Search

CN-122000009-A - Medical equipment fault prediction and full life cycle benefit optimization method, system, electronic equipment and storage medium

CN122000009ACN 122000009 ACN122000009 ACN 122000009ACN-122000009-A

Abstract

The invention provides a medical equipment fault prediction and full life cycle benefit optimization method, a system, electronic equipment and a storage medium. Belongs to the technical field of medical equipment management. The method comprises the steps of collecting multidimensional data such as equipment running state, environment and full life cycle management, preprocessing the data, extracting key features including environment temperature and humidity cooperative coefficients, constructing a multi-algorithm cooperative model by adopting an LSTM (least squares) and random forest and gradient lifting tree, conducting fault prediction, dynamically correcting model parameters based on prediction deviation, generating a personalized maintenance scheduling scheme based on fault risk level and greedy algorithm, constructing a full life cycle benefit assessment model, generating equipment running control instructions according to assessment results, and feeding back to an equipment end. According to the invention, through multi-source data fusion and multi-algorithm collaborative analysis, the problems of low prediction precision and stiff maintenance mode in the prior art are solved, closed-loop management from benefit evaluation to technical regulation is realized, and operation and maintenance cost is reduced.

Inventors

  • DAI YUEYUE

Assignees

  • 江苏康医通科技有限公司

Dates

Publication Date
20260508
Application Date
20251230

Claims (10)

  1. 1. A method for predicting faults and optimizing full life cycle benefits of medical equipment, the method comprising: Step S1, multi-dimensional data acquisition and integration, namely acquiring equipment running state data of medical equipment in real time through a data acquisition interface, acquiring environment data of the environment where the equipment is located through an environment sensor, and acquiring full life cycle management data from a hospital management system; step S2, data preprocessing and feature engineering, namely cleaning and standardizing acquired data, and constructing a fault feature set by combining field knowledge and statistical analysis, wherein the fault feature set at least comprises an environmental temperature and humidity synergistic coefficient; Step 3, multi-algorithm collaborative fault prediction, namely inputting the fault feature set into a pre-trained multi-algorithm collaborative prediction model to output a fault prediction result, wherein the multi-algorithm collaborative prediction model carries out weighted fusion based on the results of a plurality of basic prediction models and has a dynamic correction mechanism based on prediction deviation; Step S4, personalized maintenance scheduling optimization, namely generating a personalized maintenance scheduling scheme by combining diagnosis and treatment plan constraint and operation and maintenance resource states according to the fault risk level determined by the fault prediction result; and S5, full life cycle benefit dynamic evaluation and optimization, namely constructing a multi-dimensional benefit evaluation model based on full flow data, calculating benefit indexes, generating equipment operation control instructions according to the benefit evaluation results and feeding back the equipment operation control instructions to the medical equipment, so as to realize technical closed-loop regulation.
  2. 2. The method according to claim 1, wherein in step S2, the calculation formula of the environmental temperature and humidity synergy coefficient is: Wherein C is an environmental temperature and humidity synergistic coefficient, T is a current environmental temperature, T 0 is an optimal operation environmental temperature of the equipment, H is a current environmental relative humidity, H 0 is an optimal operation environmental relative humidity of the equipment, alpha and beta are weight coefficients calibrated for different types of medical equipment, and exp () is an exponential function.
  3. 3. The method according to claim 1, wherein in step S3, the multi-algorithm collaborative prediction model includes an LSTM timing prediction model, a random forest fault classification model, and a gradient lift tree risk assessment model, the weighted fusion employs dynamic weight adjustment, and a weight update formula is: Wherein W i,t is the weight of the ith basic model in the t period; A cci,t−1 is the actual prediction precision of the ith basic model in the previous period, accbase is a preset precision threshold value, and the updated weight is subjected to normalization treatment; adding a physical constraint term in a loss function of the LSTM model: Wherein Q pred is the residual heat capacity predicted by the model, Q rated is the rated heat capacity, P (t) is the real-time thermal power, eta is the heat dissipation efficiency, and the maintenance time is arranged according to the predicted residual heat capacity.
  4. 4. A method according to claim 3, wherein the dynamic correction mechanism comprises: calculating a deviation value of the predicted fault frequency and the actual fault frequency; When the deviation value exceeds a preset threshold value, positioning a basic model to be adjusted according to a deviation source; Adjusting the number of hidden layer neurons or the Dropout probability if the deviation is from an LSTM model, adjusting the number of decision trees or the depth of the trees if the deviation is from a random forest model, and adjusting the learning rate or the number of trees if the deviation is from a gradient lifting tree model; and adjusting the step length to be the preset proportion of the current parameter value each time by adopting a progressive strategy until the verification deviation meets the requirement.
  5. 5. The method according to claim 1, wherein in step S4, the generating a personalized maintenance scheduling scheme uses a greedy algorithm, with a minimum comprehensive operation and maintenance cost as an objective function: Wherein Z is comprehensive operation and maintenance cost, closs is equipment shutdown loss, clabor is maintenance labor cost, cpart is spare part allocation cost, and k 1 ,k 2 ,k 3 is a weight coefficient; The constraint conditions at least comprise that maintenance operation time is not overlapped with a core diagnosis and treatment period, the skill matching degree of operation and maintenance personnel meets the requirement, and the key spare part stock is available; a physical-data dual drive model is constructed for critical components of the medical device having thermal damage accumulation characteristics.
  6. 6. The method of claim 1, further comprising the step of adapting the threshold based on an adaptive threshold of environmental awareness: setting a dynamic threshold V (H, T, N) of fault alarm: Wherein V base is a reference threshold, N is the equipment daily average load, N 0 is the standard daily average load, ka, kb, kc are influence coefficients, and when the environmental parameters deviate from the optimal values or the load increases, the alarm threshold is automatically tightened.
  7. 7. The method according to claim 1, wherein in step S5, the generating device operation control instructions includes: When the full life cycle benefit evaluation result meets the life cycle end judgment condition, generating a degradation operation instruction, and locking the maximum operation power or limiting function module of the equipment through the equipment control interface; And when the part replacement cost ratio exceeds a threshold value, automatically adjusting the characteristic weight of the fault prediction model and the parameters of the maintenance period.
  8. 8. A medical device fault prediction and full life cycle benefit optimization system for implementing the method of any of claims 1-7, comprising: The data acquisition module is used for acquiring equipment running state data, environment data and full life cycle management data; the data preprocessing module is used for cleaning, standardizing and carrying out characteristic engineering processing on the data; The fault prediction module is used for carrying out fault prediction based on the multi-algorithm cooperative model; the maintenance scheduling module is used for generating a personalized maintenance scheduling scheme; The full life cycle benefit evaluation module is used for evaluating the benefit index and generating a control instruction; And the terminal display module is used for displaying the evaluation result and the control instruction state.
  9. 9. An electronic device comprising a memory storing a computer program and a processor implementing the steps of a medical device failure prediction and full life cycle benefit optimization method according to any of claims 1 to 7 when the computer program is executed by the processor.
  10. 10. A computer readable storage medium, characterized in that it has stored thereon a computer program which, when executed by a processor, implements the steps of a medical device failure prediction and full life cycle benefit optimization method according to any of claims 1 to 7.

