] In the field of smart device maintenance, fixed thresholds are often used to trigger alarms. For example, unified warning thresholds are set based on factory device parameters or historical experience. However, device degradation is a complex dynamic process, influenced by multiple factors, such as the device's operating environment, operating conditions, and aging. Fixed thresholds cannot reflect the device's actual degradation status in real time under different operating stages and environments, resulting in delayed maintenance.

[0057] In light of the above issues, this application proposes a method for dynamically determining warning thresholds. The method first determines the health index of a device within a preset time window based on multi-source sensor data within that window. A linear attenuation equation is then fitted to the health index within the preset time window, with the slope of the linear attenuation equation serving as the attenuation rate of the health index. Finally, the warning threshold for the multi-source sensor is determined based on the attenuation rate. By introducing the health index and attenuation rate, this method dynamically adjusts the warning threshold based on the device's actual operating status and degradation. This overcomes the limitation of fixed thresholds, which cannot reflect dynamic changes in the device in real time, and improves the timeliness of device maintenance.

[0058] It should be noted that the execution subject of this embodiment can be a computing service device with data processing, network communication and program running functions, such as a tablet computer, personal computer, etc., or an electronic device that can realize the above functions.

[0059] Based on this, the first embodiment proposed in this application provides a method for dynamically determining a warning threshold value. Figure 1 In this embodiment, the method for dynamically determining the warning threshold includes steps S10 to S40:

[0060] Step S10: determining a health index of the device within a preset time window based on multi-source sensor data within the preset time window.

[0061] Understandably, traditional fixed thresholds, based on static parameters or empirical values, are unable to adapt to equipment degradation due to environmental, operating, and aging factors. However, by calculating a health index based on multi-source sensor data within a preset time window and dynamically determining warning thresholds based on the health index's decay rate, we can avoid interference from noise or false alarms from single sensor data, quantify the device's comprehensive health status in real time, and dynamically reflect its degradation trajectory.

[0062] Optionally, step S10 includes steps S11 to S15:

[0063] Step S11: Acquire the operation instructions received by the device within the preset time window.

[0064] Step S12: Use the timestamp corresponding to the operation instruction as a time sequence mark and map it to the multi-source sensor data to obtain a data sequence with a time sequence mark.

[0065] It should be noted that operating instructions refer to control commands or operating signals received by a device during operation. These instructions can come from manual operation, automatic control systems, or upper-level management software, and include commands such as starting, stopping, adjusting parameters, and switching modes. By accessing the device's operation log, you can obtain information such as the timestamp and operation type of the operating instruction.

[0066] During equipment operation, multi-source sensors such as vibration, temperature, and current sensors, as well as operational commands such as start / stop, operation, and load change, typically collect data independently, potentially leading to inconsistent timing. Mapping the timestamps of operational commands as timing markers onto the multi-source sensor data allows synchronization of operational events with sensor readings. Optionally, prior to synchronization, the multi-source sensor data can be preprocessed using noise reduction filtering and missing value filling.

[0067] It's understandable that a device may have multiple operating cycles within a preset fixed time window. For example, an injection molding machine may experience intermittent downtime during production batches, and this fixed time window may contain invalid downtime data, meaning some sensor data may appear as zeros. Operational instructions can provide real-time information on changes during device operation. To ensure the accuracy of health index calculations, the timestamps of the operational instructions are mapped onto the multi-source sensor data as time series markers. This allows the time-stamped data sequence to reflect the changes in multi-source sensor data under different operating conditions of the device.

[0068] Step S13, determining the target operating time period of the device according to the timing mark corresponding to the operation instruction, and calculating the state transition probability according to the data sequence corresponding to the target operating time period by a forward-backward algorithm.

[0069] Step S14: segmenting the data sequence corresponding to the target operating time period according to the state transition probability to obtain a target data sequence.

[0070] It is understandable that when a device starts, stops, or experiences load changes, sensor data will exhibit nonlinear transition characteristics, such as increasing speed and temperature fluctuations. Calculating the health index directly using raw data will include these transient processes in the evaluation, leading to index fluctuations. For example, the vibration amplitude of a bearing at startup may briefly increase due to the lack of a lubricating oil film, but this does not mean that the bearing has degraded. Therefore, to more accurately obtain the health status of the device during stable operation, the state transition probability of the data sequence corresponding to the device during the target operating period can be calculated. The sensor data representing the changing process can be eliminated, leaving only the target data sequence that represents the device in a stable operating state.

