Digital intelligent laboratory intelligent operation and maintenance system based on Internet of Things and AI double engine

Abstract

The invention relates to the technical field of maintenance operation management, in particular to a digital intelligent laboratory intelligent operation and maintenance system based on Internet of Things and AI double engine, and the system comprises an equipment state sensing module which is used for collecting the real-time power reading and rated power parameters of an intelligent power meter connected to a refrigerator and a centrifugal machine, and calculating the energy consumption wear performance ratio value. According to the method and the device, the energy efficiency attenuation and the hidden wear of the equipment in the incomplete shutdown state are identified, and the energy waste and the sudden failure risk in the sub-health state are avoided. By traversing a connection flow direction relationship between a sample pretreatment unit and a downstream analysis instrument, counting and generating an equipment association retardation accumulation effect value, and placing an isolated equipment state in a topological network of an overall experimental process for weighted operation, the chain retardation influence of a single point of fault on upstream and downstream sample circulation can be quantified, and the accuracy of the system is improved. And the maintenance resource is ensured to be preferentially inclined to the key node which has the greatest influence on the global flux.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of maintenance operation management, and in particular to a digital laboratory intelligent operation and maintenance system based on Internet of Things and AI. BACKGROUND

[0002] Maintenance operation management technology is to monitor, maintain, repair and optimize the whole life cycle of physical assets, production equipment and infrastructure, to ensure the continuity, safety and efficiency of core business processes.

[0003] The existing maintenance operation management technology often relies on fixed time periods for preventive maintenance or passive repair after the device has substantial functional failure in actual application. This management mode is difficult to capture early signs of gradual degradation of device performance, resulting in long-term operation of the device in a state of high energy consumption and low efficiency, not only causing continuous energy waste, but also increasing the risk of damage to experimental samples due to environmental fluctuations. In the face of complex laboratory equipment network, the existing technology usually regards each device as an independent individual for state evaluation, lacks overall analysis of the upstream and downstream dependence relationship between devices and sample flow direction, and is easy to over-maintain non-critical path devices and ignore key bottleneck nodes, resulting in local faults causing large-scale experimental process congestion and sample accumulation. Therefore, improvement is needed. SUMMARY

[0004] The purpose of the present application is to solve the shortcomings in the prior art, and to provide a digital laboratory intelligent operation and maintenance system based on Internet of Things and AI.

[0005] In order to achieve the above-mentioned purpose, the present application adopts the following technical scheme: the digital laboratory intelligent operation and maintenance system based on Internet of Things and AI comprises: A device state perception module is used to collect real-time power readings and rated power parameters of intelligent power meters connected to the refrigerator and centrifuge, and to calculate energy consumption wear efficiency ratio values. An influence propagation analysis module is used to traverse the connection flow relationship between the sample pretreatment unit and the downstream analytical instrument according to the energy consumption wear efficiency ratio value, to statistically generate a device associated blocking accumulation effect value, to operate the energy consumption wear efficiency ratio value and the device associated blocking accumulation effect value, and to generate a laboratory global maintenance influence factor. A maintenance value evaluation module is used to divide the laboratory global maintenance influence factor by the spare parts delivery time length obtained from the external supply chain system to generate a spare part response weighted operation urgency, to calculate the difference between the expected electricity saving amount and the required maintenance material cost amount after maintenance, and to calculate the energy efficiency optimization maintenance investment return value in combination with the spare part response weighted operation urgency. The response decision scheduling module is configured to arrange the energy efficiency optimization maintenance return on investment values in descending order, match corresponding laboratory equipment codes, generate a laboratory intelligent operation and maintenance response sequence list, extract a device item ranked first in the laboratory intelligent operation and maintenance response sequence list, call a corresponding emergency backup machine scheduling scheme, and generate a hierarchical maintenance execution work order.

[0006] Preferably, the energy consumption and wear efficiency ratio value obtaining step is: According to the real-time power readings and rated power parameters of the refrigerator and the centrifuge, the power signal missing points and abnormal value points are removed, the difference ratio of the real-time power readings and the rated power parameters is calculated, and the power difference ratio sequence is arranged in chronological order; According to the power difference ratio sequence, the vibration speed effective value RMS sequence in the same time period is extracted from the Internet of Things vibration sensor with the time stamp as the matching index, the sensor interference section and signal drift section are screened out, and the power difference ratio and vibration speed effective value pairing sequence is generated; According to the power difference ratio and vibration speed effective value pairing sequence, the energy consumption and wear efficiency ratio value is calculated.

[0007] Preferably, the device associated blocking accumulation effect value obtaining step is: The target device whose energy consumption and wear efficiency ratio value exceeds the shutdown threshold is screened, the connection flow direction of the sample pre-processing unit and the downstream analytical instrument is traversed, the connection flow direction relationship topology matrix and the shutdown target device unit indicator vector are generated according to the device port mapping, and the connection flow direction relationship topology matrix and the shutdown target device unit indicator vector are formed. According to the connection flow direction relationship topology matrix and the shutdown target device unit indicator vector, the device associated blocking accumulation effect value is calculated.

[0008] Preferably, the laboratory global maintenance influence factor obtaining step is: The energy consumption and wear efficiency ratio value and the device associated blocking accumulation effect value calculated for each target device are respectively normalized in the set of all candidate devices, and the normalized values are synthesized to generate a laboratory global maintenance influence factor.

