An intelligent evaluation method and system for motor partial discharge detection
Patent Information
- Application Number
- CN202610622510.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-05-08
- Publication Date
- 2026-09-22
Smart Images
Figure CN122794162A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of motor condition detection and signal processing technology, specifically an intelligent evaluation method and system for detecting partial discharge in motors. Background Technology
[0002] Partial discharge detection in motors is a crucial testing method, and the accuracy of the test results is essential for evaluating motor performance. However, in real-world industrial testing environments, extremely complex external interferences exist, making it difficult to accurately extract partial discharge signals and perform highly reliable intelligent evaluation. Currently, the industry commonly uses a single pulse current detection method for evaluating partial discharge in motors, setting a fixed partial discharge alarm threshold for the system.
[0003] However, in actual testing, detection systems relying solely on a single electrical signal are highly susceptible to external high-frequency noise, frequently triggering false alarms. Furthermore, existing safety alarm thresholds are mostly fixed absolute values, failing to consider the impact of the motor's actual operating temperature on the insulation material's tolerance. Under conditions where the motor generates high temperatures and insulation performance deteriorates, fixed thresholds cannot provide a dynamic and realistic safety safeguard. Summary of the Invention
[0004] The purpose of this invention is to provide an intelligent evaluation method and system for detecting partial discharge in motors, so as to solve the problems raised in the prior art.
[0005] To achieve the above objectives, the present invention provides the following technical solution: an intelligent evaluation method and system for detecting partial discharge in motors, comprising the following steps: S1. While the motor is running, synchronously collect multi-dimensional feature data of the motor. The multi-dimensional feature data includes partial discharge simulation electrical signals, optical image signals of the partial discharge area, and real-time operating environmental variables. The real-time operating environmental variables include the real-time surface temperature of the motor and the real-time ambient humidity. S2. Perform interference suppression processing on the simulated partial discharge electrical signal and extract characteristic discharge parameters; S3. Perform image segmentation on the optical image signal of the partial discharge region and extract feature image parameters; S4. Establish a multi-dimensional fusion evaluation model with cross-validation of electrical and optical aspects, obtain the pre-configured first confidence weight and second confidence weight, normalize the characteristic discharge parameters and the characteristic image parameters respectively, input them into the multi-dimensional fusion evaluation model for weighted calculation, and output the comprehensive discharge severity index. S5. Based on the real-time surface temperature, perform linear compensation on the preset static alarm threshold to generate a dynamic safety threshold; S6. Perform spatial localization diagnosis of discharge type based on the extraction status of the characteristic discharge parameters and the characteristic image parameters; compare the comprehensive discharge severity index with the dynamic safety threshold and output real-time status warning; predict the remaining life of the motor based on the changing trend of the comprehensive discharge severity index within a continuous time window.
[0006] Furthermore, the synchronous acquisition of multi-dimensional characteristic data of the motor specifically includes: continuously acquiring simulated electrical signals of partial discharge in the partial discharge area of the motor using a current sensor; acquiring optical image signals of the partial discharge area of the motor using an ultraviolet imaging sensor; acquiring the real-time surface temperature and real-time ambient humidity of the motor using a temperature sensor and a humidity sensor; and using a synchronous trigger control circuit with a unified hardware clock source to timestamp the simulated electrical signals of partial discharge and the optical image signals.
[0007] Furthermore, the specific processing flow of the characteristic discharge parameter includes: converting the partial discharge electrical signal into a digital sequence through analog-to-digital conversion; using an adaptive digital filtering algorithm to periodically filter out narrowband interference from the digital sequence; extracting peak values from the denoised digital sequence and statistically obtaining the average discharge quantity as the characteristic discharge parameter.
[0008] Furthermore, the specific processing flow of the feature image parameter includes: extracting the gray-level distribution matrix of the optical image signal in the ultraviolet band; based on the local spatial contrast of each pixel in the gray-level distribution matrix, using a spatial threshold segmentation algorithm to locate suspected discharge areas and mark them as candidate discharge pixel clusters; performing morphological opening and closing operations on the candidate discharge pixel clusters to filter out sporadic interference noise on the image and fill in the broken holes inside the discharge spot, thereby obtaining the reconstructed discharge target connected region; calculating the equivalent area enclosed by the geometric outer contour of the discharge target connected region and using it as the feature image parameter.