[0071] Exemplarily, a hidden Markov model is used to partition a data sequence into operating cycles based on the timing signatures of operational instructions, thereby obtaining multi-source sensor data representing the device's target operating period. For example, the data sequence is partitioned based on "start" and "stop" instructions to obtain multi-source sensor data for the target operating period. A forward-backward algorithm is then used to calculate the state transition probabilities for the data sequence corresponding to the target operating period obtained after the partitioning. The data sequence corresponding to the target operating period is then further partitioned based on the state transition probabilities to obtain the target data sequence representing the device's stable operation.

[0072] To better understand the solution provided in this example, this example is further explained in conjunction with specific application scenarios.

[0073] Assuming that the data sequence is divided into operating cycles according to the timing marks corresponding to the operation instructions "start" and "stop", after obtaining the data sequence corresponding to the target operating period, the steps to obtain the target data sequence in the "stable operation" state from the data sequence corresponding to the target operating period can be as follows:

[0074] First, based on the actual operating characteristics of the equipment, the equipment's operating state during the target operating period is modeled as a Markov chain consisting of multiple hidden states to reflect the equipment's performance at different operating stages. For example, the equipment's operating state can be modeled as a Markov chain consisting of three hidden states: "Startup Warmup," "Stable Operation," and "Speed ​​Down Shutdown." Startup Warmup represents the process of gradually accelerating the equipment from a stationary state to a stable operating speed. Stable operation represents the state of normal operation of the equipment at rated load and speed. Speed ​​Down Shutdown represents the process of gradually decelerating the equipment from a stable operating state to a stop.

[0075] Next, feature data is extracted from the data sequence corresponding to the target operating period to generate an observation sequence. These features should be able to accurately reflect the operating status of the device. For example, the centroid of the vibration spectrum can be extracted from the vibration sensor data to reflect the distribution of the device's vibration energy, and the temperature change rate can be extracted from the temperature sensor data to reflect the temperature change trend of the device during operation. Feature data can also be other related features such as current effective value, pressure fluctuation, etc., depending on the sensor type or user needs, and this application does not impose specific restrictions.

[0076] Subsequently, a current moment is determined in the data sequence corresponding to the target operating period. Starting from the initial moment of the target operating period, the forward probability of the device being in each hidden state and observing the current observation sequence is calculated step by step using the hidden state transition probability and observation probability. Also, starting from the current moment, the backward probability of the device being in each hidden state and observing the subsequent observation sequence is calculated step by step using the hidden state transition probability and observation probability. The hidden state transition probability represents the probability of the device transitioning from one hidden state to another; the hidden state observation probability represents the probability of observing a certain observation state under a specific hidden state, and can be set based on experience or through model training based on historical data. Then, based on the forward and backward probabilities calculated above, the hidden state transition probability and observation probability are iteratively optimized using the Baum-Welch algorithm.

[0077] Finally, starting from the initial moment, the forward probability and the optimized state transition probabilities are used to calculate the maximum probability of the device being in each hidden state until the final moment. Based on this maximum probability, a maximum probability path for hidden state transitions in the sensor data is derived, resulting in the most likely state transition sequence during the device's operation. Based on this state transition sequence, the start and end points of each complete operating cycle of the device are identified. For example, a complete operating cycle begins with the "start-up warm-up" state and ends with the "speed reduction shutdown" state. Based on the identified start and end points of the operating cycle, the target data sequence for the "stable operation" state is obtained from the data sequence corresponding to the target operating period.

[0078] Step S15: determining the health index of the device within the preset time window according to the target data sequence.

[0079] Understandably, equipment degradation typically occurs during stable operation, not transient processes. By segmenting the target data sequence, the health index calculation can more specifically learn degradation characteristics, avoiding interference from transient processes that may contain features unrelated to degradation, thereby improving the accuracy of the health index calculation.

[0080] Optionally, step S15 includes steps S151 to S153:

[0081] Step S151 : determining the historical sensor data sequence corresponding to each historical failure mode according to the historical failure cases.