[0009] Preferably, the spare part response weighted operation and maintenance urgency obtaining step is: According to the laboratory global maintenance influence factor, the spare part arrival time field of the external supply chain system is called, the device code and the uniform time reference are aligned, and the laboratory global maintenance influence factor and the spare part arrival time pairing list are generated. According to the pairing list of the laboratory global maintenance influence factor and the spare part delivery time length, the spare part delivery time length record with zero is removed, the spare part delivery time length decimal is standardized, and the single value is calculated by multiplying the laboratory global maintenance influence factor and the spare part delivery time length for each device, and the device code is labeled, and the spare part response weighted operation and maintenance urgency is generated.

[0010] Preferably, the energy efficiency optimization maintenance investment return value obtaining step is: According to the spare part response weighted operation and maintenance urgency, the predicted electricity saving amount after maintenance and the required maintenance material cost amount are aggregated, the difference between the predicted electricity saving amount after maintenance and the required maintenance material cost amount is calculated according to the device code, and the basic investment return rate is calculated by dividing the difference by the required maintenance material cost amount, and the basic investment return rate is multiplied by the corresponding spare part response weighted operation and maintenance urgency to generate the energy efficiency optimization maintenance investment return value.

[0011] Preferably, the laboratory intelligent operation and maintenance response list obtaining step is: The energy efficiency optimization maintenance investment return value record and the laboratory device code catalog are internally connected according to the device code field, the energy efficiency optimization maintenance investment return value missing record is cleaned up, the energy efficiency optimization maintenance investment return value is arranged in descending order, and the laboratory device code is sorted in lexicographic order for the parallel items, and the laboratory intelligent operation and maintenance response list is generated.

[0012] Preferably, the hierarchical maintenance execution work order obtaining step is: According to the laboratory intelligent operation and maintenance response list, the first device item is extracted and the corresponding laboratory device code is read, the emergency backup machine dispatching scheme index table is called to locate the record according to the laboratory device code, and the scheme state field is checked to be available, the emergency backup machine type parameter and the switching step list are merged to generate the emergency backup machine dispatching scheme. According to the emergency backup machine dispatching scheme, the work order unit data is filled according to the laboratory device code, and the work order priority is set to the first level of the laboratory intelligent operation and maintenance response list, the switching operation time sequence and the maintenance inspection time sequence are prepared, and the responsibility team list is listed, the switching step list of the emergency backup machine dispatching scheme is referenced to form the hierarchical maintenance execution work order.

[0013] Compared with the prior art, the advantages and positive effects of the present application are: In the present application, the real-time power readings of the intelligent power meter connected to the refrigerator and centrifuge are collected and compared with the rated power parameters, and the energy consumption wear efficiency ratio value is calculated, which can identify the energy efficiency decay and hidden wear of the equipment in the incomplete shutdown state, avoid energy waste and sudden failure risk in the sub-health state. By traversing the connection flow direction relationship between the sample pretreatment unit and the downstream analytical instrument, the device associated blocking accumulation effect value is counted and generated, the isolated device state is placed in the topology network of the whole experimental process for weighted operation, which can quantify the chain blocking influence of single point failure on upstream and downstream sample flow, and ensure that the maintenance resources are preferentially tilted to the key nodes with the greatest impact on global flux. The spare parts response weighted operation urgency is generated by introducing the spare parts arrival time data of the external supply chain system, and the energy efficiency optimization maintenance investment return value is calculated by combining the difference between the expected electricity saving amount after maintenance and the maintenance material cost, which realizes the multi-dimensional integration of technical urgency, supply chain feasibility and economic benefit, and ensures that the generated maintenance strategy meets the logistics reality. Further, the emergency standby machine scheduling scheme is called and the hierarchical maintenance execution work order is generated, which can establish a response mechanism between new and old equipment switching and fault disposal, and reduce the experimental interruption time caused by equipment downtime. BRIEF DESCRIPTION OF DRAWINGS

[0014] Figure 1 The system flowchart of the present application. DETAILED DESCRIPTION

[0015] In order to make the purpose, technical scheme and advantages of the present application clearer, the present application will be further described in detail below in combination with the drawings and examples. It should be understood that the specific examples described herein are only used to explain the present application and do not limit the present application.

[0016] Please refer to Figure 1 The present application provides a technical scheme: a digital laboratory intelligent operation and maintenance system based on Internet of Things and AI dual engine, which comprises: A device state sensing module is used to collect the real-time power readings of the intelligent power meter connected to the refrigerator and centrifuge and the rated power parameters, and calculate the energy consumption wear efficiency ratio value; An influence propagation analysis module is used to traverse the connection flow direction relationship between the sample pretreatment unit and the downstream analytical instrument according to the energy consumption wear efficiency ratio value, count and generate the device associated blocking accumulation effect value, and calculate the energy consumption wear efficiency ratio value and the device associated blocking accumulation effect value to generate the laboratory global maintenance influence factor; A maintenance value evaluation module is used to divide the laboratory global maintenance influence factor by the spare parts arrival time obtained from the external supply chain system to generate the spare parts response weighted operation urgency, calculate the difference between the expected electricity saving amount after maintenance and the required maintenance material cost amount, and calculate the energy efficiency optimization maintenance investment return value in combination with the spare parts response weighted operation urgency. The response decision scheduling module is used to sort the energy efficiency optimization and maintenance investment return values ​​in descending order, match the corresponding laboratory equipment codes, generate a laboratory intelligent operation and maintenance response sequence list, extract the first and first equipment item in the laboratory intelligent operation and maintenance response sequence list, retrieve the corresponding emergency standby machine scheduling plan, and generate a hierarchical maintenance execution work order.

