WO2022246758A1 - 产品状态量化和剩余寿命预测方法、装置和系统 - Google Patents

产品状态量化和剩余寿命预测方法、装置和系统 Download PDF

Info

Publication number
WO2022246758A1
WO2022246758A1 PCT/CN2021/096502 CN2021096502W WO2022246758A1 WO 2022246758 A1 WO2022246758 A1 WO 2022246758A1 CN 2021096502 W CN2021096502 W CN 2021096502W WO 2022246758 A1 WO2022246758 A1 WO 2022246758A1
Authority
WO
WIPO (PCT)
Prior art keywords
water pump
product
cost
data
model
Prior art date
Application number
PCT/CN2021/096502
Other languages
English (en)
French (fr)
Inventor
李朝春
周晓舟
李奂轮
傅玲
Original Assignee
西门子股份公司
西门子（中国）有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 西门子股份公司, 西门子（中国）有限公司 filed Critical 西门子股份公司
Priority to KR1020237044776A priority Critical patent/KR20240011823A/ko
Priority to CN202180096844.9A priority patent/CN117178273A/zh
Priority to JP2023573001A priority patent/JP2024520472A/ja
Priority to EP21942334.0A priority patent/EP4325390A1/en
Priority to PCT/CN2021/096502 priority patent/WO2022246758A1/zh
Publication of WO2022246758A1 publication Critical patent/WO2022246758A1/zh

Links

Images

Classifications

    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F04—POSITIVE - DISPLACEMENT MACHINES FOR LIQUIDS; PUMPS FOR LIQUIDS OR ELASTIC FLUIDS
    • F04D—NON-POSITIVE-DISPLACEMENT PUMPS
    • F04D15/00—Control, e.g. regulation, of pumps, pumping installations or systems
    • F04D15/0088—Testing machines
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F04—POSITIVE - DISPLACEMENT MACHINES FOR LIQUIDS; PUMPS FOR LIQUIDS OR ELASTIC FLUIDS
    • F04D—NON-POSITIVE-DISPLACEMENT PUMPS
    • F04D15/00—Control, e.g. regulation, of pumps, pumping installations or systems
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2111/00—Details relating to CAD techniques
    • G06F2111/10—Numerical modelling