[0082] Step S152: Determine the weight of the target data sequence according to the similarity between the target data sequence and the historical sensor data sequence.

[0083] It is understandable that different failure modes of equipment involve different physical degradation processes. For example, bearing wear is mainly caused by surface fatigue and lubrication failure. Early characteristics may be reflected in the increase of specific frequency components in the vibration spectrum, while motor overheating may be caused by aging of winding insulation or failure of the cooling system. The effective value of current and the temperature change rate are key indicators. If all sensor data are given the same weight, the health index may misjudge the status of the equipment due to ignoring the specificity of the failure mode. For example, in the early stages of bearing wear, the sensitivity of vibration data is higher than that of temperature data. If the two are given the same weight, the health index may not reflect the degree of wear in a timely manner. Therefore, the current target data sequence can be used to determine the failure mode that the equipment may face, and then the corresponding sensor data weight can be further increased according to the possible failure mode to improve the failure specificity of the health index and more accurately calculate the health index of the equipment.

[0084] Exemplarily, past equipment failure cases are collected from equipment maintenance records, fault databases, or expert knowledge bases. Each fault case contains the fault type, such as bearing wear, gear breakage, motor overheating, etc., the time of occurrence, a description of the fault phenomenon, and sensor data before and after the fault occurs. The collected fault cases are classified and organized to establish a fault mode library. Each fault mode corresponds to one or more fault types and is associated with the corresponding sensor data sequence. From the fault mode library, the historical sensor data sequence corresponding to each fault mode is extracted. Using algorithms such as dynamic time warping, Euclidean distance, and cosine similarity, the similarity between the target data sequence and the historical sensor data sequence corresponding to each historical fault mode is calculated. Based on the similarity calculation results, the weight of the target data sequence is determined.

[0085] A higher similarity indicates a closer match between the target data sequence and a historical failure pattern. Therefore, the sensor data corresponding to that failure pattern should be given a higher weight when calculating the health index. The device's health index is then calculated by combining the weights of the target data sequence with the sensor data. This calculation can be a weighted average, such as multiplying each sensor's data by its weight and summing the results, or a more complex machine learning model, such as using weights as input features to train a health index prediction model.

[0086] To better understand the solution provided in this example, this example is further explained in conjunction with specific application scenarios.

[0087] Taking wind turbine gearbox health monitoring as an example, it is assumed that a fault mode library containing failure modes such as bearing wear and gear fracture has been established and associated with the corresponding historical sensor data sequences. A dynamic time warping algorithm is used to calculate the similarity between the target data sequence, i.e., the current gearbox operating data, and the historical sensor data sequence corresponding to each historical fault mode. Based on the similarity calculation results, the weight of the target data sequence is determined. For example, if the target data sequence has a high similarity with the historical sensor data sequence of the bearing wear failure mode, the bearing wear-related sensor data is given a higher weight. Combining the weight of the target data sequence and the sensor data, a weighted average method is used to calculate the health index of the gearbox. A low health index indicates that the gearbox may be at risk of failure, and the generation of a maintenance work order needs to be triggered.

[0088] Optionally, step S152 includes steps S1521 to S1524:

[0089] Step S1521, obtaining the initial weight of the target data sequence.

[0090] For example, the distribution of sensor data under the historical health status of the device is analyzed, and the variance or entropy of each sensor data point is calculated. This is used as the initial weight for each sensor data point in the target data sequence. For example, a larger variance in vibration sensor data indicates greater sensitivity to state changes, and thus a higher initial weight is assigned. The user can also manually set the initial weight for each sensor data point.

[0091] Step S1522: Determine a target fault mode corresponding to the target data sequence based on the similarity between the target data sequence and the historical sensor data sequence.

[0092] Step S1523 : determining, according to the faulty sensor corresponding to the target fault mode, a compensation weight of the data corresponding to the faulty sensor in the target data sequence.

[0093] Step S1524 : determining the weight of the target data sequence according to the initial weight of the target data sequence and the compensation weight of the data corresponding to the faulty sensor in the target data sequence.