[0017] The steps to obtain the energy consumption and wear efficiency ratio are as follows: Based on the real-time power readings and rated power parameters of the smart power meters of refrigerators and centrifuges, they are synchronized and paired according to the device number and timestamp, power signal missing points and outlier points are eliminated, the difference ratio between the real-time power reading and the rated power parameter is calculated and arranged in chronological order to generate a power difference ratio sequence. Based on the power difference ratio sequence, the RMS sequence of vibration velocity effective values ​​within the same time period is extracted from the IoT vibration sensor using the timestamp as the matching index. Sensor interference segments and signal drift segments are filtered out to generate a power difference ratio and vibration velocity effective value pairing sequence. Based on the pairing sequence of power difference ratio and effective vibration velocity value, the energy consumption wear efficiency ratio is calculated using the following formula: ; in, This is the energy consumption and wear efficiency ratio value. For the first The power difference ratio at each time point is the ratio of the difference between the real-time power reading and the rated power parameter to the rated power parameter. For the first The effective value of vibration velocity at each time point represents the intensity of vibration energy. This represents the peak value of the maximum vibration velocity observed throughout the entire time series. This represents the total number of samples in the power difference ratio and vibration velocity effective value paired sequence.

[0018] Specifically, according to the intelligent power meter connected with the refrigerator and the centrifuge, real-time power readings are obtained through RS485 or wireless communication protocol to set the sampling frequency (for example, once every second), at the same time, the rated power parameters of the equipment are obtained by reading the equipment nameplate or calling the equipment management database, the real-time power readings and the rated power parameters are corresponded according to the MAC address or asset number unique to the equipment, and are uniformly aligned to the millisecond timestamp of the network time protocol server, for the obtained original power data stream, data cleaning operation is performed, and null (NaN) caused by communication packet loss and abnormal peak values caused by circuit transient surge are identified and removed, the determination standard of abnormal values is set as: when the single-point power reading exceeds 3 times the standard deviation of the average value of the adjacent five points, or the power reading is negative, it is determined as an abnormal point, for the missing position after removal, the theoretical power value at this time is fitted by using three normal data points before and after the three-point spline interpolation method to fill in, so as to maintain the continuity of the time sequence, then the difference ratio calculation logic is constructed, and the cleaned real-time power readings are extracted point by point and the rated power parameters , the difference between the two is calculated and divided by the rated power parameter, the calculation expression is , the calculation result reflects the deviation degree of the equipment under the current load, if the result is positive, it indicates overload operation, if it is negative, it indicates low load operation, finally, all the calculated time point difference ratios are indexed and arranged in order according to the timestamp from small to large, forming an ordered data set containing time dimension, that is, a power difference ratio sequence is generated.

[0019] According to the power difference ratio sequence generated by the foregoing steps, the timestamp list contained therein is read as a primary key index, a data request is initiated to a piezoelectric vibration sensor installed at the compressor or motor housing of the equipment through the Internet of Things gateway, vibration acceleration raw waveform data in the time period is called, time domain integral processing is performed on the raw waveform data to convert it into vibration speed data, and the root mean square value (RMS) is calculated according to a set time window (for example, a 1-second window synchronized with the power sampling frequency), thereby extracting a vibration speed effective value sequence representing the vibration energy intensity. Before data pairing, quality control needs to be performed on the vibration data to filter out sensor interference segments and signal drift segments. The identification basis for the interference segment is that the kurtosis index of the vibration signal is detected, and if the kurtosis value in a certain time window is greater than a preset impact threshold (for example, an empirical value of 5.0, which is obtained by artificially applying a mechanical impact of a known intensity on the equipment in a stationary state and recording the response data), and the duration of the high-frequency signal is less than 0.1 second, it is determined to be external impact interference and is removed. The identification basis for the signal drift segment is that the trend item is extracted by low-pass filtering the signal, and if the monotonic change amplitude of the trend item in 1 minute exceeds 20% of the baseline (the average vibration value when the equipment is running at no load), it is determined to be sensor temperature drift or zero drift, and needs to be corrected by removing the trend item using polynomial fitting. After cleaning and correction, the timestamp of the power difference ratio sequence is used as the reference to search for the closest vibration speed effective value RMS data point within the allowed time deviation range (for example, ±50 milliseconds), and the two are corresponded one by one. If a certain time point cannot match the corresponding vibration data, the power data at that time point is discarded to ensure strict synchronization of the two sets of data in the time dimension. Finally, the aligned data rows are combined to generate a power difference ratio and vibration speed effective value pairing sequence.

[0020] In the energy consumption wear efficiency ratio numerical calculation formula, the wear efficiency of the equipment under non-ideal working conditions is quantified by combining the deviation degree of power fluctuation and the energy intensity of mechanical vibration. The first half of the formula uses a weighted structure similar to a correlation coefficient to evaluate the coupling degree of power anomalies (whether overload or low load, taking the absolute value) and vibration intensity, that is, whether high-intensity vibration is accompanied when the power deviation is large. The second half of the formula calculates the crest factor, that is, the ratio of the vibration peak value to the effective value, which is used to capture the impact characteristics in the vibration signal.

[0021] For the total sample number of the power difference ratio and vibration speed effective value pairing sequence, the parameter is obtained by counting the total number of data rows in the pairing sequence generated by the foregoing steps, which is a dimensionless integer value. In actual application, a complete working cycle or a fixed time period (such as 10 minutes) of the equipment is usually selected as the observation window. If the sampling frequency is 1 Hz, The value of the parameter is 600, which is used to balance the influence of sample size in the calculation of mean value and weighting process, to ensure that the calculation result will not be deviated by orders of magnitude due to the length of the observation period, and is the basic dimension of statistical calculation.