[0009] Further determine the extraction status of the characteristic discharge parameter and the characteristic image parameter: if the characteristic discharge parameter and the characteristic image parameter are extracted simultaneously, then the discharge type is diagnosed as exposed discharge on the motor surface; Furthermore, the comprehensive discharge severity index is obtained through the following steps: the first confidence weight is used to represent the assessment confidence of the electrical dimension, and the second confidence weight is used to represent the assessment confidence of the optical dimension; the characteristic discharge parameter is normalized and divided by a preset factory baseline discharge quantity to obtain the electrical anomaly ratio; the characteristic image parameter is normalized and divided by a preset safety baseline spot area to obtain the optical anomaly ratio; the multi-dimensional fusion assessment model is used to perform the following calculations: CSI = α × (Q) avg / Qref ) + β × (S / S ref ); Wherein, CSI is the Comprehensive Discharge Severity Index, Q avg Let Q be the characteristic discharge parameter. ref S is the preset factory standard discharge amount, and S is the feature image parameter. ref The preset safety reference spot area is α, the first confidence weight is β, and the second confidence weight is β. The Comprehensive Discharge Severity Index (CSI) is used to represent the current insulation damage level of the motor. The higher the value, the more severe the insulation damage of the motor and the greater the risk of insulation breakdown.
[0010] Furthermore, the default values of the first confidence weight and the second confidence weight are initially configured through experimental calibration during the system's factory deployment phase, specifically including: Partial discharge data from electrical and optical detection were collected under standard environmental conditions and compared with known reference parameters output by a standard partial discharge generator. The correct recognition rate and false recognition rate of the two types of detection results are statistically analyzed, and their respective confidence indices are calculated. The initial benchmark values of the first confidence weight and the second confidence weight are generated proportionally according to the relative size of the confidence indices. The values of the first confidence weight and the second confidence weight are dynamically adjusted. The specific adjustment logic is as follows: the real-time ambient humidity is read in real time; when the real-time ambient humidity data is greater than the preset humidity alarm threshold, the value of the second confidence weight is automatically reduced, and the value of the first confidence weight is increased proportionally.
[0011] Furthermore, the specific steps for generating the dynamic safety threshold are as follows: extracting the real-time surface temperature, retrieving the temperature compensation coefficient corresponding to the motor insulation material and the standard reference room temperature, and calculating the dynamic safety threshold Th. d The calculation formula is: Th d =Th b ×[1-k×(TT n )]; Among them, Th b The static alarm threshold is T, and the real-time surface temperature is T. n The standard reference room temperature is given, and k is the temperature compensation coefficient.
[0012] The temperature compensation coefficient k is obtained by: obtaining the temperature degradation curve corresponding to the insulation material of the motor, which is obtained by performing variable temperature dielectric strength tests on the same type of insulation material and fitting the test data; extracting the insulation attenuation change rate of the temperature degradation curve within the rated temperature range of the motor, and using it as the temperature compensation coefficient k.
[0013] Furthermore, the spatial positioning diagnosis of discharge type, output of real-time status warning, and prediction of the remaining life of the motor specifically include: determining the extraction status of the characteristic discharge parameters and the characteristic image parameters: if only the characteristic discharge parameters are successfully extracted, the discharge type is diagnosed as deep internal insulation discharge of the motor due to light signal obstruction; if the characteristic discharge parameters and the characteristic image parameters are extracted simultaneously, the discharge type is diagnosed as exposed discharge on the surface of the motor; comparing the comprehensive discharge severity index with the dynamic safety threshold generated in step S5 in real time, and triggering and outputting a real-time status warning signal when the comprehensive discharge severity index reaches the dynamic safety threshold; retrieving the historical comprehensive discharge severity index sequence within the continuous time window, constructing a degradation trend evolution model of the comprehensive discharge severity index increasing over time using a curve fitting algorithm, extrapolating the degradation trend evolution model in the positive direction along the time axis, and calculating the expected time node where the output value of the degradation trend evolution model intersects with the dynamic safety threshold, and taking the time span from the current moment to the expected time node as the remaining life of the motor.