[0094] Exemplarily, the dynamic time warping distance between the target data sequence and the historical sensor data sequence corresponding to each historical fault mode is calculated to quantify the similarity between the two in terms of time series. Next, after normalizing the target data sequence and the historical sensor data sequence corresponding to each historical fault mode, the cosine similarity between the target data sequence and the historical sensor data sequence corresponding to each historical fault mode is calculated to evaluate the similarity between the two in terms of direction. The similarity between the two in terms of time series and direction is weighted and summed to obtain a similarity score. If the similarity between the target data sequence and any historical fault mode exceeds a preset threshold, the historical fault mode is used as the target fault mode. The fault sensor corresponding to the target fault mode is identified in the target data sequence, and the compensation weight of the sensor data corresponding to the fault sensor is set.

[0095] Finally, the weight of the target data sequence is obtained by adding the initial weight and the compensation weight.

[0096] Step S153: Determine the health index according to the target data sequence and the weight.

[0097] After determining the weight corresponding to each sensor data in the target data sequence, the multi-source sensor data and its weight are weighted and summed to calculate the health index of the device.

[0098] Optionally, multi-source sensor data is input into the long short-term memory network model, and the long short-term memory network model learns the dependency between the sensor data and the health index based on historical training data to obtain the health index within a preset time period window output by the long short-term memory network model.

[0099] Step S20: fitting a linear attenuation equation according to the health index within the preset time window, and using the slope of the linear attenuation equation as the attenuation rate of the health index.

[0100] It should be noted that the decay rate indicates the rate at which the device health index decreases over time and can be defined as the amount of health index decrease per unit time. For example, if the device health index drops from 1.0 to 0.8 over 100 hours, the decay rate is 0.002 / hour.

[0101] Alternatively, the decay rate is determined by a linear decay model. Assume that the health index decreases linearly over time, i.e. , where λ is the decay rate, is the health index at time t, It is the health index at the initial moment within the preset time window.

[0102] Alternatively, the decay rate is determined by an exponential decay model. Assume that the health index decreases exponentially over time, i.e. .

[0103] Alternatively, for complex decay processes, models such as polynomial regression, support vector regression, or neural network can be used to fit the changing trend of the health index.

[0104] Optionally, the decay rate is determined by a preset formula:

[0105]

[0106] Where P is the preset time window, is the mean value of the health index within the i-th day, is the mean of all health indices, represents the i-th day, It represents the mean of t, which is obtained by calculating the arithmetic mean of all t, that is, .

[0107] Step S30: determining a warning threshold of the multi-source sensor according to the attenuation rate.

[0108] Traditional static thresholds fail to reflect dynamic changes in equipment health. For example, when bearings are in the early stages of wear, the vibration amplitude may be below the fixed threshold, but the decay rate has already significantly increased. When the decay rate of equipment is high, the static threshold may lag behind the actual development of the fault, resulting in a lack of timely warnings and maintenance, leading to delayed maintenance responses. However, quantifying the rate of equipment degradation by decay rate and linking the warning threshold to the degradation rate allows for more accurate early warnings based on the dynamic degradation process of the equipment.

[0109] Optionally, step S30 includes steps S31 to S33:

[0110] Step S31 : determining an initial threshold of the multi-source sensor according to the health index.

[0111] Step S32: determining a threshold adjustment amount of the multi-source sensor according to the attenuation rate.

[0112] Step S33: determining a warning threshold of the multi-source sensor according to the initial threshold and the threshold adjustment amount.

[0113] For example, the initial threshold of each multi-source sensor is determined according to the health index of the device according to the preset mapping relationship. Then, the initial threshold is corrected according to the attenuation rate to determine the threshold adjustment amount of each multi-source sensor: .in, is the initial threshold, The sum of the initial threshold and the threshold adjustment amount is then used as the warning threshold of the multi-source sensor.

[0114] In this embodiment, by introducing the health index and decay rate, the warning threshold can be dynamically adjusted according to the actual operating status and degradation of the equipment, overcoming the limitation that the fixed threshold cannot reflect the dynamic changes of the equipment in real time.

[0115] Based on the first embodiment of the present application, in the second embodiment of the present application, the same or similar contents as those in the above embodiment 1 can be referred to the above introduction and will not be described in detail later. Figure 2 After step S30, the method for dynamically determining the warning threshold further includes steps S40 to S60:

[0116] Step S40: using the health index and the target failure mode as decision variables, and using maintenance resources as constraint variables, wherein the maintenance resources include at least one of the number of equipment spare parts, the maintenance time window, and the matching degree of maintenance staff.