[0022] The power difference ratio of the time point represents the difference between the real-time power reading and the rated power parameter and the ratio of the rated power parameter, which is a dimensionless floating point number, calculated by the formula , where is the real-time power collected at the time point (unit: watt), is the rated power of the device (unit: watt), for example, the rated power of a refrigerator is 500W, and the real-time power collected at the time is 550W, then , which means that the power deviates from the rated value by 10%, and the absolute value is taken to participate in the operation in order to pay attention to the deviation amplitude rather than the direction, which directly quantifies the abnormal degree of electric energy input.

[0023] The vibration velocity effective value of the time point represents the intensity of vibration energy, with the unit of millimeters per second (mm / s), which is obtained by integrating the acceleration signal collected by the vibration sensor and calculating the root mean square (RMS), and the specific acquisition steps are as follows: collect discrete vibration velocity data points in the time window, and the calculation formula is , which reflects the average vibration intensity of the device at time .

[0024] The maximum vibration velocity peak value observed in the entire time series has the unit of millimeters per second (mm / s), which is the maximum instantaneous absolute value of the vibration velocity signal reached in the current observation period (i.e. samples), and the acquisition method is to traverse the vibration waveform data in the entire period and extract the maximum absolute value, which reflects the strongest mechanical impact or transient oscillation suffered by the device during operation.

[0025] According to the parameter, the calculation is as follows: There are 5 sample points in the observation period, i.e. .

[0026] After collection and calculation, the power difference ratio absolute value sequence dimensionless) is: [0.1, 0.2, 0.05, 0.3, 0.15].

[0027] Corresponding vibration velocity effective value sequence (unit: mm / s) is: [2.0, 4.0, 1.0, 8.0, 3.0].

[0028] Vibration peak value detected during the entire observation period .

[0029] First step, calculate the weighted sum part in the molecule : ; ; Multiply by the number of samples : ; Second step, calculate each term in the denominator: (dimensionless); ; Denominator product: ; Third step, calculate the left half of the formula (coupling coefficient): ; Fourth step, calculate the right half of the formula (peak factor term): First calculate the root mean square value of (the denominator part): The sequence is: [4.0, 16.0, 1.0, 64.0, 9.0]; ; Mean: ; Square root: ; Calculate the right half ratio: ; Fifth step, calculate the final energy consumption wear efficiency ratio value : ; The results show that the calculated energy consumption wear efficiency ratio value 3.746, which value comprehensively reflects the severity of mechanical wear of the equipment under unit energy consumption deviation. The higher the value, the higher the vibration energy accompanied by abnormal power fluctuation of the equipment, and the vibration form has significant impact (high peak factor). If the health threshold is set to 2.5 (the threshold is determined by statistical regression analysis of historical failure data), the current result 3.746 has significantly exceeded the threshold, indicating that there may be serious mechanical looseness or bearing pitting inside the equipment, causing additional physical damage while consuming additional electrical energy, and it is recommended that the system immediately trigger a red maintenance alarm.

[0030] The acquisition step of the equipment-associated blocking accumulation effect value is: Screening target equipment with energy consumption wear efficiency ratio value exceeding shutdown threshold, traversing the connection flow direction of sample pretreatment unit and downstream analytical instrument, generating connection flow direction relationship topology matrix and shutdown target equipment unit indicator vector according to equipment port mapping, forming connection flow direction relationship topology matrix and shutdown target equipment unit indicator vector; According to the connection flow direction relationship topology matrix and the shutdown target equipment unit indicator vector, the equipment-associated blocking accumulation effect value is calculated, and the calculation formula is: ; Wherein, is the equipment-associated blocking accumulation effect value, is the node importance weight vector, wherein the first element represents the importance of the first equipment, which is calculated according to the equipment output flux, operation time and maintenance priority, is the influence propagation attenuation factor, which is used to represent the proportional attenuation of influence intensity for each order of path extension, is the connection flow direction relationship topology matrix, and the element in the matrix represents the direct connection relationship of the first equipment to the first equipment, is the transpose of the matrix , which is used to represent the reverse connection, is the upper bound of path length, which is defined as the maximum order number reachable in the topology matrix, is the path order index, represents the first order power of the connection flow direction relationship topology matrix , reflecting the indirect influence degree of the target equipment signal on the downstream nodes after passing through level path, is the unit indicator vector of the first equipment, which represents the position of the target equipment in the matrix.

[0031] Specifically, according to the energy consumption wear performance ratio value obtained by the foregoing calculation, a shutdown threshold based on historical statistics is set, and the setting process of the threshold is to retrieve the peak value data of the energy consumption performance ratio of all similar equipment in the laboratory in the past three years before the equipment is shut down due to failure, calculate the 95th percentile of the distribution as a baseline, for example, the baseline value is calculated as 4.2, and the equipment exceeding the value is determined as a high-risk target equipment, and then a full network topology scanning program is started to traverse the physical connection line and logical data flow between the sample pretreatment unit (including the opening machine, centrifuge, etc.) and the downstream analytical instrument (including the biochemical analyzer, immune analyzer, etc.), the device port mapping table in the LIMS system is analyzed by using the depth-first search algorithm (DFS), the “output-input” relationship between each pair of devices is identified, and a directed graph of laboratory equipment is constructed, on the basis of which a connection flow relationship topology matrix is generated, the matrix is an N-order square matrix (N is the total number of equipment), if the equipment A has a sample flow to the equipment B, the corresponding position of the matrix is marked as 1, otherwise 0, at the same time, for each identified target equipment, a shutdown target equipment unit indicator vector is generated, the length of the vector is N, only the index position corresponding to the target equipment is 1, and the rest is 0, which is used for positioning the fault source point in the subsequent matrix operation, and finally the connection flow relationship topology matrix and the shutdown target equipment unit indicator vector are collected and output.