[0014] To achieve the above objectives, a second aspect of the present invention provides an intelligent evaluation system for detecting partial discharge in a motor, comprising: The data synchronization acquisition module is used to synchronously acquire multi-dimensional feature data of the motor while the motor is running. The multi-dimensional feature data includes partial discharge simulation electrical signals, optical image signals of the partial discharge area, and real-time operating environmental variables, including the real-time surface temperature and real-time ambient humidity of the motor. The electrical feature extraction module is used to perform interference suppression processing on the partial discharge simulated electrical signal and extract characteristic discharge parameters; The optical feature extraction module is used to segment the optical image signal of the partial discharge region and extract feature image parameters. The multidimensional fusion evaluation module is used to establish a multidimensional fusion evaluation model that cross-validates electrical and optical aspects. It obtains the pre-configured first confidence weight and second confidence weight, normalizes the characteristic discharge parameters and the characteristic image parameters respectively, and then inputs them into the multidimensional fusion evaluation model for weighted calculation, outputting a comprehensive discharge severity index. The dynamic threshold generation module is used to linearly compensate the preset static alarm threshold based on the real-time surface temperature to generate a dynamic safety threshold. The comprehensive diagnosis and prediction module is used to perform spatial localization diagnosis of discharge type based on the extraction status of the characteristic discharge parameters and the characteristic image parameters; compare the comprehensive discharge severity index with the dynamic safety threshold and output real-time status warning; and predict the remaining life of the motor based on the changing trend of the comprehensive discharge severity index within a continuous time window.
[0015] Compared with the prior art, the beneficial effects of the present invention are: This invention enables cross-validation of electrical and optical data, overcoming the problem of single electrical signals being susceptible to interference; by introducing dynamic weighting of environmental humidity and temperature compensation thresholds based on material degradation characteristics, the system maintains extremely high alarm accuracy under harsh operating conditions of high temperature and high humidity; and achieves rapid diagnosis of internal and external discharges and quantitative extrapolation of the remaining lifespan of equipment. Attached Figure Description
[0016] Figure 1 This is a flowchart of an intelligent evaluation method for detecting partial discharge in a motor, as disclosed in an embodiment of the present invention. Detailed Implementation
[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0018] Example: Figure 1 As shown, the present invention provides a technical solution, an intelligent evaluation method for detecting partial discharge in motors, comprising the following steps: S1. While the motor is running, a high-frequency pulse current sensor installed on the motor's grounding wire continuously collects simulated partial discharge electrical signals, while an ultraviolet imaging sensor deployed at the motor end collects optical image signals of the area. To ensure alignment of the multidimensional data, a unified hardware clock source is used to timestamp both data at the microsecond level. Simultaneously, a temperature sensor collects the real-time surface temperature of the motor (85℃) and a humidity sensor collects the real-time ambient humidity (85%RH).
[0019] S2. Due to strong periodic narrowband interference such as inverter switching noise in industrial environments, the acquired partial discharge analog electrical signal is first converted from analog to digital to obtain the original digital sequence. In this embodiment, an LMS adaptive filter is constructed to suppress interference. The LMS filter automatically tracks and cancels out strongly correlated periodic narrowband interference, outputting a denoised digital sequence with a high signal-to-noise ratio. Finally, pulse peaks are extracted from the denoised digital sequence, and the average discharge amount within one power frequency cycle is calculated and used as a characteristic discharge parameter. In this embodiment, the characteristic discharge parameter is calculated to be 400 pC.
[0020] S3. Extract the grayscale distribution matrix of the optical image signal in the ultraviolet band; based on the local spatial contrast of each pixel in the grayscale distribution matrix, locate the suspected discharge region using a spatial threshold segmentation algorithm and mark it as a candidate discharge pixel cluster; perform morphological opening and closing operations on the candidate discharge pixel clusters to obtain the reconstructed discharge target connected region; calculate the equivalent area enclosed by the geometric outer contour of the discharge target connected region and use it as a feature image parameter. In this embodiment, the feature image parameter is calculated to be 150 pixels.