Description

Medical equipment fault prediction and full life cycle benefit optimization method, system, electronic equipment and storage medium Technical Field The invention relates to the technical field of medical equipment management and digital operation and maintenance, in particular to a method, a system, electronic equipment and a storage medium for predicting faults of medical equipment and optimizing full life cycle benefits. Background In the digital transformation process of the medical industry, high-end medical equipment such as a CT machine, a magnetic resonance apparatus (MRI), an ultrasonic diagnostic apparatus and the like are core supports for clinical diagnosis and treatment. Currently, the main stream management schemes of medical equipment are mainly divided into two categories, namely regular preventive maintenance, namely checking and replacing parts according to fixed time intervals, and passive response based on basic data monitoring, namely triggering an alarm when parameters such as voltage, current and the like are out of limit. However, the prior art has significant disadvantages: 1. The 'one-cut' maintenance mode is rigidified, namely the actual load difference of different equipment cannot be adapted to the regular maintenance, the high-frequency equipment is easy to fail in the maintenance period, and the low-frequency equipment is subjected to excessive maintenance, so that the cost is increased. 2. The early warning mechanism is simple, the false alarm rate is high, the judgment is only carried out by relying on a single parameter threshold, the coupling relation among parameters and the environmental influence are not considered, and the instantaneous fluctuation and the fault precursor are difficult to distinguish. 3. The whole life cycle is lacked, the existing scheme focuses on maintenance links, purchase, operation and maintenance, diagnosis and treatment benefits and scrapping whole-flow data are not integrated, so that the equipment management decision is lacked scientific basis, and the phenomenon of 'heavy purchase and light management' is common. 4. The single algorithm has poor adaptability, the operation data of the medical equipment has nonlinear and non-stable characteristics, the single threshold judgment or the traditional machine learning model prediction precision is insufficient, and the hidden faults are difficult to identify in advance. Disclosure of Invention The invention provides a method, a system, electronic equipment and a storage medium for predicting faults of medical equipment and optimizing full life cycle benefits, which are used for solving the problems of low precision of fault prediction, rigidification of a maintenance mode and loss of full life cycle benefit management and control in the prior art. The invention discloses a medical equipment fault prediction and full life cycle benefit optimization method, which comprises the following steps: Step S1, multi-dimensional data acquisition and integration, namely acquiring equipment running state data of medical equipment in real time through a data acquisition interface, acquiring environment data of the environment where the equipment is located through an environment sensor, and acquiring full life cycle management data from a hospital management system; step S2, data preprocessing and feature engineering, namely cleaning and standardizing acquired data, and constructing a fault feature set by combining field knowledge and statistical analysis, wherein the fault feature set at least comprises an environmental temperature and humidity synergistic coefficient; Step 3, multi-algorithm collaborative fault prediction, namely inputting the fault feature set into a pre-trained multi-algorithm collaborative prediction model to output a fault prediction result, wherein the multi-algorithm collaborative prediction model carries out weighted fusion based on the results of a plurality of basic prediction models and has a dynamic correction mechanism based on prediction deviation; Step S4, personalized maintenance scheduling optimization, namely generating a personalized maintenance scheduling scheme by combining diagnosis and treatment plan constraint and operation and maintenance resource states according to the fault risk level determined by the fault prediction result; and S5, full life cycle benefit dynamic evaluation and optimization, namely constructing a multi-dimensional benefit evaluation model based on full flow data, calculating benefit indexes, generating equipment operation control instructions according to the benefit evaluation results and feeding back the equipment operation control instructions to the medical equipment, so as to realize technical closed-loop regulation. Preferably, in step S2, the calculation formula of the environmental temperature and humidity synergistic coefficient is: Wherein C is an environmental temperature and humidity synergistic coefficient, T is a current environmenta