[0032] In the device correlation blockage accumulation effect value calculation formula, the bidirectional propagation process of the simulated fault in the complex network is simulated, the forward logistics blockage (i.e. no sample available downstream) is captured, the reverse logistics accumulation (i.e. sample cannot be sent out upstream) is captured, the comprehensive blockage pressure caused by the shutdown of the target equipment to other high-value nodes in the whole network is quantified by weighted summation, and the local fault risk is mapped to the global maintenance urgency.

[0033] The node importance weight vector has a dimension of , and the th element represents the comprehensive business value of the th equipment. In order to ensure dimensional consistency, the parameter must be strictly normalized, and the steps are as follows: collect three original data: output flux (sample / hour), historical utilization rate (%), and maintenance priority (1-5 integer), respectively, normalize the maximum and minimum of the three data in the whole laboratory equipment range, and obtain and , and finally synthesize them by using a weighted formula: Here, the weights 0.5, 0.3 and 0.2 are set according to the “efficiency first” operation strategy of the laboratory, and is a dimensionless value between 0 and 1.

[0034] is a dimensionless coefficient between 0 and 1, used to simulate the physical law that the impact of a failure decreases as the network distance increases. The setting of this parameter is based on the following method: simulate a single-point shutdown in the simulation environment, and observe the queue length growth rate of the first-order adjacent nodes and the second-order adjacent nodes. If the first-order growth rate is 100% and the second-order growth rate is 60%, set , usually taking a value of 0.5 to 0.7.

[0035] is a binary matrix of the connection flow relationship topology matrix, , element indicates the existence of direct sample flow from device to device , which directly maps the physical connection relationship and is dimensionless.

[0036] is the transpose of matrix , element , which mathematically represents reverse connection and physically is used to calculate the “reverse back pressure” or “accumulation effect” of the failure on the upstream node.

[0037] is the upper bound of the path length, an integer, representing the maximum number of levels for calculating the impact, which is usually set according to the average cascade length of the laboratory pipeline. For example, if the longest pipeline contains 5 devices, then is set to 3 or 4, ignoring very remote and weak impacts.

[0038] is the path order index, a summation variable, from 1 to , respectively calculating the 1st-order direct impact, 2nd-order indirect impact, etc.

[0039] indicates the power of matrix , element indicates the number of paths with a length of from to , which is used here to quantify the impact pathways of multiple downstream orders.

[0040] is the unit indicator vector of the th device, with a dimension of , only the th element is 1, which is used to extract the column vector related to the target device in matrix operations.

[0041] Calculation according to parameters: Build 3-node network: Device 1 Device 2 Device 3.

[0042] Target device is Device 2 (middle node, ).

[0043] Parameter preparation: After normalizing the device flux of the whole network, for example, Device 1 (sample addition) , Device 2 (centrifugation) , Device 3 (analysis) , then .

[0044] Topology matrix (1 connected to 2, 2 connected to 3).

[0045] Transposed matrix .

[0046] Let the attenuation factor be , the upper bound of the path .

[0047] Target vector .

[0048] Matrix summation operation ( ): .

[0049] Extract the target column: .

[0050] The first element in the result vector is 1, indicating that the shutdown of Device 2 directly blocks the upstream Device 1 (due to the effect of ); the third element is 1, indicating that the shutdown of Device 2 directly cuts off the downstream Device 3 (due to the effect of ).

[0051] Weighted evaluation: .

[0052] .

[0053] The calculated device-related blocking accumulation effect value is 1.3, which clearly indicates that Device 2 as the central node, its failure will simultaneously affect the upstream Device 1 (importance 0.6) and the downstream Device 3 (importance 0.7), causing a total network cumulative loss of 1.3 units of importance. If compared to another edge device (such as Device 0 connected only to Device 1, weight 0.5), its The value can be only 0.5 (only affecting downstream device 1), so the value 1.3 reveals the key position of device 2 in the topology, and the system should list it as a priority maintenance object to prevent single-point failure from spreading to a full-line failure.

[0054] The acquisition step of the laboratory global maintenance influence factor is: The energy consumption wear efficiency ratio value calculated for each target device and the device associated with the blocking accumulation effect value are normalized by minimum maximum value in the set of all candidate devices, and the normalized values are synthesized to generate the laboratory global maintenance influence factor.

[0055] Specifically, the energy consumption wear efficiency ratio value (denoted as ) and the device associated with the blocking accumulation effect value (denoted as ) are calculated according to the foregoing steps, respectively, before performing global synthesis, the difference in value distribution between the two needs to be eliminated, all candidate devices in the current maintenance period are traversed, and the energy consumption efficiency ratio set and the blocking effect value set are constructed, respectively, the maximum value and the minimum value are extracted from them, the minimum maximum value normalization formula is applied to map the value and the value of each device to the dimensionless interval of , ensuring that the two indicators are comparable on the same order of magnitude, then define the synthesis weight, for example, set the weight of the device health dimension to 0.4 and the weight of the network influence dimension to 0.6, linearly weight the sum of the normalized two indicators according to the weight, that is, , and finally generate the laboratory global maintenance influence factor of each device as the basis for sorting maintenance work orders.