[0021] S4. In this embodiment, characteristic discharge parameters and characteristic image parameters are extracted simultaneously, and the discharge type is determined to be exposed discharge on the motor surface. The characteristic discharge parameters are normalized and divided by a preset factory baseline discharge amount to obtain the electrical anomaly percentage; the characteristic image parameters are normalized and divided by a preset safety baseline spot area to obtain the optical anomaly percentage; the multi-dimensional fusion evaluation model is used to perform the following calculations: CSI = α × (Q) avg / Q ref ) + β × (S / S ref ); Wherein, CSI is the Comprehensive Discharge Severity Index, Q avg Let Q be the characteristic discharge parameter. ref S is the preset factory standard discharge amount, and S is the feature image parameter. ref The preset safety reference spot area is α, where α is the first confidence weight and β is the second confidence weight.
[0022] The preset factory reference discharge level and safety reference spot area are determined by injecting a known intensity reference partial discharge pulse into a motor of the same model using a standard partial discharge generator under standard conditions of 25°C and 40%RH. The average value stably collected by the electrical sensor is recorded as the factory reference discharge level, and the spot area stably identified by the optical sensor is recorded as the safety reference spot area. Under the same standard conditions, the correct recognition rates of electrical and optical detection in multiple calibration experiments are statistically analyzed, and their respective confidence indices are calculated. Initial reference values for the first and second confidence weights are generated proportionally based on the relative magnitude of the confidence indices. In this embodiment, the factory reference discharge level is 500 pC, the safety reference spot area is 200 pixels, the correct recognition rate of both electrical and optical detection in multiple calibration experiments is above 95%, the relative magnitude of the confidence indices is allocated in a 1:1 ratio, and the first and second confidence weights are both 0.5. The preset humidity alarm threshold is 60%RH.
[0023] During the real-time operation of the motor, the ambient humidity measured in step S1 is 85%RH, far exceeding the preset humidity alarm threshold. Ultraviolet photons undergo severe scattering under extremely high humidity, triggering the system's dynamic adjustment logic. In this embodiment, the system has a pre-set continuous linear penalty algorithm with a humidity attenuation coefficient λ of 0.008. When the real-time ambient humidity H exceeds the humidity alarm threshold H... th At that time, the system dynamically adjusts the weights according to the following formula: The formula for the weights of the electrical dimension is: α = α1 + λ × (HH) th ); The formula for weighting the optical dimension is: β = β1 + λ × (HH) th ); Where α1 is the first confidence weight and β1 is the initial second confidence weight, the calculated first confidence weight is 0.7 and the second confidence weight is 0.3.
[0024] In this embodiment, CSI is calculated to be 0.785.
[0025] S5. The specific steps for generating the dynamic safety threshold are as follows: extract the real-time surface temperature, retrieve the temperature compensation coefficient corresponding to the motor insulation material and the standard reference room temperature, and calculate the dynamic safety threshold Th. d The calculation formula is: Th d =Th b ×[1-k×(TT n )]; Among them, Th b The static alarm threshold is T, and the real-time surface temperature is T. nThe standard reference room temperature is denoted as , and k is the temperature compensation coefficient.
[0026] The maximum safe discharge amount allowed by the motor during factory testing at standard reference room temperature is obtained and used as the static alarm threshold of the system. The temperature degradation curve corresponding to the insulation material of the motor is obtained. The temperature degradation curve is obtained by performing variable temperature dielectric strength tests on the same type of insulation material and fitting the test data. The insulation attenuation change rate of the temperature degradation curve within the rated temperature range of the motor is extracted and used as the temperature compensation coefficient.
[0027] In this embodiment, the real-time surface temperature is 85°C, the standard reference room temperature is 25°C, the motor insulation material is epoxy mica tape, the corresponding temperature compensation coefficient is 0.005, and the static alarm threshold is 1. The calculated dynamic safety threshold is 0.7.