Definitions

  • the invention relates to the field of industrial digitization, in particular to a method, device and system for product state quantification and remaining life prediction.
  • pump degradation is caused by many factors, including mechanical, hydraulic, electromagnetic, vibration, and temperature, which can make it difficult to model physically.
  • operational data is often not available due to missing sensors or being available only for a short period of time.
  • detailed design data for pumps are often not available, such as pictures.
  • the first aspect of the present invention provides a product state quantification and remaining life prediction method, which includes the following steps: T1, obtaining product design data and operating data; T2, establishing a product design model based on the design data, and according to the Perform data fusion of design data and operation data to obtain the predicted performance curve of the product, and select the optimal working state point in the predicted performance curve; T3, establish the cost model and the operation elasticity model of the product based on the predicted performance curve, and according to the product's Performance requirements or operating costs are calculated to predict the remaining life of the product.
  • the product is a water pump
  • the step T2 further includes the following steps: T21, performing data preprocessing on the design data and operating data of the water pump to remove abnormal data and error data; T22, establishing The design model of the water pump, and perform the design performance evaluation and analysis of the product according to the design model, wherein the design model is a multi-scale digital twin model; T23, perform using a data fusion algorithm based on the design data and operating data Data fusion to obtain the predicted performance curve of the water pump, and improve the accuracy of the predicted performance curve by optimizing the parameters of the data fusion algorithm; T24, calculate the load of the water pump and its power and efficiency, and at the same time meet the increasing work of the water pump The optimal matching relationship within the range is used to select the optimal working state point in the predicted performance curve.
  • step T3 further includes the following steps: T31, based on the predicted performance curve, perform a regression calculation to generate a performance decay model of the water pump; T32, model the cost of the product based on the performance decay model, To calculate the full life cycle economic cost of the product and the replacement cost of the product; T33, based on the performance attenuation model and the operating data and cost parameters of the water pump, use the optimization algorithm to calculate the replacement time of the water pump; T34, based on the full load operation parameters of the water pump and decay function to calculate the operating elasticity of the pump.
  • the data fusion algorithm is MFK algorithm.
  • b and c represent parameters used for least squares curve fitting.
  • C ic is initial cost
  • C in is installation cost
  • C e is energy cost
  • C o is system management cost
  • C m is maintenance and repair cost
  • C s failure cost
  • C env is environmental cost
  • C d Removal and disposal costs.
  • LCC rep is the full life cycle economic cost of replacing the new water pump
  • LCC original is the full life cycle economic cost of the old water pump
  • T end is the time when the service life of the old water pump expires
  • T rep is the time to replace the new water pump
  • C e.rep is the energy cost of replacing the new water pump
  • C res,rep is the energy cost of using the new water pump
  • C res,original is the residual value of the old water pump
  • C res,rep is the residual value of the new water pump
  • the energy cost C e, orginal and C e.rep can be calculated by multiplying the energy consumption by the electricity price
  • Q is the pump flow
  • N is the pump speed
  • Density(Q, N) is the joint probability density function of the pump flow and speed
  • PE(t, Q, N) is the power attenuation model under different pump flow and speed
  • PriceE(t) is the predicted electricity price with time as variable.
  • the full load operating parameters include the minimum requirement Q d of water pump flow and the minimum pressure head requirement H d ,
  • f 1( ⁇ t ) represents the operating elasticity of the pump head
  • f 2( ⁇ t ) represents the operating elasticity of the pump flow
  • a second aspect of the present invention provides a product status quantification and remaining life prediction system, which includes: a processor; and a memory coupled to the processor, the memory having instructions stored therein, the instructions being executed by the processor
  • the electronic device When executed, the electronic device is made to perform an action, and the action includes: T1, acquiring product design data and operating data; T2, establishing a product design model based on the design data, and executing data based on the design data and operating data Fusion to obtain the predicted performance curve of the product, select the optimal working state point in the predicted performance curve; T3, establish the cost model and operation elasticity model of the product based on the predicted performance curve, and calculate the product according to the performance requirements or operating costs of the product Prediction of the remaining life of the product described above.
  • the product is a water pump
  • the action T2 further includes the following steps: T21, performing data preprocessing on the design data and operating data of the water pump, removing abnormal data and error data; T22, establishing The design model of the water pump, and perform the design performance evaluation and analysis of the product according to the design model, wherein the design model is a multi-scale digital twin model; T23, perform using a data fusion algorithm based on the design data and operating data Data fusion to obtain the predicted performance curve of the water pump, and improve the accuracy of the predicted performance curve by optimizing the parameters of the data fusion algorithm; T24, calculate the load of the water pump and its power and efficiency, and at the same time meet the increasing work of the water pump The optimal matching relationship within the range is used to select the optimal working state point in the predicted performance curve.
  • the action T3 further includes the following steps: T31, based on the predicted performance curve, perform a regression calculation to generate a performance decay model of the water pump; T32, model the cost of the product based on the performance decay model, To calculate the full life cycle economic cost of the product and the replacement cost of the product; T33, based on the performance attenuation model and the operating data and cost parameters of the water pump, use the optimization algorithm to calculate the replacement time of the water pump; T34, based on the full load operation parameters of the water pump and decay function to calculate the operating elasticity of the pump.
  • the data fusion algorithm is MFK algorithm.
  • b and c represent parameters used for least squares curve fitting.
  • C ic is initial cost
  • C in is installation cost
  • C e is energy cost
  • C o is system management cost
  • C m is maintenance and repair cost
  • C s failure cost
  • C env is environmental cost
  • C d Removal and disposal costs.
  • LCC rep is the full life cycle economic cost of replacing the new water pump
  • LCC original is the full life cycle economic cost of the old water pump
  • T end is the time when the service life of the old water pump expires
  • T rep is the time to replace the new water pump
  • C e.rep is the energy cost of replacing the new water pump
  • C res,rep is the energy cost of using the new water pump
  • C res,original is the residual value of the old water pump
  • C res,rep is the residual value of the new water pump
  • the energy cost C e, orginal and C e.rep can be calculated by multiplying the energy consumption by the electricity price
  • Q is the flow rate of the water pump
  • N is the speed of the water pump
  • Density(Q, N) is the joint probability density function of the water pump flow and speed
  • P E (t, Q, N) is the power attenuation under different water pump flow and speed
  • Price E (t) is the predicted electricity price with time as a variable.
  • the full load operating parameters include the minimum requirement Q d of water pump flow and the minimum pressure head requirement H d ,
  • f 1 ( ⁇ t) represents the operating elasticity of the pump head
  • f 2 ( ⁇ t) represents the operating elasticity of the pump flow
  • the third aspect of the present invention provides a product state quantification and remaining life prediction device, which includes: an acquisition device, which obtains product design data and operating data; a fusion device, which establishes a product design model based on the design data, and according to Perform data fusion of the design data and operating data to obtain a predicted performance curve of the product, and select an optimal working state point in the predicted performance curve; a computing device, which establishes a product cost model and an operating elasticity model based on the predicted performance curve , calculating the remaining life prediction of the product according to the performance requirement or operating cost of the product.
  • a fourth aspect of the present invention provides a computer program product tangibly stored on a computer-readable medium and comprising computer-executable instructions which, when executed, cause at least one processor to perform the The method described in the first aspect of the present invention.
  • a fifth aspect of the present invention provides a computer readable medium having stored thereon computer executable instructions which, when executed, cause at least one processor to perform the method according to the first aspect of the present invention.
  • the product state quantification and remaining life prediction mechanism of the present invention is reliable, accurate, and capable of performing quantification and prediction based on the dynamic performance of the product.
  • the invention is particularly applicable to water pumps.
  • the invention can quantify the dynamic performance state of the water pump based on the dynamic digital model of the water pump, and can quantify the remaining service life of the water pump based on the attenuation model of the water pump.
  • the present invention can provide products for process automation solution providers, or provide value-added service packages for water pump producers.
  • Figure 1 is the pressure head curve, power curve and efficiency curve of the water pump
  • Fig. 2 is a schematic diagram of a performance curve generated based on water pump design data and operating data for state quantification and remaining life prediction of a product according to a specific embodiment of the present invention
  • Fig. 3 is the water pump pressure head decay curve of the state quantification and remaining life prediction of the product according to a specific embodiment of the present invention
  • Fig. 4 is a comparison diagram of the replacement of old and new water pumps according to the status quantification and remaining life prediction of products according to a specific embodiment of the present invention
  • Fig. 5 is a schematic structural diagram of a product state quantification and remaining life prediction system according to a specific embodiment of the present invention.
  • Fig. 6 is the water pump power increase curve of the state quantification and remaining life prediction of the product according to a specific embodiment of the present invention.
  • FIG. 7 is an operating elasticity curve for condition quantification and remaining life prediction of a product according to a specific embodiment of the present invention.
  • the present invention provides a state quantification and remaining life prediction mechanism which is especially suitable for water pumps.
  • the invention models the design performance and predicted performance of the water pump based on digital twin technology.
  • the present invention combines the water pump operation data and performance curve to establish a data-driven digital twin model of the water pump to evaluate the performance of the water pump in operation, so as to obtain the remaining life prediction of the water pump in terms of performance requirements or operating costs.
  • the present invention will be described below by taking a water pump as an example.
  • the first aspect of the present invention provides a method for product state quantification and remaining life prediction, which includes step T1, step T2 and step T3.
  • Fig. 5 is a schematic structural diagram of a state quantification and remaining life prediction system according to a specific embodiment of the present invention.
  • the system 100 for product state quantification and remaining life prediction includes a performance analysis device 110 and a remaining life prediction device 120 .
  • the input data of the performance analysis device 110 is the design data D and the operation data O of the water pump, and the output data is the optimal working state point P.
  • the performance analysis device 110 includes a data preprocessing device 111 , a design modeling device 112 , a data fusion device 113 and an optimization device 114 .
  • the performance analysis device 110 outputs the predicted performance curve of the water pump to the remaining service life prediction device 120, and the remaining service life prediction device 120 outputs the remaining life prediction calculated according to the performance requirements or operating costs of the water pump.
  • the remaining service life prediction device 120 includes a performance decay modeling device 121 , a cost modeling device 122 , a cost optimization device 123 and an operation elasticity modeling device 124 .
  • step T1 is executed to obtain the design data and operation data of the product.
  • the design data includes the speed, flow, pressure and power of the water pump
  • the operating data also includes the speed, flow, pressure and power of the water pump under different wear conditions.
  • Design data refers to the data value that you want to achieve when designing the pump, which is idealized.
  • the running data is the corresponding data generated when the water pump is actually running, and it is the live data.
  • the design data includes the lift of the water pump.
  • the lift of a centrifugal pump is also called the head of the pump, which refers to the energy obtained by the fluid per unit weight through the pump.
  • the head of the pump depends on the structure of the pump, such as the diameter of the impeller, the bending of the blades, etc., and the speed.
  • the pressure head of the pump cannot be calculated accurately in theory, and it is generally determined by experimental methods.
  • the operating data includes the speed and flow of the water pump, and the time series data is classified according to the speed.
  • step T2 the performance analysis device 110 establishes a design model of the product based on the design data, and performs data fusion according to the design data and operating data to obtain a predicted performance curve of the product, and selects the optimization work in the predicted performance curve status point.
  • the step T2 further includes sub-step T21, sub-step T22, sub-step T23 and sub-step T24.
  • the data preprocessing device 111 performs data preprocessing on the design data and operating data to remove abnormal data and error data.
  • the data pre-execution module collects the design data and execution data of the water pump, and eliminates outliers.
  • outliers include using algorithms to remove burrs, remove outliers, and so on.
  • the design modeling device 112 establishes a product design model based on the design data, and performs design performance evaluation and analysis of the product according to the design model, wherein the design model is a multi-scale digital twin Model.
  • the design performance refers to the factory performance of the pump, which is an idealized performance.
  • a multi-scale digital twin refers to a live performance digital twin including a water pump head curve, power curve and efficiency curve
  • the water pump performance digital twin model utilizes a multi-scale data-driven model.
  • it has three different curves to evaluate performance, including head curve, power curve and efficiency curve.
  • the above three curves are usually used in the selection of pumps in the design stage of pump production lines. However, the above curves will not be continuously used during subsequent system operation.
  • the invention combines the water pump operation data and the water pump pressure head curve, power curve and efficiency curve to establish a data-driven digital twin model of the water pump to evaluate the performance of the water pump in operation.
  • multi-scale modeling technology is applied to the above modeling process.
  • Figure 1 shows the pressure head curve, power rate curve and efficiency curve of the water pump, based on which the design model of the water pump can be implemented. Among them, different scales of the model describe the dynamic system of the water pump.
  • the vertical coordinates represent the pressure head, efficiency (%) and input/output (KW)
  • the horizontal coordinates represent the flow per minute of the pump
  • the curve S 3 represents the performance of the pump
  • the curve S 4 represents the efficiency
  • the curve S 5 represents the motor input (KW) .
  • step T23 the data fusion device 113 uses a data fusion algorithm to perform data fusion according to the design data and operating data to obtain a predicted performance curve of the product, and improves the accuracy of the predicted performance curve by optimizing parameters of the data fusion algorithm.
  • Fig. 2 is a schematic diagram of generating a performance curve based on water pump design data and operating data for state quantification and remaining life prediction according to a specific embodiment of the present invention.
  • the design data includes the lift of the pump.
  • the lift of the centrifugal pump is also called the head of the pump, which refers to the energy obtained by the fluid per unit weight through the pump.
  • the head of the pump depends on the structure of the pump, such as the diameter of the impeller, the bending of the blades, etc., and the speed.
  • the pressure head of the pump cannot be calculated accurately in theory, and it is generally determined by experimental methods.
  • the operating data includes the speed and flow of the water pump, and the time series data is classified according to the speed.
  • the performance curve of the pump can be obtained by using the MFK algorithm (Multi-Fidelity Kriging) based on the above design data and operating data. Specifically, as shown in the right figure of Fig. 2, according to the above-mentioned design data of Yang Cheng and the operation data of speed and flow rate, the performance curve can be obtained according to the MFK algorithm. It can be seen that the actual data of the water pump is attenuated after a period of time. It is different from the expected design data when it was shipped from the factory. The historical performance curve and the real-time performance curve of the water pump will be obtained by executing the above process iteratively.