[0056] The acquisition step of the spare part response weighted operation urgency is: According to the laboratory global maintenance influence factor, call the spare part delivery time field of the external supply chain system, align the two fields according to the device code and the unified time reference, and generate a list of laboratory global maintenance influence factors and spare part delivery time pairs; According to the laboratory global maintenance influence factor and the spare part delivery time pairing list, remove the spare part delivery time field with a value of zero and standardize the spare part delivery time field, calculate a single value by multiplying the laboratory global maintenance influence factor by the spare part delivery time for each device, and label the device code, and generate the spare part response weighted operation urgency.

[0057] Specifically, according to the laboratory global maintenance influence factor, an application programming interface (API) connection with an external supply chain management system (SCM) is established, a query request containing the current device model and spare parts list is sent, and the spare parts inventory status and the expected arrival time field (in hours) contained in the feedback data packet are obtained. The device code in the laboratory equipment management database is used as the primary key, and the logistics information of the corresponding spare parts is parsed from the SCM feedback data to ensure that each maintenance influence factor record can find the corresponding supply chain response data. If a device involves multiple spare parts, the longest arrival time is selected as the final arrival time parameter of the device to reflect the actual situation of the maintenance activity being subject to the short board effect. Then, according to the unified network time protocol (NTP) timestamp, the generation time of the maintenance influence factor and the acquisition time of the supply chain data are calibrated to ensure that they reflect the same time state of the operation and maintenance state. The outdated data with a time deviation of more than 1 hour is removed, and finally the cleaned two sets of data are merged row by row according to the device code to generate a laboratory global maintenance influence factor and spare parts arrival time pairing list.

[0058] According to the laboratory global maintenance influence factor and the spare parts arrival time pairing list, each record in the list is traversed, the spare parts arrival time field is checked, and abnormal records with a value of zero (usually representing data missing or system error) are directly removed. For valid records, considering that the spare parts arrival time is usually in hours or days and may have decimals (such as 1.5 days), it is uniformly converted to hours and kept to one decimal place, for example, 1.5 days is converted to 36.0 hours. Then the spare parts response weighted operation logic is defined, and the corresponding laboratory global maintenance influence factor (denoted as ) and the standardized spare parts arrival time (denoted as ) of each device are extracted to calculate the product of the two as the operation and maintenance urgency index of the device, and the calculation formula is , where is the spare parts response weighted operation and maintenance urgency, and this calculation process reflects the coupling relationship between risk and response time: even if the device failure influence factor is high, if the spare parts can be delivered immediately (T is small), the overall urgency is relatively controllable; otherwise, if the high-influence-factor device is faced with a long-period of waiting for materials (T is large), the urgency is multiplied, and finally the value of T is marked in the corresponding device record to generate the spare parts response weighted operation and maintenance urgency.

[0059] The steps for obtaining the energy efficiency optimization maintenance return on investment value are: Based on the weighted maintenance urgency of spare parts response, the estimated electricity cost savings after maintenance and the cost of required repair materials are aggregated. The difference between the estimated electricity cost savings and the cost of required repair materials is calculated according to the equipment code. The basic return on investment is calculated by dividing the difference by the cost of required repair materials. The basic return on investment is then multiplied by the corresponding spare parts response weighted maintenance urgency to generate the energy efficiency optimized maintenance return on investment value.

[0060] Specifically, based on the weighted maintenance urgency of spare parts response, the financial management module is queried to obtain the expected energy efficiency improvement data for each piece of equipment after maintenance and repair. Combined with the local industrial electricity price (e.g., 0.8 yuan / kWh), the estimated electricity cost savings due to energy efficiency restoration within a future maintenance cycle (e.g., 365 days) are calculated (denoted as...). Simultaneously, retrieve the Bill of Materials (BOM) for repair, add up the purchase price of all replacement parts and labor service costs, and calculate the required cost of repair materials (denoted as...). Perform the difference calculation according to the device code: The difference This reflects the net financial return, followed by the calculation of the underlying rate of return on investment. This ratio quantifies the energy efficiency benefit multiple generated by a unit maintenance investment. To incorporate the urgency of maintenance into decision-making, the basic return on investment is further weighted with the spare parts response calculated above to determine the maintenance urgency. To perform a product operation, that is... ,in The system calculates the final energy efficiency optimization maintenance return on investment. The higher the value, the more economical the maintenance of the equipment will be, and the more urgent the maintenance bottleneck will be. Based on this, the system prioritizes all equipment to be maintained and generates the energy efficiency optimization maintenance return on investment.

[0061] The steps for obtaining the laboratory intelligent operation and maintenance response sequence list are as follows: The records of energy efficiency optimization and maintenance investment return values ​​are internally linked with the laboratory equipment code directory based on the equipment code field. Missing records of energy efficiency optimization and maintenance investment return values ​​are cleaned up. The records are sorted in descending order by energy efficiency optimization and maintenance investment return values ​​and the parallel items are sorted in lexicographical order by laboratory equipment code. A laboratory intelligent operation and maintenance response sequence table is generated.