[0117] Exemplarily, the target failure mode is converted into a one-hot encoding or embedding vector, and the health index sequence and the failure mode encoding are concatenated into a high-dimensional vector as a decision variable.

[0118] Step S50: input the decision variables and the constraint variables into a preset work order generation model, generate at least one reference maintenance plan through the preset work order generation model, and calculate the maintenance cost corresponding to the reference maintenance plan.

[0119] For example, Figure 3 As shown in Figure 1, a multi-objective genetic algorithm such as NSGA-II can be used as a preset work order generation model. The multi-objective genetic model is used to optimize the variables in the maintenance plan based on the decision variables and constraint variables to determine the target maintenance plan with the lowest maintenance cost.

[0120] Specifically, the chromosome structure of a multi-objective genetic algorithm can be first encoded as a maintenance plan, for example, chromosome = [maintenance execution time, maintenance operation, spare parts list, maintenance staff]. Multiple sets of chromosomes are randomly generated within the range of constrained variables as alternative maintenance plans, and a maintenance cost function F is defined. The variables in the chromosomes can be determined based on multiple historical maintenance plans corresponding to the target failure mode. The multi-objective genetic algorithm then performs genetic operations to output a non-dominated solution set as a reference maintenance plan. Genetic operations simulate the natural selection, crossover, and mutation processes in biological evolution and are used to generate a new generation of chromosomes, i.e., a new maintenance plan, based on the decision variables and the initial randomly generated sets of chromosomes. A non-dominated solution set refers to a set of solutions in a multi-objective optimization problem for which no better solution exists. In this embodiment, the solution refers to the maintenance plan. The maintenance cost corresponding to the non-dominated solution set is then calculated based on the defined maintenance cost function F.

[0121] Step S60 : determining a target maintenance plan from the reference maintenance plan according to the maintenance cost, and generating a maintenance work order according to the target maintenance plan.

[0122] For example, a maintenance cost threshold can be defined, and a reference plan with a maintenance cost below the threshold can be identified as the target maintenance plan. A maintenance work order is then generated based on the target maintenance plan, including at least the variables in the target maintenance plan. The urgency of the maintenance work order can also be determined based on the equipment's health index; the lower the health index, the higher the urgency of the work order.

[0123] Optionally, when the data collected by the multi-source sensors exceeds the warning threshold and the device health index is less than a first threshold, an alarm is triggered and a maintenance work order is generated. When the data collected by the multi-source sensors exceeds the warning threshold and the device health index is greater than a second threshold, a maintenance work order is generated. The first threshold is less than the second threshold.

[0124] Based on the above embodiments of the present application, in the third embodiment of the present application, the same or similar contents as those in the above embodiments can be referred to the above introduction and will not be described in detail later. Figure 4 After step S30, the method for dynamically determining the warning threshold further includes steps S70 to S100:

[0125] Step S70: creating a health index sequence according to the health index within the preset time window.

[0126] Step S80: performing packet decomposition on the health index sequence to obtain a low-frequency component corresponding to the health index sequence.

[0127] For example, the health indexes within the preset time window are arranged in ascending order by timestamp to generate an equally spaced time series HI= A wavelet basis is selected based on the signal characteristics, and the number of decomposition levels is determined based on the length of the health index sequence. The HI sequence is decomposed using discrete wavelet transform to obtain the corresponding low-frequency and high-frequency components of the HI sequence. The low-frequency component can represent the long-term trend of equipment degradation, such as the gradual process of bearing wear, while the high-frequency component can represent sudden noise or fault impact of the sensor.

[0128] Specifically, the health index sequence is first convolved using the low-pass and high-pass filters corresponding to the wavelet basis. After each decomposition layer, the low-pass filter coefficients are downsampled by a factor of 2. This process is repeated for the low-frequency components retained after each decomposition layer until the desired number of decomposition layers is reached.

[0129] Step S90: determining the Hurst index corresponding to the low-frequency component, and determining the adjustment amount of the attenuation rate according to the Hurst index, wherein the Hurst index represents the possibility that the changing trend of the health index sequence will continue in the future.

[0130] Step S100: taking the sum of the attenuation rate and the adjustment amount of the attenuation rate as the new attenuation rate.