  • MFK algorithm Multi-Fidelity Kriging
  • the accuracy of the predicted performance curve can be improved by optimizing the parameters of the data fusion algorithm.
  • f(x) ⁇ GP(m(x), k(x, x')) can be described by an input vector x, an average function m(x) and a covariance function k(x, x').
  • the sub-element model method is also called Co-Kriging in the field of geostatistics.
  • the present invention considers four different Co-Kriging methods, which utilize x1 as the LFM input vector, x2 as the HFM input vector, and f1 and f2 as the respective function values:
  • ARCK autoregressive Co-Kriging
  • DMGP deep multi-scale Gaussian function
  • SCK Simple Co-Kriging
  • SMGP Simple Multi-fidelity Gaussian Processes
  • kij is a specific covariance function for the total covariance matrix K , ARCK and DMGP use:
  • k 11 k 1 (x 1 , x 1 )
  • k 12 ⁇ k 1 (x 1 , x 2 )
  • k 21 ⁇ k 1 (x 2 , x 1 )
  • k 22 ⁇ 2 k 1 (x 2 , x 1 )+k 2 (x 2 , x 2 )
  • ⁇ is the autoregressive factor
  • D is the dimension
  • k1 and k2 are sum functions in the following way:
  • DMGP utilizes a neural network as an input deformation h(x) instead of the raw input vector x.
  • SCK and SMGP utilize:
  • k 12 a 1 b 1 k 1 (x 1 , x 2 )+a 2 b 2 k 2 (x 1 , x 2 )
  • k 21 a 1 b 1 k 1 k 1 (x 2 , x 1 )+a 2 b 2 k 2 (x 2 , x 1 )
  • SMGP utilizes a neural network as an input transform h(x) instead of the raw input vector x.
  • the present invention utilizes a sequential multi-fidelity meta-modeling (SMF, sequential multi-fidelity meta-modeling) method.
  • SMF sequential multi-fidelity meta-modeling
  • SMF is an optimization algorithm that starts with certain initial samples and adds samples from LFM and HFM taking into account the computational cost of different samples. Mainly, samples are added where the confidence interval of the learned meta-model is largest. In the end, we get the most probable metamodel, constrained by the maximum computational budget.
  • the optimization device 114 calculates the best matching relationship between the load of the product and its power and efficiency while satisfying the growing working range of the product, so as to select the optimal working state point in the predicted performance curve .
  • the current performance curve is used to judge which load efficiency is the highest and the power is the best when the current pump is running, and the best operating point of the water pump is found on the curve, that is, at what speed the water pump is most cost-effective. That is, to find the best matching relationship between the rotational speed, output power and efficiency, while satisfying that the pressure and flow are within the normal working range.
  • step T3 is executed, the remaining service life prediction device 120 establishes a product cost model and an operating elasticity model based on the predicted performance curve, and calculates the remaining service life prediction of the product according to the product performance requirements or operating costs.
  • the step T3 includes sub-step T31, sub-step T32, sub-step T33 and sub-step T34.
  • the performance degradation modeling means 121 performs a regression calculation to generate a performance degradation model of the water pump based on the predicted performance curve. Specifically, by performing step T2 iteratively, the historical performance curve and the current performance curve of the water pump, as well as the performance curves of the water pump operating under different wear conditions can be obtained based on the predicted performance curve.
  • the pressure head attenuation model of the pump is described by the formula as a time-series pressure head loss model as follows:
  • b and c represent parameters used for least squares curve fitting.
  • b and c are the regression parameters calculated by the least squares regression of four points in the historical performance curve of the water pump.
  • the pump head decay curve can be obtained, the ordinate is the pump head loss (%), and the abscissa is time (year). As shown in Figure 3, with the increase of the pump usage time, the loss value of the pressure head increases gradually and greatly.
  • the cost modeling device 122 models the cost of the product based on the performance decay model, so as to calculate the full life cycle economic cost of the product and the replacement cost of the product.
  • the remaining service life is calculated from two aspects of economy and operational flexibility.
  • life cycle cost of the pump (LCC, Life Cycle Cost) is the cost of the entire life cycle of the pump, such as procurement, installation, operation, maintenance and disposal.
  • the replacement cost (RC, Replacement Cost) of the water pump is the additional cost due to the replacement of the water pump at a specific time in the future, taking into account different life cycle costs.
  • the remaining useful life from an economic point of view can be defined in terms of the length of time between the present and a certain point in the future at which the replacement cost of the pump is minimal.
  • the first design life cycle LC 11 and the second design life cycle LC 12 of the above-mentioned old water pump represent the factory design life of the water pump, so they do not take into account the actual working conditions and wear conditions, and are idealized life unreal lifespan.
  • the old water pump is replaced with a new water pump
  • from time T 0 to T rep1 represents the first design life cycle LC 21 of the new water pump
  • from time T rep1 to time T rep2 represents the second design life cycle LC 12 of the new water pump
  • the first design life cycle LC 21 and the second design life cycle LC 22 of the above-mentioned new water pump represent the actual life of the new water pump, which is the real life in consideration of the actual working conditions and wear conditions.
  • the life cycle cost of the water pump is:
  • C ic is the initial cost, including the price of pumps, systems, pipelines, procurement and additional services.
  • C in is the installation cost.
  • C e is the cost of energy, the estimated cost of system operation for pump drives, controls and any additional services.
  • C o is the human operation cost such as system management.
  • C m is maintenance and repair costs, where maintenance includes routine and predictive maintenance.
  • C s is the failure cost, such as lost production.
  • C env is the environmental cost such as removal of contaminants from pumped fluids and accessories.
  • Cd is the cost of demolition and disposal, including the cost of restoration of the local environment and disposal of additional services.
  • the replacement cost of the pump is therefore:
  • LCC rep is the cost of replacing the new water pump
  • LCC original is the cost of the old water pump
  • T end is the time when the service life of the old water pump expires
  • T rep is the time to replace the new water pump
  • C e.rep is the time to replace the new water pump C res,rep is the energy cost of the new water pump
  • C res,original is the residual value of the old water pump
  • C res,rep is the residual value of the new water pump, so as time goes by, the energy cost of the old water pump C e, orginal is rising, and the energy cost C res, rep of the new water pump is decreasing.
  • C ic , C in , and C d are one-time expenses.
  • C o , C m , C s , and C env are fixed costs that are continuous in time. Therefore, no matter whether the water pump is replaced or not, the above-mentioned costs C o , C m , C s , and C env are the same, so they can be ignored.
  • the energy cost C e, orginal and C e.rep can be calculated by multiplying the energy consumption by the electricity price
  • Density(Q, N) is the joint probability density function of pump flow Q and speed N
  • PE (t, Q, N) is the power updated by the power decay model under different pump flow Q and speed N
  • Price E (t) is the predicted electricity price with time as a variable.
  • the cost optimization device 123 uses an optimization algorithm to calculate the replacement time of the water pump based on the performance attenuation model, the operation data of the water pump, and the cost parameters. That is, the pump replacement time ⁇ t should be chosen to minimize the replacement cost RC or the highest economic benefit, that is, make the replacement cost RC the smallest.
  • the optimization algorithm includes intelligent optimization algorithm, probability density function or golden section interpolation.
  • the operation data of the water pump includes historical operation data of the flow rate and rotational speed of the water pump, the transmission efficiency of the water pump, and the click efficiency of the water pump.
  • Cost parameters include: the initial cost C ic of the one-time cost of the water pump, the installation cost C in , the dismantling and disposal cost C d ; the remaining service life of the water pump T end1 -T now , where T end1 represents the first life cycle LC of the old water pump 11 planting time, T now means the current time; electricity price.
  • the remaining service life ⁇ t can be calculated using an average optimization algorithm to minimize the pump renewal cost.
  • the abscissa of Fig. 6 is time, and the ordinate is the power of the water pump.
  • Curve S 1 is the water pump performance attenuation curve predicted at the current moment.
  • Curve S 2 is the performance attenuation curve assuming that the old water pump will be replaced by a new water pump at a certain point in the future.
  • T end1 represents the planting of LC 11 in the first life cycle of the old water pump Time, T now means the current time, and T rep means the time when the old water pump is replaced with a new water pump.
  • the replacement cost RC( ⁇ t) needs to be minimized by the following function:
  • the operation elasticity modeling device 124 evaluates the operation elasticity of the water pump based on the actual requirements of the operation of the water pump, that is, the remaining service life of the water pump under extreme usage conditions. Wherein, the extreme conditions of use are full-time and complex conditions, if the extreme conditions of use are not met under full-time and full-load, the water pump needs to be replaced.
  • the required data include the minimum demand of the flow rate Q d of the water pump, and the minimum pressure head demand H d under the minimum flow rate.
  • the operating elasticity curve shown in Figure 7 can be obtained.
  • the vertical coordinate in Figure A represents the flow rate of the water pump
  • the vertical coordinate in the figure below represents the pressure head of the water pump
  • the horizontal coordinate represents time.
  • the curve S 11 represents the water pump flow under the condition of full load of the water pump
  • the curve S 12 represents the water pump flow under a specific pressure head
  • the dotted line represents the minimum flow demand Q d .
  • Curve S 21 represents the water pump pressure head under the condition of full load of the water pump
  • curve S 22 represents the water pump pressure head under a specific flow condition
  • the dotted line represents the minimum pressure head demand H d .
  • the zero edge points P1 and P2 indicate that the headroom of the water pump is gone.
  • the water pump is always in a dynamic substantive operation during daily operation to meet the load demand in the later stage. Therefore, the pump needs some optional flexibility to respond to pressure and flow up and down.
  • the flexibility advantage diminishes over time as the performance of the pump degrades. Since the pump performance margin from the pressure head or flow demand in the time period from now to a certain point in the future is zero at that time, the remaining service life in terms of flexibility can be defined.
  • the Qd velocity is the minimum requirement for flow and Hd is the minimum head requirement at the minimum flow Qd .
  • the following formulas give the optional pump head and flow operating elasticity:
  • Q d is the minimum flow requirement
  • H d is the minimum pressure head requirement at the minimum flow rate
  • Q(H, N, t) is the flow decay function
  • H(Q, N, t) is the pressure decay function
  • the operating elasticity of the pressure head of the water pump or the operating elasticity of the water pump flow rate is less than a predetermined value, it is necessary to send a reminder that the water pump needs to be replaced.
  • f 1 ( ⁇ t) or f 2 ( ⁇ t ) is equal to 0, it means that the water pump needs to be replaced. Therefore, when the above formula is 0, the operating elastic remaining service life can be calculated. By calculating the remaining useful life from these two aspects, the end user can choose which to rotate to make more decisions in the future, such as shutting down the water pump or performing a replacement on the water pump.
  • a second aspect of the present invention provides a product status quantification and remaining life prediction system, which includes: a processor; and a memory coupled to the processor, the memory having instructions stored therein, the instructions being executed by the processor
  • the electronic device When executed, the electronic device is made to perform an action, and the action includes: T1, acquiring product design data and operating data; T2, establishing a product design model based on the design data, and executing data based on the design data and operating data Fusion to obtain the predicted performance curve of the product, select the optimal working state point in the predicted performance curve; T3, establish the cost model and operation elasticity model of the product based on the predicted performance curve, and calculate the product according to the performance requirements or operating costs of the product Prediction of the remaining life of the product described above.
  • the product is a water pump
  • the action T2 further includes the following steps: T21, performing data preprocessing on the design data and operating data of the water pump, removing abnormal data and error data; T22, establishing The design model of the water pump, and perform the design performance evaluation and analysis of the product according to the design model, wherein the design model is a multi-scale digital twin model; T23, perform using a data fusion algorithm based on the design data and operating data Data fusion to obtain the predicted performance curve of the water pump, and improve the accuracy of the predicted performance curve by optimizing the parameters of the data fusion algorithm; T24, calculate the load of the water pump and its power and efficiency, and at the same time meet the increasing work of the water pump The optimal matching relationship within the range is used to select the optimal working state point in the predicted performance curve.
  • the action T3 further includes the following steps: T31, based on the predicted performance curve, perform a regression calculation to generate a performance decay model of the water pump; T32, model the cost of the product based on the performance decay model, To calculate the full life cycle economic cost of the product and the replacement cost of the product; T33, based on the performance attenuation model and the operating data and cost parameters of the water pump, use the optimization algorithm to calculate the replacement time of the water pump; T34, based on the full load operation parameters of the water pump and decay function to calculate the operating elasticity of the pump.
  • the data fusion algorithm is MFK algorithm.
  • b and c represent parameters used for least squares curve fitting.
  • C ic is initial cost
  • C in is installation cost
  • C e is energy cost
  • C o is system management cost
  • C m is maintenance and repair cost
  • C s failure cost
  • C env is environmental cost
  • C d Removal and disposal costs.
  • LCC rep is the full life cycle economic cost of replacing the new water pump
  • LCC original is the full life cycle economic cost of the old water pump
  • T end is the time when the service life of the old water pump expires
  • T rep is the time to replace the new water pump
  • C e.rep is the energy cost of replacing the new water pump
  • C res,rep is the energy cost of using the new water pump
  • C res,original is the residual value of the old water pump
  • C res,rep is the residual value of the new water pump
  • the energy cost C e, orginal and C e.rep can be calculated by multiplying the energy consumption by the electricity price
  • Q is the flow rate of the water pump
  • N is the speed of the water pump
  • Density(Q, N) is the joint probability density function of the water pump flow and speed
  • P E (t, Q, N) is the power attenuation under different water pump flow and speed
  • Price E (t) is the predicted electricity price with time as a variable.
  • the full load operating parameters include the minimum requirement Q d of water pump flow and the minimum pressure head requirement H d ,
  • f 1( ⁇ t ) represents the operating elasticity of the pump head
  • f 2( ⁇ t ) represents the operating elasticity of the pump flow
  • the third aspect of the present invention provides a product state quantification and remaining life prediction device, which includes: an acquisition device, which obtains product design data and operating data; a fusion device, which establishes a product design model based on the design data, and according to Perform data fusion of the design data and operating data to obtain a predicted performance curve of the product, and select an optimal working state point in the predicted performance curve; a computing device, which establishes a product cost model and an operating elasticity model based on the predicted performance curve , calculating the remaining life prediction of the product according to the performance requirement or operating cost of the product.
  • a fourth aspect of the present invention provides a computer program product tangibly stored on a computer-readable medium and comprising computer-executable instructions which, when executed, cause at least one processor to perform the The method described in the first aspect of the present invention.
  • a fifth aspect of the present invention provides a computer readable medium having stored thereon computer executable instructions which, when executed, cause at least one processor to perform the method according to the first aspect of the present invention.
  • the product state quantification and remaining life prediction mechanism of the present invention is reliable, accurate, and capable of performing quantification and prediction based on the dynamic performance of the product.
  • the invention is particularly applicable to water pumps.
  • the invention can quantify the dynamic performance state of the water pump based on the dynamic digital model of the water pump, and can quantify the remaining service life of the water pump based on the attenuation model of the water pump.
  • the present invention can provide products for process automation solution providers, or provide value-added service packages for water pump producers.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Hardware Design (AREA)
  • Evolutionary Computation (AREA)
  • Geometry (AREA)
  • General Physics & Mathematics (AREA)
  • Mechanical Engineering (AREA)
  • Control Of Positive-Displacement Pumps (AREA)
  • Control Of Non-Positive-Displacement Pumps (AREA)
  • Details And Applications Of Rotary Liquid Pumps (AREA)