[0062] Specifically, an inner join is performed between the records of energy efficiency optimization and maintenance investment return values ​​and the laboratory equipment code directory based on the equipment code field. Using a hash join strategy in a relational database engine, the "equipment code" is used as the hash key to construct an in-memory hash table for efficient matching of the two datasets. Each joined record is traversed, and a rigorous data cleaning operation is performed on the "energy efficiency optimization and maintenance investment return value" field, removing all records marked as NULL, NaN (not a number), or negative infinity values ​​resulting from calculation errors, ensuring that all data participating in the sorting has actual physical meaning. Subsequently, a multi-level sorting logic is initiated, first setting "energy efficiency optimization and maintenance investment return value" as the primary sorting key and employing the Quick Sort algorithm. The `Sort` function sorts the entire dataset in descending order, placing the equipment with the highest return on investment (ROI) at the top of the list, reflecting a "efficiency-first" operation and maintenance strategy. To address the possibility of identical values ​​in the sorting process, the "laboratory equipment code" is introduced as a secondary sorting key, and the equipment is sorted in ascending order according to the ASCII code table. For example, when two devices both have an ROI of 12.5, the device with the code "L-01" will be placed before "L-02". This deterministic sorting rule eliminates the influence of randomness on the operation and maintenance order. Each item in the list is assigned a unique integer index that increments from 1. The final output is a structured dataset containing the equipment code, ROI value, and sorting index, generating a laboratory intelligent operation and maintenance response sequence list.

[0063] The steps for obtaining hierarchical maintenance work orders are as follows: Based on the laboratory intelligent operation and maintenance response sequence list, extract the first and most listed equipment item and read the corresponding laboratory equipment code. Call the emergency standby machine scheduling scheme index table to locate the record according to the laboratory equipment code and verify that the scheme status field is available. Merge the emergency standby machine model parameters and the switching step list to generate an emergency standby machine scheduling scheme. Based on the emergency standby machine scheduling plan, the work unit data is filled in according to the laboratory equipment code and the work order priority is set as the first level of the laboratory intelligent operation and maintenance response sequence list. The switching operation sequence and maintenance inspection sequence are compiled and the list of responsible teams is listed. The switching step list of the emergency standby machine scheduling plan is used to form a hierarchical maintenance execution work order.

[0064] Specifically, according to the laboratory intelligent operation and maintenance response sequence table, the first row record in the list is directly accessed through the index pointer, the "laboratory equipment code" corresponding to the record is extracted as the core search key, the emergency backup machine scheduling scheme index table stored in the configuration database is called, the index table is a key-value pair structure, which maps the relationship between the main device and the backup resource, and the binary search method is used to quickly locate the corresponding configuration record in the index table, the "scheme state" field in the record is read, and a string comparison is performed with the pre-defined enumeration value "Status_Ready" (indicating ready for use). If the state value does not match (for example, it shows "Status_Maintenance" or "Status_Occupied"), automatically jump to query the sub-optimal backup machine or trigger the out-of-stock alarm. After confirming that the state check is passed, the static attribute data of the backup machine is read from the record, including the device model, rated power, interface type and storage location coordinates. At the same time, according to the "operation process ID" field in the record, the corresponding structured text data is called from the standard operation procedure (SOP) library, including detailed shutdown locking steps, pipeline disconnection steps, backup machine access steps and initialization debugging parameters. These scattered attribute data and process text are serialized and merged to construct a JSON format data package containing complete scheduling instructions, and an emergency backup machine scheduling scheme is generated.

[0065] According to the emergency backup machine scheduling scheme, a new maintenance work order object is instantiated, the JSON content in the scheme data package is parsed, the code of the main device, the model of the backup machine and the location coordinates of the fault occurrence are automatically mapped and filled into the "metadata" part of the work order, the "emergency level" attribute of the work order is directly assigned as the highest level defined by the system (for example, "P0 level - immediate response") according to the first ranking of the device in the response sequence table. Then enter the time planning stage, read the pre-set standard working hour parameters in the scheme (for example, switching time 30 minutes, self-checking time 15 minutes), take the work order generation time as the reference point (T0), derive the expected completion time of each key node through time addition operation, compile the switching operation time sequence and maintenance inspection time sequence including "T0+30min complete physical switching" and "T0+45min complete function verification", then access the scheduling management system through the API interface, match the skill labels required by the work order (such as "centrifuge maintenance", "high voltage electrician") with the skill certificates of the current on-duty personnel, screen out the list of qualified and currently idle technical personnel, add them to the "responsibility team" field of the work order, and finally embed the standard operation step (SOP) text referred in the scheme into the execution description area of the work order for on-site reference by maintenance personnel, forming a hierarchical maintenance execution work order.

Claims

1. A digital and intelligent laboratory operation and maintenance system based on the dual engines of IoT and AI, characterized in that: The system includes: The equipment status sensing module is used to collect real-time power readings and rated power parameters of the smart power meters connected to refrigerators and centrifuges, and to calculate the energy consumption and wear efficiency ratio. The impact propagation analysis module is used to traverse the connection flow relationship between the sample pretreatment unit and the downstream analytical instrument based on the energy consumption and wear efficiency ratio value, statistically generate the equipment-related blockage and accumulation effect value, and calculate the energy consumption and wear efficiency ratio value and the equipment-related blockage and accumulation effect value to generate the laboratory global maintenance impact factor. The maintenance value assessment module is used to divide the laboratory's overall maintenance impact factor by the spare parts delivery time obtained from the external supply chain system, generate a spare parts response weighted maintenance urgency, calculate the difference between the expected electricity cost savings after maintenance and the cost of the required repair materials, and calculate the energy efficiency optimized maintenance investment return value in combination with the spare parts response weighted maintenance urgency. The response decision scheduling module is used to sort the energy efficiency optimization maintenance investment return values ​​in descending order, match the corresponding laboratory equipment codes, generate a laboratory intelligent operation and maintenance response sequence table, extract the first-ranked equipment item in the laboratory intelligent operation and maintenance response sequence table, retrieve the corresponding emergency standby machine scheduling plan, and generate a hierarchical maintenance execution work order.