[0131] A low-frequency component is obtained at each level of decomposition. The obtained low-frequency component is divided into N subintervals of length τ. For each subinterval, the range R and standard deviation S of the cumulative deviation are calculated, and then the rescaled range R / S value of each subinterval is calculated. The power law relationship between the rescaled range R / S value and the time window length is then calculated to obtain the Hurst exponent H: . Where C is a constant.

[0132] The Hurst exponent is a measure of the long-term memory of a time series, reflecting the likelihood that a current trend will persist or reverse in the future. A high Hurst exponent indicates that the health index series has long-term memory and that the current trend is likely to persist for some time to come. In other words, the health index will continue to decline in line with its current trend.

[0133] For example, if the Hurst exponent H>0.5, it means that the device degradation trend may continue and the attenuation rate is not adjusted, that is, = ,in, Indicates the adjustment amount of the attenuation rate. 0.5 indicates that the equipment degradation trend is less likely to continue and may reverse, that is, the current attenuation rate may overestimate future risks. In this case, the attenuation rate should be reduced to avoid triggering maintenance too early. ,in, is the adjustment coefficient. Finally, the sum of the attenuation rate and the adjustment amount of the attenuation rate is taken as the new attenuation rate: .

[0134] Through the above steps, combined with wavelet decomposition and Hurst exponent analysis, the changing trend and characteristics of the equipment health index can be considered more comprehensively, and the calculation accuracy of the attenuation rate can be improved.

[0135] Based on the above embodiments of the present application, in the fourth embodiment of the present application, the same or similar contents as those in the first embodiment can be referred to above and will not be described in detail. On this basis, after step S60, the method for dynamically determining the warning threshold further includes steps S110 to S140:

[0136] Step S110: Obtain the health index and maintenance time of the equipment after maintenance.

[0137] Step S120 , calculating a reward function value corresponding to the target maintenance plan according to the health index after maintenance and the maintenance time.

[0138] Step S130: If the reward function value is less than a preset reward value, the attenuation rate of the device is re-determined according to the health index after maintenance.

[0139] Step S140 : re-determine the warning threshold of the multi-source sensor according to the re-determined attenuation rate.

[0140] Collect multi-source sensor data within a preset time period after maintenance and calculate a post-maintenance health index based on this data. Also, record the actual maintenance time. Substitute the post-maintenance health index and maintenance time into a defined reward function to calculate the reward value corresponding to the target maintenance plan.

[0141] For example, the reward function can be .in, is the reward function value; is the health index after maintenance; is the health index before maintenance; The maintenance time window for the target maintenance plan; The actual maintenance process takes time; For maintenance costs; 、 and is the weight coefficient.

[0142] If the calculated reward function value is less than the preset reward value, it means that the current maintenance plan has not achieved the expected effect and the possibility of equipment failure is still relatively high. At this time, the attenuation rate of the equipment is re-determined based on the health index after maintenance, and the warning threshold of the multi-source sensor is re-determined based on the re-determined attenuation rate, or the work order generation rules are readjusted.

[0143] In this embodiment, the maintenance effect is quantified through a reward function. When the maintenance effect does not meet expectations, the system automatically triggers parameter recalculation to adapt to the new degradation state of the equipment and provide timely warnings, which can effectively reduce the response delay of equipment maintenance.

[0144] For example, in order to help understand the implementation process of the method for dynamically determining the warning threshold obtained by combining this embodiment with the above embodiments, please refer to Figure 5 , Figure 5 A brief flowchart of a method for dynamically determining an early warning threshold is provided. Specifically:

[0145] First, multi-source sensor data is collected within a preset time window. After preprocessing the multi-source sensor data, the device's health index is calculated based on the multi-source sensor data. The device's decay rate is then calculated based on the health index, and the sensor's warning threshold is dynamically updated based on the decay rate. When the real-time data collected by the multi-source sensor data exceeds the warning threshold, a health index threshold determination is triggered. When the device's health index falls below the first threshold, an alarm is triggered and a maintenance work order is generated. When the device's health index exceeds the second threshold, a maintenance work order is generated. The first threshold is lower than the second threshold.

[0146] It should be noted that the above examples are only used to understand the present application and do not constitute a limitation on the method for dynamically determining the warning threshold of the present application. More simple transformations based on this technical concept are all within the scope of protection of the present application.