Abstract

一种产品状态量化和剩余寿命预测方法，获取产品的设计数据和运行数据，基于设计数据建立产品的设计模型，并根据设计数据和运行数据执行数据融合以获得产品的预测性能曲线，选取预测性能曲线中优化工作状态点，基于预测性能曲线建立产品的费用模型和操作弹性模型，根据产品的性能需求或者运行成本计算产品的剩余寿命预测，并且能够基于产品的动态性能执行量化和预测。此外，还公开了实现方法的系统、装置。

Description

产品状态量化和剩余寿命预测方法、装置和系统 技术领域
本发明涉及工业数字化领域，尤其涉及产品状态量化和剩余寿命预测方法、装置和系统。
背景技术
许多公司(例如自来水厂所有者)拥有自己的泵设施，只考虑到泵系统的最初采购和安装费用。对于产线设计者或者管理者来说，在安装主要新设备或者执行主要维修以前评估不同方案的效果是基本工作。这样的评估会验证最具经济吸引力的选择。然而，在整个生命周期或者遭受一些失效的过程中泵的效果是逐渐衰减的，其会造成一些水切断和意外关闭的损失。由于国家和全球市场会继续变得更有竞争力，生产线所有者必须持续探索降低成本，以改善其运营的利润率。由于节约成本的来源，特别是最小化能源消耗和生产线停工期，生产线设备运营会受到关注。因此，客户或者所有者想要知道现有性能状态和为泵维持和替换服务保持有用的数量上的使用寿命。
因此，对于量化泵性能和剩余使用寿命来说还存在很多挑战。例如，泵的衰退由很多因素导致，包括机械上的、液压的、电磁的、震动和温度等因素，其会导致难于建立物理模型。例如，由于缺少传感器或者仅在一个短时间周期内有效，运行数据通常不能完成。再例如，泵的细节设计数据通常并不可用，例如图片。
发明内容
本发明第一方面提供了产品状态量化和剩余寿命预测方法，其中，包括如下步骤：T1,获取产品的设计数据和运行数据；T2,基于所述设计数据建立产品的设计模型，并根据所述设计数据和运行数据执行数据融合以获得产品的预测性能曲线，选取所述预测性能曲线中的优化工作状态点；T3,基于所述预测性能曲线建立产品的费用模型和操作弹性模型，根据产品的性能需求或者运行成本计算所述产品的剩余寿命预测。
进一步地，所述产品为水泵，所述步骤T2还包括如下步骤：T21，对所述水泵的设计数据和运行数据执行数据预处理，去除异常数据和错误数据；T22，基于所述设计数据建立水泵的设计模型，并且根据所述设计模型执行所述产品的设计性能评价和分析，其中，所述设计模型是多尺度数字双胞胎模型；T23，根据所述设计数据和运行数据利用数据融合算法执行数据融合以获得水泵的预测性能曲线，并且通过优化数据融合算法的参数提高所述预测性能曲线的准确性；T24，计算水泵的负荷及其功率和效率之间并同时满足所述水泵的增长工作范围内的优化匹配关系，以选取所述预测性能曲线中的优化工作状态点。
进一步地，所述步骤T3还包括如下步骤：T31，基于所述预测性能曲线，执行一个回归计算来产生一个水泵的性能衰减模型；T32，基于所述性能衰减模型对产品的成本进行建模，以计算所述产品的全生命周期经济成本和产品的替换成本；T33，基于性能衰减模型和水泵的运行数据、成本参数，利用优化算法计算水泵的替换时间；T34，基于水泵的满负荷运行参数和衰减函数计算水泵的操作弹性。
进一步地，所述数据融合算法为MFK算法。
进一步地，所述泵的性能衰减模型通过公式描述成时序的压头损失的模型如下：
P(t)＝P N-bt c
其中，P N表示初始性能，也就是t＝0时的设计点，即水泵刚出厂时的性能。b和c表示用于最小二乘法曲线拟合的参数。
进一步地，所述全生命周期经济成本为：
LCC＝C ic+C in+C e+C o+C m+C s+C env+C d
其中，C ic为初始成本，C in是安装成本，C e是能量成本，C o是系统管理成本，C m是维持和维修成本，C s是故障成本,C env是环境成本,C d是拆除和处理成本。
进一步地，所述替换成本为：
Figure PCTCN2021096502-appb-000001
其中，LCC rep是替换新水泵的全生命周期经济成本，LCC original是旧水泵的全 生命周期经济成本，其中，T end是旧水泵使用周期到期的时间，T rep是替换新水泵的时间，C e.rep是替换新水泵的能量成本，C res，rep是用新水泵的能量成本，C res，original是旧水泵的剩余价值，C res，rep是新水泵的剩余价值，
其中，能量成本C e，orginal和C e.rep能够用能量功耗乘以电价来计算，
Figure PCTCN2021096502-appb-000002
其中，Q是水泵流量，N是水泵转速 ，Density(Q，N)是水泵流量和转速的联合概率密度函数， PE(t，Q，N)是在不同水泵流量和转速下的由功率衰减模型更新的功率， PriceE(t)是时间作为变量的预测电价。
进一步地，所述满负荷运行参数包括水泵流量的最小要求Q d和最小压头需求H d，
其中，f 1(Δt )表示水泵压头的操作弹性，f 2(Δt )表示水泵流量的操作弹性：
f 1(Δt)＝|H(Δt)-H d|
f 2(Δt)＝|Q(Δt)-Q d|
当所述水泵压头的操作弹性f 1(Δt )或者水泵流量的操作弹性f 2(Δt )小于一个预定阈值时，则需要发送需要更换水泵提示。
本发明第二方面提供了产品状态量化和剩余寿命预测系统，其中，包括：处理器；以及与所述处理器耦合的存储器，所述存储器具有存储于其中的指令，所述指令在被处理器执行时使所述电子设备执行动作，所述动作包括：T1,获取产品的设计数据和运行数据；T2,基于所述设计数据建立产品的设计模型，并根据所述设计数据和运行数据执行数据融合以获得产品的预测性能曲线，选取所述预测性能曲线中的优化工作状态点；T3,基于所述预测性能曲线建立产品的费用模型和操作弹性模型，根据产品的性能需求或者运行成本计算所述产品的剩余寿命预测。
进一步地，所述产品为水泵，所述动作T2还包括如下步骤：T21，对所述水泵的设计数据和运行数据执行数据预处理，去除异常数据和错误数据；T22，基于所述设计数据建立水泵的设计模型，并且根据所述设计模型执行所述产品的设计性能评价和分析，其中，所述设计模型是多尺度数字双胞胎模型；T23，根据所述设计数据和运行数据利用数据融合算法执行数据融合以获得水泵的预测性能曲线，并且通过优化数据融合算法的参数提高所述预测性 能曲线的准确性；T24，计算水泵的负荷及其功率和效率之间并同时满足所述水泵的增长工作范围内的优化匹配关系，以选取所述预测性能曲线中的优化工作状态点。
进一步地，所述动作T3还包括如下步骤：T31，基于所述预测性能曲线，执行一个回归计算来产生一个水泵的性能衰减模型；T32，基于所述性能衰减模型对产品的成本进行建模，以计算所述产品的全生命周期经济成本和产品的替换成本；T33，基于性能衰减模型和水泵的运行数据、成本参数，利用优化算法计算水泵的替换时间；T34，基于水泵的满负荷运行参数和衰减函数计算水泵的操作弹性。
进一步地，所述数据融合算法为MFK算法。
进一步地，所述泵的性能衰减模型通过公式描述成时序的压头损失的模型如下：
P(t)＝P N-bt c
其中，P N表示初始性能，也就是t＝0时的设计点，即水泵刚出厂时的性能。b和c表示用于最小二乘法曲线拟合的参数。
进一步地，所述全生命周期经济成本为：
LCC＝C ic+C in+C e+C o+C m+C s+C env+C d
其中，C ic为初始成本，C in是安装成本，C e是能量成本，C o是系统管理成本，C m是维持和维修成本，C s是故障成本,C env是环境成本,C d是拆除和处理成本。
进一步地，所述替换成本为：
Figure PCTCN2021096502-appb-000003
其中，LCC rep是替换新水泵的全生命周期经济成本，LCC original是旧水泵的全生命周期经济成本，其中，T end是旧水泵使用周期到期的时间，T rep是替换新水泵的时间，C e.rep是替换新水泵的能量成本，C res，rep是用新水泵的能量成本，C res，original是旧水泵的剩余价值，C res，rep是新水泵的剩余价值，
其中，能量成本C e，orginal和C e.rep能够用能量功耗乘以电价来计算，
Figure PCTCN2021096502-appb-000004
其中，Q是水泵流量，N是水泵转速，Density(Q，N)是水泵流量和转速的联合概率密度函数，P E(t，Q，N)是在不同水泵流量和转速下的由功率衰减模型更新的功率，Price E(t)是时间作为变量的预测电价。
进一步地，所述满负荷运行参数包括水泵流量的最小要求Q d和最小压头需求H d，
其中，f 1(Δt)表示水泵压头的操作弹性，f 2(Δt)表示水泵流量的操作弹性：
f 1(Δt)＝|H(Δt)-H d|
f 2(Δt)＝|Q(Δt)-Q d|
当所述水泵压头的操作弹性f 1(Δt)或者水泵流量的操作弹性f 2(Δt)小于一个预定阈值时，则需要发送需要更换水泵提示。
本发明第三方面提供了产品状态量化和剩余寿命预测装置，其中，包括：获取装置，其获取产品的设计数据和运行数据；融合装置,其基于所述设计数据建立产品的设计模型，并根据所述设计数据和运行数据执行数据融合以获得产品的预测性能曲线，选取所述预测性能曲线中的优化工作状态点；计算装置,其基于所述预测性能曲线建立产品的费用模型和操作弹性模型，根据产品的性能需求或者运行成本计算所述产品的剩余寿命预测。
本发第四方面提供了计算机程序产品，所述计算机程序产品被有形地存储在计算机可读介质上并且包括计算机可执行指令，所述计算机可执行指令在被执行时使至少一个处理器执行根据本发明第一方面所述的方法。
本发明第五方面提供了计算机可读介质，其上存储有计算机可执行指令，所述计算机可执行指令在被执行时使至少一个处理器执行根据本发明第一方面所述的方法。
本发明的产品状态量化和剩余寿命预测机制可靠、精确，并且能够基于产品的动态性能执行量化和预测。本发明特别适用于水泵。本发明能够基于水泵的动态数字模型量化水泵的动态性能状态，并能够基于水泵衰减模型量化水泵剩余使用寿命。本发明能够提供给工艺自动化解决方案提供者产品，或者为水泵生产者提供价值增加服务包。
附图说明
图1是水泵的压头曲线、功率曲线和效率曲线；
图2是根据本发明一个具体实施例的产品的状态量化和剩余寿命预测的基于水泵设计数据和运行数据产生性能曲线的示意图；
图3是根据本发明一个具体实施例的产品的状态量化和剩余寿命预测的水泵压头衰减曲线；
图4是根据本发明一个具体实施例的产品的状态量化和剩余寿命预测的水泵替换新旧对比图；
图5是根据本发明一个具体实施例的产品的状态量化和剩余寿命预测系统的结构示意图；
图6是根据本发明一个具体实施例的产品的状态量化和剩余寿命预测的水泵功率增加曲线；
图7是根据本发明一个具体实施例的产品的状态量化和剩余寿命预测的操作弹性曲线。
具体实施方式
以下结合附图，对本发明的具体实施方式进行说明。
本发明提供了一种状态量化和剩余寿命预测机制，其尤其适合于水泵。本发明基于数字双胞胎技术对水泵的设计性能和预测性能建模。其中，本发明结合水泵运行数据以及性能曲线来建立一个水泵的数据驱动的数字双胞胎模型来评估水泵在运行中的性能，以获取所述水泵在性能需求或者运行成本方面的剩余寿命预测。下面以水泵为例，对本发明进行说明。
本发明第一方面提供了产品状态量化和剩余寿命预测方法，其中，包括步骤T1、步骤T2和T3。