2. The intelligent laboratory operation and maintenance system based on IoT and AI dual engines as described in claim 1, characterized in that, The steps for obtaining the energy consumption and wear efficiency ratio are as follows: Based on the real-time power readings and rated power parameters of the smart power meters of refrigerators and centrifuges, they are synchronized and paired according to the device number and timestamp, power signal missing points and outlier points are eliminated, the difference ratio between the real-time power reading and the rated power parameter is calculated and arranged in chronological order to generate a power difference ratio sequence. Based on the power difference ratio sequence, the RMS sequence of vibration velocity values ​​within the same time period is extracted from the IoT vibration sensor using the timestamp as the matching index. Sensor interference segments and signal drift segments are filtered out to generate a power difference ratio and vibration velocity RMS pairing sequence. The energy consumption and wear efficiency ratio is calculated based on the pairing sequence of the power difference ratio and the effective value of vibration velocity.

3. The intelligent laboratory operation and maintenance system based on IoT and AI dual engines as described in claim 1, characterized in that, The steps for obtaining the device-related hindrance and accumulation effect value are as follows: Target equipment whose energy consumption and wear efficiency ratio exceeds the shutdown threshold is screened, the connection flow between the sample preprocessing unit and the downstream analytical instrument is traversed, and a connection flow relationship topology matrix and a shutdown target equipment unit indicator vector are generated based on the equipment port mapping to form a connection flow relationship topology matrix and a shutdown target equipment unit indicator vector. Based on the connection flow relationship topology matrix and the shutdown target equipment unit indication vector, calculate the equipment-related obstruction and accumulation effect value.

4. The intelligent laboratory operation and maintenance system based on IoT and AI dual engines as described in claim 1, characterized in that, The steps for obtaining the global maintenance impact factor of the laboratory are as follows: The energy consumption and wear efficiency ratio values ​​calculated for each target device and the associated resistance and accumulation effect values ​​of the device are normalized to the minimum and maximum values ​​within the set of all candidate devices. The normalized values ​​are then synthesized to generate the laboratory global maintenance impact factor.

5. The intelligent laboratory operation and maintenance system based on IoT and AI dual engines as described in claim 1, characterized in that, The steps for obtaining the weighted maintenance urgency of the spare parts response are as follows: Based on the laboratory global maintenance impact factor, the spare parts arrival time field of the external supply chain system is called, and the two fields are aligned according to the equipment code and the unified time benchmark to generate a pairing list of laboratory global maintenance impact factor and spare parts arrival time. Based on the pairing list of laboratory global maintenance impact factor and spare parts delivery time, records with spare parts delivery time of zero are removed and the decimal places of spare parts delivery time are standardized. For each device, the laboratory global maintenance impact factor is multiplied by the spare parts delivery time to calculate a single value and the device code is marked to generate a spare parts response weighted maintenance urgency.

6. The intelligent laboratory operation and maintenance system based on IoT and AI dual engines as described in claim 1, characterized in that, The steps for obtaining the return on investment for energy efficiency optimization and maintenance are as follows: Based on the weighted maintenance urgency of the spare parts response, the estimated electricity cost savings after maintenance and the cost of required repair materials are aggregated. The difference between the estimated electricity cost savings and the cost of required repair materials is calculated according to the equipment code. The basic return on investment is then calculated by dividing the difference by the cost of required repair materials. The basic return on investment is then multiplied by the corresponding weighted maintenance urgency of the spare parts response to generate an energy-efficient optimized maintenance return on investment value.

7. The intelligent laboratory operation and maintenance system based on IoT and AI dual engines as described in claim 1, characterized in that, The steps for obtaining the laboratory intelligent operation and maintenance response sequence list are as follows: The records of energy efficiency optimization and maintenance investment return values ​​are internally linked with the laboratory equipment code directory based on the equipment code field. Missing records of energy efficiency optimization and maintenance investment return values ​​are cleaned up. The records are sorted in descending order by energy efficiency optimization and maintenance investment return values, and the parallel items are sorted in lexicographical order by laboratory equipment code to generate a laboratory intelligent operation and maintenance response sequence table.

8. The intelligent laboratory operation and maintenance system based on IoT and AI dual engines as described in claim 1, characterized in that, The steps for obtaining the hierarchical maintenance execution work order are as follows: Based on the laboratory intelligent operation and maintenance response sequence table, extract the first and second equipment item and read the corresponding laboratory equipment code. Call the emergency standby machine scheduling scheme index table to locate the record according to the laboratory equipment code and verify that the scheme status field is available. Merge the emergency standby machine model parameters and the switching step list to generate an emergency standby machine scheduling scheme. According to the emergency standby machine scheduling scheme, the work unit data is filled in according to the laboratory equipment code and the work order priority is set as the first level of the laboratory intelligent operation and maintenance response sequence list. The switching operation sequence and maintenance inspection sequence are compiled and the list of responsible teams is listed. The switching step list of the emergency standby machine scheduling scheme is used to form a hierarchical maintenance execution work order.

Images