[0147] The present application provides a device for dynamically determining a warning threshold, and the device for dynamically determining a warning threshold includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method for dynamically determining the warning threshold in the above-mentioned embodiment one.

[0148] Reference below Figure 6, which shows a schematic diagram of the structure of a device suitable for implementing a dynamic determination of a warning threshold in an embodiment of the present application. The device for dynamically determining a warning threshold in an embodiment of the present application may include, but is not limited to, mobile terminals such as laptop computers and tablet computers (PADs), as well as fixed terminals such as desktop computers. Figure 6 The device for dynamically determining the warning threshold shown is merely an example and should not limit the functions and scope of use of the embodiments of the present application.

[0149] like Figure 6 As shown, the device for dynamically determining a warning threshold may include a processing device 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes based on a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage device 1003 into a random access memory (RAM) 1004. RAM 1004 also stores various programs and data required for the operation of the device for dynamically determining a warning threshold. Processing device 1001, ROM 1002, and RAM 1004 are interconnected via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems may be connected to I / O interface 1006: input device 1007 including, for example, a touch screen, touchpad, keyboard, mouse, image sensor, microphone, accelerometer, gyroscope, etc.; output device 1008 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; storage device 1003 including, for example, a magnetic tape, hard disk, etc.; and communication device 1009. Communication device 1009 may allow the device for dynamically determining the warning threshold to communicate with other devices wirelessly or by wire to exchange data. While the figure illustrates the device for dynamically determining the warning threshold with various systems, it should be understood that implementation or presence of all illustrated systems is not required. More or fewer systems may alternatively be implemented or present.

[0150] In particular, according to the embodiments disclosed in the present application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, the embodiments disclosed in the present application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program comprising program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are executed.

[0151] The device for dynamically determining a warning threshold provided in this application, employing the method for dynamically determining a warning threshold in the above-described embodiment, can address the technical problem of improving the timeliness of equipment maintenance. Compared to the prior art, the beneficial effects of the device for dynamically determining a warning threshold provided in this application are the same as those of the method for dynamically determining a warning threshold provided in the above-described embodiment. Other technical features of the device for dynamically determining a warning threshold are the same as those disclosed in the method in the above-described embodiment and are not further elaborated here.

[0152] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0153] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

[0154] The present application provides a computer-readable storage medium having computer-readable program instructions (ie, a computer program) stored thereon, wherein the computer-readable program instructions are used to execute the method for dynamically determining the warning threshold in the above-mentioned embodiment.

[0155] The computer-readable storage medium provided herein may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, systems, or devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to, an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM) or flash memory, optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including, but not limited to, wires, optical cables, radio frequency (RF), etc., or any suitable combination thereof.

[0156] The computer-readable storage medium may be included in the device for dynamically determining the early warning threshold value; or may exist independently without being assembled into the device for dynamically determining the early warning threshold value.

[0157] The computer-readable storage medium carries one or more programs. When the one or more programs are executed by the device for dynamically determining the warning threshold, the device can write computer program code for performing the operations of the present application in one or more programming languages ​​or a combination thereof. The programming languages ​​include object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as "C" or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, or as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (for example, via the Internet using an Internet service provider).

[0158] The flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, program segment or part of code, and the module, program segment or part of code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, as well as the combination of boxes in the block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or can be implemented using a combination of dedicated hardware and computer instructions.

[0159] The modules described in the embodiments of the present application may be implemented in software or hardware, wherein the name of a module does not necessarily limit the unit itself.

[0160] The computer-readable storage medium provided in this application stores computer-readable program instructions (i.e., a computer program) for executing the aforementioned method for dynamically determining a warning threshold. This computer-readable storage medium can address the technical problem of improving the timeliness of equipment maintenance. Compared to the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the method for dynamically determining a warning threshold provided in the aforementioned embodiments, and will not be further elaborated here.

[0161] The above description is only part of the embodiments of the present application and does not limit the patent scope of the present application. All equivalent structural transformations made by using the contents of the present application specification and drawings under the technical concept of the present application, or direct / indirect application in other related technical fields are included in the patent protection scope of the present application.