图5是根据本发明一个具体实施例的状态量化和剩余寿命预测系统的结构示意图。如图7所示，产品的状态量化和剩余寿命预测系统100包括性能分析装置110和剩余使用寿命预测装置120。其中，所述性能分析装置110的输入数据是水泵的设计数据D和运行数据O，输出数据是优化工作状态点P。具体地，所述性能分析装置110包括数据预处理装置111、设计建模装置112、数据融合装置113和优化装置114。所述性能分析装置110输出水泵的预测性 能曲线至所述剩余使用寿命预测装置120，剩余使用寿命预测装置120输出的是根据水泵的性能需求或者运行成本计算的剩余寿命预测。具体地，所述剩余使用寿命预测装置120包括性能衰减建模装置121、费用建模装置122、费用优化装置123和操作弹性建模装置124。
首先执行步骤T1，获取产品的设计数据和运行数据。在本实施例中，设计数据包括水泵的转速、流量、压力和功率等，运行数据也包括在不同磨损条件下的水泵的转速、流量、压力和功率。设计数据指的是设计水泵时想要达到的数据值，是理想化的。运行数据是水泵实际运行时产生的相应数据，是实况数据。
示例性地，如图2所示，在本实施例中，设计数据包括水泵的杨程，扬程离心泵的扬程又称为泵的压头，是指单位重量流体经泵所获得的能量。水泵的扬程大小取决于泵的结构，如叶轮直径的大小，叶片的弯曲情况等、转速。对泵的压头不能从理论上作出精确的计算，一般用实验方法测定。运行数据包括水泵的转速和流量，其中，时序数据按照转速做了分类。
然后执行步骤T2,性能分析装置110基于所述设计数据建立产品的设计模型，并根据所述设计数据和运行数据执行数据融合以获得产品的预测性能曲线，选取所述预测性能曲线中的优化工作状态点。
具体地，所述步骤T2还包括子步骤T21、子步骤T22、子步骤T23和子步骤T24。
在子步骤T21中，数据预处理装置111对所述设计数据和运行数据执行数据预处理，去除异常数据和错误数据。其中，数据预执行模块收集水泵的设计数据和执行数据，并且消除异常值。具体地，异常值包括利用算法去除毛刺，去除异常点等。
在子步骤T22中，设计建模装置112基于所述设计数据建立产品的设计模型，并且根据所述设计模型执行所述产品的设计性能评价和分析，其中，所述设计模型是多尺度数字双胞胎模型。其中，设计性能指的是水泵的出厂性能，是理想化的性能。
具体地，在本实施例中，多尺度数字双胞胎(multi-fidelity digital twin)是指一个包括水泵压头曲线、功率曲线和效率曲线的实况性能数字双胞胎
具体地，水泵性能数字双胞胎模型利用了多尺度数据驱动模型。针对水泵，其具有三个不同曲线来评估性能，包括压头曲线、功率曲线和效率曲线。 上述三种曲线通常应用于水泵产线设计阶段的水泵选择。然而，在后续系统运行过程中上述曲线不会持续使用。本发明结合水泵运行数据以及水泵压头曲线、功率曲线和效率曲线来建立一个水泵的数据驱动数字双胞胎模型来评估水泵在运行中的性能。其中，多尺度建模技术应用于上述建模过程。
图1是水泵的压头曲线、功率(power rate)曲线和效率曲线，可据此执行水泵的设计模型建模。其中，模型的不同尺度描述了水泵的动态系统。竖坐标分别表示压头、效率(％)和输入/输出(KW)，横坐标为水泵的每分钟流量，曲线S 3表示水泵性能，曲线S 4表示效率，曲线S 5表示电机输入(KW)。
在步骤T23中，数据融合装置113根据所述设计数据和运行数据利用数据融合算法执行数据融合以获得产品的预测性能曲线，并且通过优化数据融合算法的参数提高所述预测性能曲线的准确性。
图2是根据本发明一个具体实施例的状态量化和剩余寿命预测的基于水泵设计数据和运行数据产生性能曲线的示意图。如图2所示，在本实施例中，设计数据包括水泵的杨程，扬程离心泵的扬程又称为泵的压头，是指单位重量流体经泵所获得的能量。水泵的扬程大小取决于泵的结构，如叶轮直径的大小，叶片的弯曲情况等、转速。对泵的压头不能从理论上作出精确的计算，一般用实验方法测定。运行数据包括水泵的转速和流量，其中，时序数据按照转速做了分类。因此，基于上述设计数据和运行数据采用MFK算法(Multi-Fidelity Kriging)可以得到水泵的性能曲线。具体地，如图2的右图所示，根据上述杨程的设计数据和转速和流量的运行数据，按照MFK算法能够得到性能曲线，可知，水泵用过一段时间，实际数据是有衰减的，和刚出厂时预想的设计数据不一样。迭代执行上述过程会得到水泵的历史性能曲线和实时性能曲线。其中，为了使预测的性能曲线和实际的性能曲线一致，运行数据如果越靠近性能曲线就表示性能曲线的预测越准，因此可以通过优化数据融合算法的参数提高所述预测性能曲线的准确性。
下面以MFK算法为例，对本发明的数据融合步骤进行说明
如图1所示，
Figure PCTCN2021096502-appb-000005
代表一个分析或者数字的高尺度模型运算子，其不是实时的但是非常细节化、高尺度化并且基于自然法则，
Figure PCTCN2021096502-appb-000006
相应的索引为H。
Figure PCTCN2021096502-appb-000007
代表难度更低的低尺度模型运算子，其是实时的，并且描述了
Figure PCTCN2021096502-appb-000008
的行为，
Figure PCTCN2021096502-appb-000009
相应的索引为L。对于上述模型由此产生的方程也包括相应的模型误差
Figure PCTCN2021096502-appb-000010
Figure PCTCN2021096502-appb-000011
Figure PCTCN2021096502-appb-000012
将较低尺度模型作为例子，并应用图5所示的状态监视器方法执行建模中的数据融合。其中，
Figure PCTCN2021096502-appb-000013
是融合运算子。通过利用上个测量输入
Figure PCTCN2021096502-appb-000014
作为低尺度模型的输入，并将其融合到实际测试数据
Figure PCTCN2021096502-appb-000015
其输出能够用如下公式估算：
Figure PCTCN2021096502-appb-000016
Figure PCTCN2021096502-appb-000017
利用高斯过程回归(Gaussian Process Regression)来融合从低尺度和高尺度模型来的信息。f(x)～GP(m(x)，k(x，x′))能够用输入矢量x，一个平均函数m(x)和协方差函数k(x，x′)描述。
子元模型方法也叫做地质统计学领域的Co-Kriging。特别地，本发明考虑了四个不同Co-Kriging方法，其利用了x 1作为LFM输入矢量，x 2作为HFM输入矢量，f 1和f 2作为各自的函数值：
Figure PCTCN2021096502-appb-000018
上述方法不同之处在于GP的协方差矩阵和贺函数：自回归Co-Kriging(ARCK)，深度多尺度高斯函数(DMGP，Deep Multi-fidelity Gaussian Processes)，S简单Co-Kriging(SCK，Simple Co-Kriging)和简单多尺度高斯函数(SMGP，Simple Multi-fidelity Gaussian Processes)。以上被描述为：
Figure PCTCN2021096502-appb-000019
其中，k i.j是针对总的协方差矩阵 K的特定协方差函数，ARCK和DMGP利用：
k 11＝k 1(x 1，x 1)
k 12＝λk 1(x 1，x 2)
k 21＝λk 1(x 2，x 1)
k 22＝λ 2k 1(x 2，x 1)+k 2(x 2，x 2)
其中，λ是自回归因素，D是维度，k 1和k 2是以下方式的和函数：
Figure PCTCN2021096502-appb-000020
两个Co-Kriging方法在输入矢量的表现不同。DMGP利用了神经网络作为输入变形h(x)代替未处理的输入矢量x。与此相反，SCK和SMGP利用：
Figure PCTCN2021096502-appb-000021
k 12＝a 1b 1k 1(x 1，x 2)+a 2b 2k 2(x 1，x 2)
k 21＝a 1b 1k 1k 1(x 2，x 1)+a 2b 2k 2(x 2，x 1)
Figure PCTCN2021096502-appb-000022
由于协方差矩阵 K的a i、b i是超函数。再次，SMGP利用神经网络作为输入转换h(x)代替未处理的输入矢量x。此外，本发明利用连续多尺度元建模(SMF，sequential multi-fidelity meta-modeling)方法。基本上，SMF是一个优化算法，其用特定初始样本开始，并考虑不同样本计算成本的情况下添加LFM和HFM的样本。主要地，样本在习得的元建模的可靠区间最大处添加。最后，我们会得到最可能的元建模，其被最大计算预算约束。
在子步骤T24中，优化装置114计算产品的负荷及其功率和效率之间并同时满足所述产品的增长工作范围内的最佳匹配关系，以选取所述预测性能曲线中的优化工作状态点。具体地，通过当前性能曲线判断当前泵运行在哪个负荷效率最高功率最优，在曲线上找水泵最佳运行点，即在什么样的转速下水泵性价比最高。也就是，找出转速和输出功率和效率之间的最佳匹配关系，同时满足压力和流量在正常工作范围内。
最后执行步骤T3,剩余使用寿命预测装置120基于所述预测性能曲线建立产品的费用模型和操作弹性模型，根据产品的性能需求或者运行成本计算所述产品的剩余寿命预测。
具体地，所述步骤T3包括子步骤T31、子步骤T32、子步骤T33和子步骤T34。
在子步骤T31中，性能衰减建模装置121基于所述预测性能曲线，执行 一个回归计算来产生一个水泵的性能衰减模型。具体地，迭代执行步骤T2，可以基于预测性能曲线得到水泵的历史性能曲线以及当前的性能曲线，以及在水泵在不同磨损条件下运行的性能曲线。
其中，泵的压头衰减模型通过公式描述成时序的压头损失的模型如下：
P(t)＝P N-bt c
其中，P N表示初始性能，也就是t＝0时的设计点，即水泵刚出厂时的性能。b和c表示用于最小二乘法曲线拟合的参数。例如，b和c通过水泵的历史性能曲线中的四个点的最小二乘回归计算得到的回归参数。
因此，如3所示，通过不停地用历史数据来执行预测，可以得到泵的压头衰减曲线，其纵坐标是水泵的压头损失(％)，横坐标为时间(年)。如图3所示，随着水泵使用时间的增加，压头的损失值是逐渐大幅增加的。
在子步骤T32中，费用建模装置122基于所述性能衰减模型对产品的成本进行建模，以经计算所述产品的全生命周期经济成本和产品的替换成本。
其中，在本发明中，剩余使用寿命通过经济上和操作弹性两个方面计算。其中，水泵的生命周期成本(LCC，Life Cycle Cost)是整个水泵生命周期的花销，例如采购、安装、运行、维持和处理等。水泵的替换成本(RC，Replacement Cost)是由于水泵在考虑到不同的生命周期成本的情况下未来一个特定时间替换导致的额外成本。从经济角度考虑的剩余使用寿命能够用从现在到未来一个特定时间点的时间长度来定义，而在这个时间点水泵的替换成本最小。