[0028] S6. Compare the comprehensive discharge severity index of 0.785 with the dynamic safety threshold of 0.7. Since 0.785 > 0.7, the system outputs a real-time status warning signal. The historical comprehensive discharge severity index sequence within the continuous time window is retrieved, and a curve fitting algorithm is used to construct a degradation trend evolution model of the comprehensive discharge severity index as it increases over time. This degradation trend evolution model is extrapolated along the time axis in a positive direction, and it is calculated that the curve will cross the smooth line of the dynamic safety threshold after 45 days. The system outputs that the remaining lifespan of the motor is 45 days.
[0029] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.
Claims
1. An intelligent evaluation method for detecting partial discharge in motors, characterized in that: Includes the following steps: S1. While the motor is running, synchronously collect multi-dimensional feature data of the motor. The multi-dimensional feature data includes partial discharge simulation electrical signals, optical image signals of the partial discharge area, and real-time operating environmental variables. The real-time operating environmental variables include the real-time surface temperature of the motor and the real-time ambient humidity. S2. Perform interference suppression processing on the simulated partial discharge electrical signal and extract characteristic discharge parameters; S3. Perform image segmentation on the optical image signal of the partial discharge region and extract feature image parameters; S4. Establish a multi-dimensional fusion evaluation model with cross-validation of electrical and optical aspects, obtain the pre-configured first confidence weight and second confidence weight, normalize the characteristic discharge parameters and the characteristic image parameters respectively, input them into the multi-dimensional fusion evaluation model for weighted calculation, and output the comprehensive discharge severity index. S5. Based on the real-time surface temperature, perform linear compensation on the preset static alarm threshold to generate a dynamic safety threshold; S6. Compare the comprehensive discharge severity index with the dynamic safety threshold, and output a real-time status warning; The remaining lifespan of the motor is predicted based on the changing trend of the comprehensive discharge severity index within a continuous time window.
2. The intelligent evaluation method for partial discharge detection of motors according to claim 1, characterized in that: In step S1, the synchronous acquisition of multi-dimensional feature data of the motor specifically includes: continuously acquiring simulated electrical signals of partial discharge in the partial discharge region of the motor using a current sensor; acquiring optical image signals of the partial discharge region of the motor using an ultraviolet imaging sensor; acquiring the real-time surface temperature and real-time ambient humidity of the motor using a temperature sensor and a humidity sensor; and using a synchronous trigger control circuit with a unified hardware clock source to timestamp the simulated electrical signals of partial discharge and the optical image signals.
3. The intelligent evaluation method for partial discharge detection of motors according to claim 1, characterized in that: In step S2, the specific processing flow of the characteristic discharge parameter includes: converting the partial discharge electrical signal into a digital sequence through analog-to-digital conversion; using an adaptive digital filtering algorithm to filter out periodic narrowband interference from the digital sequence; extracting peak values from the denoised digital sequence and statistically obtaining the average discharge quantity as the characteristic discharge parameter.
4. The intelligent evaluation method for partial discharge detection of motors according to claim 1, characterized in that: In step S3, the specific processing flow of the feature image parameter includes: extracting the gray-level distribution matrix of the optical image signal in the ultraviolet band; based on the local spatial contrast of each pixel in the gray-level distribution matrix, locating the suspected discharge region using a spatial threshold segmentation algorithm, and marking it as a candidate discharge pixel cluster; performing morphological opening and closing operations on the candidate discharge pixel cluster to obtain the reconstructed discharge target connected region; calculating the equivalent area enclosed by the geometric outer contour of the discharge target connected region, and using it as the feature image parameter.
5. The intelligent evaluation method for partial discharge detection of motors according to claim 1, characterized in that: In step S4, the comprehensive discharge severity index is obtained through the following steps: the first confidence weight is used to represent the assessment confidence of the electrical dimension, and the second confidence weight is used to represent the assessment confidence of the optical dimension; the characteristic discharge parameter is normalized and divided by a preset factory baseline discharge quantity to obtain the electrical anomaly ratio; the characteristic image parameter is normalized and divided by a preset safety baseline spot area to obtain the optical anomaly ratio; the multi-dimensional fusion assessment model is used to perform the following calculations: CSI=α×(Q avg / Q ref )+β×(S / S ref ); Wherein, CSI is the Comprehensive Discharge Severity Index, Q avg Let Q be the characteristic discharge parameter. ref S is the preset factory standard discharge amount, and S is the feature image parameter. ref The preset safety reference spot area is α, where α is the first confidence weight and β is the second confidence weight.