Claims

1. A method for dynamically determining a warning threshold, characterized in that: The method for dynamically determining the warning threshold includes: Obtaining the operation instructions received by the device within a preset time window; The timestamp corresponding to the operation instruction is used as a time sequence mark and mapped to the multi-source sensor data to obtain a data sequence with a time sequence mark; Determining a target operating period of the device according to a timing mark corresponding to the operating instruction, calculating a state transition probability according to a data sequence corresponding to the target operating period using a forward-backward algorithm, and segmenting the data sequence corresponding to the target operating period according to the state transition probability to obtain a target data sequence, including: using a hidden Markov model to divide the data sequence into operating periods according to the timing mark of the operating instruction, calculating a state transition probability for the data sequence corresponding to the target operating period obtained after the segmentation using a forward-backward algorithm, and further segmenting the data sequence corresponding to the target operating period according to the state transition probability to obtain the target data sequence; Determine the historical sensor data sequence corresponding to each historical failure mode based on historical failure cases; Obtaining an initial weight of the target data sequence; Determining a target fault mode corresponding to the target data sequence based on a similarity between the target data sequence and the historical sensor data sequence; determining, according to the faulty sensor corresponding to the target fault mode, a compensation weight for data corresponding to the faulty sensor in the target data sequence; determining a weight of the target data sequence according to an initial weight of the target data sequence and a compensation weight of data corresponding to the faulty sensor in the target data sequence; determining a health index according to the target data sequence and the weight; Fitting a linear attenuation equation according to the health index within the preset time window, and using the slope of the linear attenuation equation as the attenuation rate of the health index; An early warning threshold of the multi-source sensor is determined according to the attenuation rate.

2. The method for dynamically determining the warning threshold according to claim 1, wherein: After the step of fitting a linear attenuation equation according to the health index within the preset time window and using the slope of the linear attenuation equation as the attenuation rate of the health index, the method further includes: Creating a health index sequence according to the health index within the preset time window; performing packet decomposition on the health index sequence to obtain a low-frequency component corresponding to the health index sequence; Determining a Hurst exponent corresponding to the low-frequency component, and determining an adjustment amount for the attenuation rate based on the Hurst exponent, wherein the Hurst exponent represents the likelihood that a changing trend of the health index sequence will continue in the future; The sum of the attenuation rate and the adjustment amount of the attenuation rate is used as the new attenuation rate.

3. The method for dynamically determining a warning threshold according to any one of claims 1 to 2, characterized in that: The step of determining the warning threshold of the multi-source sensor according to the attenuation rate includes: determining an initial threshold value of the multi-source sensor according to the health index; determining a threshold adjustment amount of the multi-source sensor according to the attenuation rate; An early warning threshold of the multi-source sensor is determined according to the initial threshold and the threshold adjustment amount.

4. The method for dynamically determining the warning threshold according to claim 1, wherein: After the step of determining the warning threshold of the multi-source sensor according to the attenuation rate, the method further includes: Using the health index and the target failure mode as decision variables, and using maintenance resources as constraint variables, wherein the maintenance resources include at least one of the number of equipment spare parts, the maintenance time window, and the matching degree of maintenance personnel; Inputting the decision variables and the constraint variables into a preset work order generation model, generating at least one reference maintenance plan through the preset work order generation model, and calculating the maintenance cost corresponding to the reference maintenance plan; A target maintenance plan is determined in the reference maintenance plan according to the maintenance cost, and a maintenance work order is generated according to the target maintenance plan.

5. The method for dynamically determining the warning threshold according to claim 4, wherein: After the steps of determining a target maintenance plan from the reference maintenance plan based on the maintenance cost and generating a maintenance work order based on the target maintenance plan, the method further includes: Obtaining the health index and maintenance time of the equipment after maintenance; Calculating a reward function value corresponding to the target maintenance plan according to the health index after maintenance and the maintenance time; If the reward function value is less than a preset reward value, re-determining the decay rate of the device according to the health index after maintenance; The warning threshold of the multi-source sensor is re-determined according to the re-determined attenuation rate.

6. A device for dynamically determining an early warning threshold, characterized in that: The device comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the method for dynamically determining a warning threshold according to any one of claims 1 to 5.

7. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the method for dynamically determining the warning threshold according to any one of claims 1 to 5 are implemented.

Images