如图4所示，从时间T 0到T end1表示旧水泵的第一个设计生命周期LC 11，从时间T end1到时间T end2表示旧水泵的第二个设计生命周期LC 12。其中，上述旧水泵的第一个设计生命周期LC 11和第二个设计生命周期LC 12表示的是水泵的出厂设计寿命，因此并未考虑到实际工况和磨损条件，是理想化的寿命而非真实寿命。如果把旧水泵替换成新水泵，那么从时间T 0到T rep1表示新水泵的第一个设计生命周期LC 21，从时间T rep1到时间T rep2表示新水泵的第二个设计生命周期LC 12。其中，上述新水泵的第一个设计生命周期LC 21和第二个设计生命周期LC 22表示的是新水泵的实际寿命，考虑到了实际工况和磨损条件，是真实寿命。
具体地，水泵的生命周期成本为：
LCC＝C ic+C in+C e+C o+C m+C 2+C env+C d
其中，C ic为初始成本，包括水泵、系统、管道、采购和附加服务的价格。 C in是安装成本。C e是能量成本，水泵驱动、控制和任何附加服务的系统运行预测成本。C o是系统管理等人力运行成本。C m是维持和维修费用，其中维修包括日常和预测性维修。C s是故障成本，例如生产损失。C env是环境成本，例如泵出液体和附件的污染物去除。C d是拆除和处理费用，包括修复本地环境和附加服务处理的费用。
因此该水泵的替换成本为：
Figure PCTCN2021096502-appb-000023
其中，LCC rep是替换新水泵的成本，LCC original是旧水泵的成本，其中，T end是旧水泵使用周期到期的时间，T rep是替换新水泵的时间，C e.rep是替换新水泵的能量成本，C res，rep是用新水泵的能量成本，C res，original是旧水泵的剩余价值，C res，rep是新水泵的剩余价值，因此随着时间流逝，旧水泵的能量成本C e，orginal是上升的，新水泵的能量成本C res，rep是降低的。其中，上述C ic、C in、C d是一次性花费。C o、C m、C s、C env是时间连续的固定花费。因此，不论是否更换水泵，上述成本C o、C m、C s、C env是一样的，因此可以忽略不计。
能量成本C e，orginal和C e.rep能够用能量功耗乘以电价来计算，
Figure PCTCN2021096502-appb-000024
Density(Q，N)是水泵流量Q的和转速N的联合概率密度函数，P E(t，Q，N)是在不同水泵流量Q和转速N下的由功率衰减模型更新的功率，Price E(t)是时间作为变量的预测电价。
基于上述算法来对水泵执行费用建模。
在子步骤T33中，费用优化装置123基于性能衰减模型和水泵的运行数据、成本参数，利用优化算法计算水泵的替换时间。即，泵替换时间Δt取多 少在经济上替换成本RC最小或经济收益最高，即使得替换成本RC最小。其中，所述优化算法包括智能优化算法、概率密度函数或黄金分割插值。
具体地，水泵的运行数据包括水泵的流量和转速的历史运行数据，水泵的传输效率，水泵的点击效率。成本参数包括：水泵的一次性成本初始成本C ic、安装成本C in、拆除和处理费用C d；水泵的剩余使用寿命T end1-T now，其中，T end1表示旧水泵第一个生命周期LC 11的种植时间，T now表示现在的时间；电价。
然后，把上述数据带入概率密度函数P t(Q，N，t)，通过历史数据分析得到水泵的功耗分布和流量、转速。
为了获得最佳经济效益，剩余使用寿命Δt能够用平均优化算法来最小化水泵更新成本来计算。图6的横坐标为时间，纵坐标为水泵的功率。曲线S 1是当前时刻预测的水泵性能衰减曲线，曲线S 2是假设在未来某个时间点要旧水泵替换为新水泵的性能衰减曲线，T end1表示旧水泵第一个生命周期LC 11的种植时间，T now表示现在的时间，T rep表示旧水泵替换为新水泵的时间。如图6所示，对于能量消耗部分，曲线S 1和S 2、T rep和T end1围成的面积S表示旧水泵替换成新水泵的经济利益，面积S最大化，这意味着会从能源节约获得更多利益，这是由于水泵性能由于更换得到了提高。因此，替换成本RC(Δt)需要通过以下函数最小化：
Min f(Δt)＝RC(Δt)
在子步骤T34中，操作弹性建模装置124基于水泵运行的实际要求来评估水泵的操作弹性，即在极端使用条件下的水泵的剩余使用寿命。其中，所述极端使用条件为满工蛮复杂情况下，如果满工满负载都达不到极端使用条件，则需要执行水泵替换。
其中所需数据包括水泵的流量Q d的最小需求，在最小流量下的最小压头需求H d。将上述数据带入水泵的流量衰减函数Q(H，N，t)和压头衰减函数H(Q，N，t)，能得到如图7所示的操作弹性曲线。如图7所示，图A的竖坐标表示水泵的流量，下图的竖坐标表示水泵的压头，横坐标为时间。其中曲线S 11表示水泵满负荷条件下的水泵流量，曲线S 12表示在特定压头情况下的水泵流量，虚线表示最小流量需求Q d。曲线S 21表示水泵满负荷条件下的水泵压头，曲线S 22表示在特定流量情况下的水泵压头，虚线表示最小压头需求H d。零边缘点P 1和P 2表示水泵的余量没有了。
具体地，水泵总是日常运行过程中在一个动态实质性运行中，来满足后阶段的负载需要量。因此，水泵需要一些可选的灵活性来响应压力和流量的上下变化。随着水泵的性能衰退，灵活性优势随着时间也会下降。由于从现在到未来一个特定时间点的时间段内水泵绩效利润从压头或者流动需求就在那时为0，灵活性方面的剩余使用寿命能够定义。我们定义两个弹性界限[Q d，H d]。Q d速度是流量的最小要求，H d是在最小流量Q d下的最小压头需求。以下公式给出了可选的水泵的压头和流量操作弹性：
f 1(Δt)＝|H(Δt)-H d|
f 2(Δt)＝|Q(Δt)-Q d|
其中，Q d是流量的最小需求，H d是在最小流量下的最小压头需求，Q(H，N，t)是流量衰减函数，H(Q，N，t)是压力衰减函数。
当所述水泵压头的操作弹性或者水泵流量的操作弹性小于一个预定与之时，则需要发送需要更换水泵提示。优选地，当f 1(Δt)或f 2(Δt )任一等于0时，则表示水泵需要替换。因此，当上述公式为0时，操作弹性剩余使用寿命能够计算。通过这两方面计算剩余使用寿命，最终用户能够选择轮转哪个来在未来做更多决定，例如关闭水泵或者对水泵执行替换。
本发明第二方面提供了产品状态量化和剩余寿命预测系统，其中，包括：处理器；以及与所述处理器耦合的存储器，所述存储器具有存储于其中的指令，所述指令在被处理器执行时使所述电子设备执行动作，所述动作包括：T1,获取产品的设计数据和运行数据；T2,基于所述设计数据建立产品的设计模型，并根据所述设计数据和运行数据执行数据融合以获得产品的预测性能曲线，选取所述预测性能曲线中的优化工作状态点；T3,基于所述预测性能曲线建立产品的费用模型和操作弹性模型，根据产品的性能需求或者运行成本计算所述产品的剩余寿命预测。
进一步地，所述产品为水泵，所述动作T2还包括如下步骤：T21，对所述水泵的设计数据和运行数据执行数据预处理，去除异常数据和错误数据；T22，基于所述设计数据建立水泵的设计模型，并且根据所述设计模型执行所述产品的设计性能评价和分析，其中，所述设计模型是多尺度数字双胞胎模型；T23，根据所述设计数据和运行数据利用数据融合算法执行数据融合以获得水泵的预测性能曲线，并且通过优化数据融合算法的参数提高所述预测性能曲线的准确性；T24，计算水泵的负荷及其功率和效率之间并同时满足所述水泵的增长工作范围内的优化匹配关系，以选取所述预测性能曲线中的优化 工作状态点。
进一步地，所述动作T3还包括如下步骤：T31，基于所述预测性能曲线，执行一个回归计算来产生一个水泵的性能衰减模型；T32，基于所述性能衰减模型对产品的成本进行建模，以计算所述产品的全生命周期经济成本和产品的替换成本；T33，基于性能衰减模型和水泵的运行数据、成本参数，利用优化算法计算水泵的替换时间；T34，基于水泵的满负荷运行参数和衰减函数计算水泵的操作弹性。
进一步地，所述数据融合算法为MFK算法。
进一步地，所述泵的性能衰减模型通过公式描述成时序的压头损失的模型如下：
P(t)＝P N-bt c
其中，P N表示初始性能，也就是t＝0时的设计点，即水泵刚出厂时的性能。b和c表示用于最小二乘法曲线拟合的参数。
进一步地，所述全生命周期经济成本为：
LCC＝C ic+C in+C e+C o+C m+C s+C env+C d
其中，C ic为初始成本，C in是安装成本，C e是能量成本，C o是系统管理成本，C m是维持和维修成本，C s是故障成本,C env是环境成本,C d是拆除和处理成本。
进一步地，所述替换成本为：
Figure PCTCN2021096502-appb-000025
其中，LCC rep是替换新水泵的全生命周期经济成本，LCC original是旧水泵的全生命周期经济成本，其中，T end是旧水泵使用周期到期的时间，T rep是替换新水泵的时间，C e.rep是替换新水泵的能量成本，C res，rep是用新水泵的能量成本，C res，original是旧水泵的剩余价值，C res，rep是新水泵的剩余价值，
其中，能量成本C e，orginal和C e.rep能够用能量功耗乘以电价来计算，
Figure PCTCN2021096502-appb-000026
其中，Q是水泵流量，N是水泵转速，Density(Q，N)是水泵流量和转速的联合概率密度函数，P E(t，Q，N)是在不同水泵流量和转速下的由功率衰减模型更新的功率，Price E(t)是时间作为变量的预测电价。
进一步地，所述满负荷运行参数包括水泵流量的最小要求Q d和最小压头需求H d，
其中，f 1(Δt )表示水泵压头的操作弹性，f 2(Δt )表示水泵流量的操作弹性：
f 1(Δt)＝|H(Δt)-H d|
f 2(Δt)＝|Q(Δt)-Q d|
当所述水泵压头的操作弹性f 1(Δt )或者水泵流量的操作弹性f 2(Δt )小于一个预定阈值时，则需要发送需要更换水泵提示。
本发明第三方面提供了产品状态量化和剩余寿命预测装置，其中，包括：获取装置，其获取产品的设计数据和运行数据；融合装置,其基于所述设计数据建立产品的设计模型，并根据所述设计数据和运行数据执行数据融合以获得产品的预测性能曲线，选取所述预测性能曲线中的优化工作状态点；计算装置,其基于所述预测性能曲线建立产品的费用模型和操作弹性模型，根据产品的性能需求或者运行成本计算所述产品的剩余寿命预测。
本发第四方面提供了计算机程序产品，所述计算机程序产品被有形地存储在计算机可读介质上并且包括计算机可执行指令，所述计算机可执行指令在被执行时使至少一个处理器执行根据本发明第一方面所述的方法。
本发明第五方面提供了计算机可读介质，其上存储有计算机可执行指令，所述计算机可执行指令在被执行时使至少一个处理器执行根据本发明第一方面所述的方法。
本发明的产品状态量化和剩余寿命预测机制可靠、精确，并且能够基于产品的动态性能执行量化和预测。本发明特别适用于水泵。本发明能够基于水泵的动态数字模型量化水泵的动态性能状态，并能够基于水泵衰减模型量化水泵剩余使用寿命。本发明能够提供给工艺自动化解决方案提供者产品，或者为水泵生产者提供价值增加服务包。
尽管本发明的内容已经通过上述优选实施例作了详细介绍，但应当认识到上述的描述不应被认为是对本发明的限制。在本领域技术人员阅读了上述内容后，对于本发明的多种修改和替代都将是显而易见的。因此，本发明的保护范围应由所附的权利要求来限定。此外，不应将权利要求中的任何附图标记视为限制所涉及的权利要求；“包括”一词不排除其它权利要求或说明书中 未列出的装置或步骤；“第一”、“第二”等词语仅用来表示名称，而并不表示任何特定的顺序。