6. The intelligent evaluation method for partial discharge detection of motors according to claim 5, characterized in that: The default values of the first confidence weight and the second confidence weight are initially configured through experimental calibration during the system's factory manufacturing phase. Specifically, this includes: collecting partial discharge data from electrical and optical detection under standard environmental conditions and comparing them with known benchmark parameters output by a standard partial discharge generator; calculating the correct recognition rate of the two types of detection results, calculating their respective confidence indices, and proportionally allocating the initial benchmark values of the first confidence weight and the second confidence weight according to the relative magnitude of the confidence indices. The values of the first confidence weight and the second confidence weight are dynamically adjusted. The specific adjustment logic is as follows: the real-time ambient humidity is read in real time; when the real-time ambient humidity data is greater than the preset humidity alarm threshold, the value of the second confidence weight is automatically reduced, and the value of the first confidence weight is increased proportionally.
7. The intelligent evaluation method for partial discharge detection of motors according to claim 1, characterized in that: In step S5, the specific steps for generating the dynamic safety threshold are as follows: extract the real-time surface temperature, retrieve the temperature compensation coefficient corresponding to the motor insulation material and the standard reference room temperature, and calculate the dynamic safety threshold Th. d The calculation formula is: Silk d =Th b ×[1-k×(TT n )]; Among them, Th b The static alarm threshold is T, and the real-time surface temperature is T. n The standard reference room temperature is given, and k is the temperature compensation coefficient. The temperature compensation coefficient k is obtained by: obtaining the temperature degradation curve corresponding to the insulation material of the motor, which is obtained by performing variable temperature dielectric strength tests on the same type of insulation material and fitting the test data; extracting the insulation attenuation change rate of the temperature degradation curve within the rated temperature range of the motor, and using it as the temperature compensation coefficient k.
8. The intelligent evaluation method for partial discharge detection of motors according to claim 1, characterized in that: In step S6, the output of real-time status warning and prediction of the remaining lifespan of the motor specifically includes: comparing the comprehensive discharge severity index with the dynamic safety threshold generated in step S5 in real time; when the comprehensive discharge severity index reaches the dynamic safety threshold, outputting a real-time status warning signal; retrieving the historical comprehensive discharge severity index sequence within the continuous time window; constructing a degradation trend evolution model of the comprehensive discharge severity index increasing over time using a curve fitting algorithm; extrapolating the degradation trend evolution model in the positive direction along the time axis; calculating the expected time node where the output value of the degradation trend evolution model intersects with the dynamic safety threshold; and taking the time span from the current moment to the expected time node as the remaining lifespan of the motor.
9. An intelligent evaluation system for detecting partial discharge in motors, characterized in that, include: The data synchronization acquisition module is used to synchronously acquire multi-dimensional feature data of the motor while the motor is running. The multi-dimensional feature data includes partial discharge simulation electrical signals, optical image signals of the partial discharge area, and real-time operating environmental variables, including the real-time surface temperature and real-time ambient humidity of the motor. The electrical feature extraction module is used to perform interference suppression processing on the partial discharge simulated electrical signal and extract characteristic discharge parameters; The optical feature extraction module is used to segment the optical image signal of the partial discharge region and extract feature image parameters. The multidimensional fusion evaluation module is used to establish a multidimensional fusion evaluation model that cross-validates electrical and optical aspects. It obtains the pre-configured first confidence weight and second confidence weight, normalizes the characteristic discharge parameters and the characteristic image parameters respectively, and then inputs them into the multidimensional fusion evaluation model for weighted calculation, outputting a comprehensive discharge severity index. The dynamic threshold generation module is used to linearly compensate the preset static alarm threshold based on the real-time surface temperature to generate a dynamic safety threshold. The comprehensive diagnosis and prediction module is used to compare the comprehensive discharge severity index with the dynamic safety threshold and output real-time status warnings. The remaining lifespan of the motor is predicted based on the changing trend of the comprehensive discharge severity index within a continuous time window.