Claims (19)

  1. 产品状态量化和剩余寿命预测方法，其中，包括如下步骤：
    T1,获取产品的设计数据和运行数据；
    T2,基于所述设计数据建立产品的设计模型，并根据所述设计数据和运行数据执行数据融合以获得产品的预测性能曲线，选取所述预测性能曲线中的优化工作状态点；
    T3,基于所述预测性能曲线建立产品的费用模型和操作弹性模型，根据产品的性能需求或者运行成本计算所述产品的剩余寿命预测。
  2. 根据权利要求1所述的产品状态量化和剩余寿命预测方法，其特征在于，所述产品为水泵，所述步骤T2还包括如下步骤：
    T21，对所述水泵的设计数据和运行数据执行数据预处理，去除异常数据和错误数据；
    T22，基于所述设计数据建立水泵的设计模型，并且根据所述设计模型执行所述产品的设计性能评价和分析，其中，所述设计模型是多尺度数字双胞胎模型；
    T23，根据所述设计数据和运行数据利用数据融合算法执行数据融合以获得水泵的预测性能曲线，并且通过优化数据融合算法的参数提高所述预测性能曲线的准确性；
    T24，计算水泵的负荷及其功率和效率之间并同时满足所述水泵的增长工作范围内的优化匹配关系，以选取所述预测性能曲线中的优化工作状态点。
  3. 根据权利要求1所述的产品状态量化和剩余寿命预测方法，其特征在于，所述步骤T3还包括如下步骤：
    T31，基于所述预测性能曲线，执行一个回归计算来产生一个水泵的性能衰减模型；
    T32，基于所述性能衰减模型对产品的成本进行建模，以计算所述产品的全生命周期经济成本和产品的替换成本；
    T33，基于性能衰减模型和水泵的运行数据、成本参数，利用优化算法计算水泵的替换时间；
    T34，基于水泵的满负荷运行参数和衰减函数计算水泵的操作弹性。
  4. 根据权利要求1所述的产品状态量化和剩余寿命预测方法，其特征在 于，所述数据融合算法为MFK算法。
  5. 根据权利要求3所述的产品状态量化和剩余寿命预测方法，其特征在于，所述泵的性能衰减模型通过公式描述成时序的压头损失的模型如下：
    P(t)＝P N-bt c
    其中，P N表示初始性能，也就是t＝0时的设计点，即水泵刚出厂时的性能。b和c表示用于最小二乘法曲线拟合的参数。
  6. 根据权利要求3所述的产品状态量化和剩余寿命预测方法，其特征在于，所述全生命周期经济成本为：
    LCC＝C ic+C in+C e+C o+C m+C s+C env+C d
    其中，C ic为初始成本，C in是安装成本，C e是能量成本，C o是系统管理成本，C m是维持和维修成本，C s是故障成本,C env是环境成本,C d是拆除和处理成本。
  7. 根据权利要求6所述的产品状态量化和剩余寿命预测方法，其特征在于，所述替换成本为：
    Figure PCTCN2021096502-appb-100001
    其中，LCC rep是替换新水泵的全生命周期经济成本，LCC original是旧水泵的全生命周期经济成本，其中，T end是旧水泵使用周期到期的时间，T rep是替换新水泵的时间，C e，rep是替换新水泵的能量成本，C res，rep是用新水泵的能量成本，C res，original是旧水泵的剩余价值，C res，rep是新水泵的剩余价值，
    其中，能量成本C e，orginal和C e，rep能够用能量功耗乘以电价来计算，
    Figure PCTCN2021096502-appb-100002
    其中，Q是水泵流量，N是水泵转速，Density(Q，N)是水泵流量和转速的联合概率密度函数，P E(t，Q，N)是在不同水泵流量和转速下的由功率衰减模型更新的功率，Price E(t)是时间作为变量的预测电价。
  8. 根据权利要求3所述的产品状态量化和剩余寿命预测方法，其特征在于，所述满负荷运行参数包括水泵流量的最小要求Q d和最小压头需求H d，
    其中，f 1(Δt)表示水泵压头的操作弹性，f 2(Δt)表示水泵流量的操作弹性：
    f 1(Δt)＝|H(Δt)-H d|
    f 2(Δt)＝|Q(Δt)-Q d|
    当所述水泵压头的操作弹性f 1(Δt)或者水泵流量的操作弹性f 2(Δt)小于一个预定阈值时，则需要发送需要更换水泵提示。
  9. 产品状态量化和剩余寿命预测系统，其中，包括：
    处理器；以及
    与所述处理器耦合的存储器，所述存储器具有存储于其中的指令，所述指令在被处理器执行时使所述电子设备执行动作，所述动作包括：
    T1,获取产品的设计数据和运行数据；
    T2,基于所述设计数据建立产品的设计模型，并根据所述设计数据和运行数据执行数据融合以获得产品的预测性能曲线，选取所述预测性能曲线中的优化工作状态点；
    T3,基于所述预测性能曲线建立产品的费用模型和操作弹性模型，根据产品的性能需求或者运行成本计算所述产品的剩余寿命预测。
  10. 根据权利要求9所述的产品状态量化和剩余寿命预测系统，其特征在于，所述产品为水泵，所述动作T2还包括如下步骤：
    T21，对所述水泵的设计数据和运行数据执行数据预处理，去除异常数据和错误数据；
    T22，基于所述设计数据建立水泵的设计模型，并且根据所述设计模型执行所述产品的设计性能评价和分析，其中，所述设计模型是多尺度数字双胞胎模型；
    T23，根据所述设计数据和运行数据利用数据融合算法执行数据融合以获得水泵的预测性能曲线，并且通过优化数据融合算法的参数提高所述预测性能曲线的准确性；
    T24，计算水泵的负荷及其功率和效率之间并同时满足所述水泵的增长工作范围内的优化匹配关系，以选取所述预测性能曲线中的优化工作状态点。
  11. 根据权利要求9所述的产品状态量化和剩余寿命预测系统，其特征在于，所述动作T3还包括如下步骤：
    T31，基于所述预测性能曲线，执行一个回归计算来产生一个水泵的性能衰减模型；
    T32，基于所述性能衰减模型对产品的成本进行建模，以计算所述产品的 全生命周期经济成本和产品的替换成本；
    T33，基于性能衰减模型和水泵的运行数据、成本参数，利用优化算法计算水泵的替换时间；
    T34，基于水泵的满负荷运行参数和衰减函数计算水泵的操作弹性。
  12. 根据权利要求9所述的产品状态量化和剩余寿命预测系统，其特征在于，所述数据融合算法为MFK算法。
  13. 根据权利要求11所述的产品状态量化和剩余寿命预测系统，其特征在于，所述泵的性能衰减模型通过公式描述成时序的压头损失的模型如下：
    P(t)＝P N-bt c
    其中，P N表示初始性能，也就是t＝0时的设计点，即水泵刚出厂时的性能。b和c表示用于最小二乘法曲线拟合的参数。
  14. 根据权利要求11所述的产品状态量化和剩余寿命预测系统，其特征在于，所述全生命周期经济成本为：
    LCC＝C ic+C in+C e+C o+C m+C s+C env+C d
    其中，C ic为初始成本，C in是安装成本，C e是能量成本，C o是系统管理成本，C m是维持和维修成本，C s是故障成本,C env是环境成本,C d是拆除和处理成本。
  15. 根据权利要求14所述的产品状态量化和剩余寿命预测系统，其特征在于，所述替换成本为：
    Figure PCTCN2021096502-appb-100003
    其中，LCC rep是替换新水泵的全生命周期经济成本，LCC original是旧水泵的全生命周期经济成本，其中，T end是旧水泵使用周期到期的时间，T rep是替换新水泵的时间，C e，rep是替换新水泵的能量成本，C res，rep是用新水泵的能量成本，C res，original是旧水泵的剩余价值，C res，rep是新水泵的剩余价值，
    其中，能量成本C e，orginal和C e，rep能够用能量功耗乘以电价来计算，
    Figure PCTCN2021096502-appb-100004
    其中，Q是水泵流量，N是水泵转速，Density(Q，N)是水泵流量和转速的联合概率密度函数，P E(t，Q，N)是在不同水泵流量和转速下的由功率衰减模型更新的功率，Price E(t)是时间作为变量的预测电价。
  16. 根据权利要求11所述的产品状态量化和剩余寿命预测系统，其特征在于，所述满负荷运行参数包括水泵流量的最小要求Q d和最小压头需求H d，
    其中，f 1(Δt)表示水泵压头的操作弹性，f 2(Δt)表示水泵流量的操作弹性：
    f 1(Δt)＝|H(Δt)-H d|
    f 2(Δt)＝|Q(Δt)-Q d|
    当所述水泵压头的操作弹性f 1(Δt)或者水泵流量的操作弹性f 2(Δt)小于一个预定阈值时，则需要发送需要更换水泵提示。
  17. 产品状态量化和剩余寿命预测装置，其中，包括：
    获取装置，其获取产品的设计数据和运行数据；
    融合装置,其基于所述设计数据建立产品的设计模型，并根据所述设计数据和运行数据执行数据融合以获得产品的预测性能曲线，选取所述预测性能曲线中的优化工作状态点；
    计算装置,其基于所述预测性能曲线建立产品的费用模型和操作弹性模型，根据产品的性能需求或者运行成本计算所述产品的剩余寿命预测。
  18. 计算机程序产品，所述计算机程序产品被有形地存储在计算机可读介质上并且包括计算机可执行指令，所述计算机可执行指令在被执行时使至少一个处理器执行根据权利要求1至8中任一项所述的方法。
  19. 计算机可读介质，其上存储有计算机可执行指令，所述计算机可执行指令在被执行时使至少一个处理器执行根据权利要求1至8中任一项所述的方法。
PCT/CN2021/096502 2021-05-27 2021-05-27 产品状态量化和剩余寿命预测方法、装置和系统 WO2022246758A1 (zh)

Priority Applications (5)

Application Number Priority Date Filing Date Title
KR1020237044776A KR20240011823A (ko) 2021-05-27 2021-05-27 제품 상태 정량화 및 잔여 수명 예측 방법, 및 장치 및 시스템
CN202180096844.9A CN117178273A (zh) 2021-05-27 2021-05-27 产品状态量化和剩余寿命预测方法、装置和系统
JP2023573001A JP2024520472A (ja) 2021-05-27 2021-05-27 製品状態定量化および残存寿命予測のための方法、装置ならびにシステム
EP21942334.0A EP4325390A1 (en) 2021-05-27 2021-05-27 Product state quantification and remaining life prediction method, and apparatus and system
PCT/CN2021/096502 WO2022246758A1 (zh) 2021-05-27 2021-05-27 产品状态量化和剩余寿命预测方法、装置和系统

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2021/096502 WO2022246758A1 (zh) 2021-05-27 2021-05-27 产品状态量化和剩余寿命预测方法、装置和系统

Publications (1)

Publication Number Publication Date
WO2022246758A1 true WO2022246758A1 (zh) 2022-12-01

Family

ID=84229431

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/096502 WO2022246758A1 (zh) 2021-05-27 2021-05-27 产品状态量化和剩余寿命预测方法、装置和系统

Country Status (5)

Country Link
EP (1) EP4325390A1 (zh)
JP (1) JP2024520472A (zh)
KR (1) KR20240011823A (zh)
CN (1) CN117178273A (zh)
WO (1) WO2022246758A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117332993A (zh) * 2023-11-29 2024-01-02 深圳市北辰德科技股份有限公司 基于物联网的金融机具控制管理方法及系统

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20050005186A1 (en) * 2003-06-23 2005-01-06 General Electric Company Method, system and computer product for estimating a remaining equipment life
US20120143564A1 (en) * 2010-12-01 2012-06-07 Xerox Corporation System and method for predicting remaining useful life of device components
CN106089753A (zh) * 2016-07-01 2016-11-09 太原理工大学 一种离心泵剩余寿命预测方法
CN106777577A (zh) * 2016-11-30 2017-05-31 中国西电电气股份有限公司 一种基于运行数据预测高压开关产品剩余寿命的方法

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20050005186A1 (en) * 2003-06-23 2005-01-06 General Electric Company Method, system and computer product for estimating a remaining equipment life
US20120143564A1 (en) * 2010-12-01 2012-06-07 Xerox Corporation System and method for predicting remaining useful life of device components
CN106089753A (zh) * 2016-07-01 2016-11-09 太原理工大学 一种离心泵剩余寿命预测方法
CN106777577A (zh) * 2016-11-30 2017-05-31 中国西电电气股份有限公司 一种基于运行数据预测高压开关产品剩余寿命的方法

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117332993A (zh) * 2023-11-29 2024-01-02 深圳市北辰德科技股份有限公司 基于物联网的金融机具控制管理方法及系统
CN117332993B (zh) * 2023-11-29 2024-03-22 深圳市北辰德科技股份有限公司 基于物联网的金融机具控制管理方法及系统

Also Published As

Publication number Publication date
CN117178273A (zh) 2023-12-05
EP4325390A1 (en) 2024-02-21
KR20240011823A (ko) 2024-01-26
JP2024520472A (ja) 2024-05-24

Similar Documents

Publication Publication Date Title
US11119451B2 (en) Apparatus, method, program, and recording medium
JP6784745B2 (ja) リアルタイムデータ駆動型の電力測定およびコスト推定システム
Behrooz et al. Dynamic optimization of natural gas networks under customer demand uncertainties
Sanusi et al. Simultaneous monitoring of magnitude and time-between-events data with a Max-EWMA control chart
JPWO2020179849A1 (ja) 計画装置、制御装置、方法、およびプログラム
JP6427357B2 (ja) 診断支援システム及び診断支援方法
US20120316852A1 (en) Arrangement and method for optimizing the operation of a supply network
CN102486833A (zh) 装置的效能预测及故障检测的方法
JP6773012B2 (ja) 操業改善効果算出装置、操業改善効果算出方法、操業改善効果算出プログラム、および記録媒体
WO2022246758A1 (zh) 产品状态量化和剩余寿命预测方法、装置和系统
JP2022177561A (ja) 製造業における循環型・低炭素型生産システムの構築方法
Petrucci et al. From real to virtual sensors, an artificial intelligence approach for the industrial phase of end-of-line quality control of GDI pumps
Johnson et al. Data analytics and pump control in a wastewater treatment plant
US12008497B2 (en) Demand sensing and forecasting
JP6615892B2 (ja) 物理システムの経時変化プロファイリングエンジン
Yoo et al. A clustering-based equipment condition model of chemical vapor deposition process
JP2020177571A (ja) 異常予兆診断装置、異常予兆診断方法及び異常予兆診断プログラム
JP6880864B2 (ja) エネルギー管理システムおよびエネルギー管理方法
US20200332773A1 (en) A prognostics and health management model for predicting wind turbine oil filter wear level
Mathumitha et al. Intelligent deep learning techniques for energy consumption forecasting in smart buildings: a review
Fernandez-Anakabe et al. An attribute oriented induction based methodology for data driven predictive maintenance
Chen et al. Research on prediction algorithm of ship equipment heath condition
Chen Integration of building information modeling and internet of things for facility maintenance management
Yildirim et al. Evaluating LMP forecasting with LSTM networks: a deep learning approach to analyzing electricity prices during unpredictable events
Groenewald et al. An Autoregressive Fault Model for Condition Monitoring of Electrical Machines in Deep-level Mines

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21942334

Country of ref document: EP

Kind code of ref document: A1

WWE Wipo information: entry into national phase

Ref document number: 2021942334

Country of ref document: EP

ENP Entry into the national phase

Ref document number: 2021942334

Country of ref document: EP

Effective date: 20231113

WWE Wipo information: entry into national phase

Ref document number: 2023573001

Country of ref document: JP

ENP Entry into the national phase

Ref document number: 20237044776

Country of ref document: KR

Kind code of ref document: A

WWE Wipo information: entry into national phase

Ref document number: 1020237044776

Country of ref document: KR

NENP Non-entry into the national phase

Ref country code: DE