Laboratory safety performance level dynamic evaluation management method based on management efficiency factor

Abstract

The invention discloses a laboratory safety performance level dynamic evaluation management method based on a management efficiency factor, and belongs to the technical field of laboratory safety management. The method comprises the following steps: firstly, establishing a theoretical risk value TR of a laboratory through weighted calculation so as to quantify an inherent risk level; secondly, dynamically collecting potential safety hazard data by deploying monitoring equipment, and calculating an actual risk value AR by combining factors such as rectification efficiency and repeated occurrence; the core lies in that the performance of a manager and the inherent risk difficulty of a laboratory are decoupled through a formula management efficiency coefficient MEF = TR / AR, and fair evaluation is realized; and finally, according to the MEF value, automatically judging a performance level A / B / C, and triggering differentiated management measures and incentive strategies. According to the method, the problem of incentive distortion caused by traditional static and score deduction assessment is solved, transformation from a punishment problem to efficient reward management is achieved, and fairness and initiative of laboratory safety management are remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of computer-aided management and laboratory safety, and particularly relates to a laboratory safety performance fair grading method based on dynamic assessment of management effectiveness. Specifically, it combines performance analysis in administrative management, office automation, and safety perception technology realized through video monitoring, and is particularly suitable for laboratory management of educational institutions. BACKGROUND

[0002] Currently, laboratory safety management generally adopts an assessment mode based on static inspection and deduction system, which typically embodies the core features of the prior art: static inspection, result orientation, and one-size-fits-all.

[0003] Static and periodic inspection: the assessment relies on periodic or non-periodic on-site inspection, which simplifies the complex and dynamic laboratory safety state into a "snapshot" score at the moment of inspection. Safety management requirements and on-site implementation requirements are fixed as checklists, and the safety state during the inspection interval is unknown and uncontrollable.

[0004] Deduction system and result orientation: this method is completely based on "non-conformance" deduction. The evaluation focus is completely on the result of "how many problems are found", while ignoring the prevention efforts of management personnel before the problem occurs and the rectification efficiency after the problem occurs.

[0005] The above prior art, although it establishes an assessment system in form, has serious defects in its internal logic, leading to management failure and distorted incentives:

[0006] Serious unfairness, dampening enthusiasm: the system completely ignores the inherent risk differences of different laboratories. A laboratory involving high-risk chemicals and large power equipment (high theoretical risk) naturally has more potential hidden points than a computer room for theoretical calculation (low theoretical risk). Under the deduction system, the management personnel of the high-risk laboratory, even if they double their efforts, are more likely to have higher absolute deduction scores than the low-risk laboratory, leading to their long-term performance lag and seriously dampening their enthusiasm for active management.

[0007] Static lag, lack of dynamic perception: periodic inspection cannot capture the dynamically changing risks in the daily operation of the laboratory. Safety management is reduced to "meeting the inspection", and as soon as the inspection is over, the safety state may quickly deteriorate, failing to achieve real process supervision.

[0008] One-sided evaluation, ignoring management effectiveness: the existing method only cares about "whether the problem exists", but does not evaluate "whether the management behavior is effective", and cannot evaluate and motivate efficient management behavior. SUMMARY

[0009] The present application aims to overcome the above-mentioned defects of the prior art, and provides a laboratory safety performance level dynamic evaluation management method based on management efficiency factors, so as to solve the problems of unfairness, static lag and neglect of management efficiency of the traditional examination mode, and establish a fair dynamic evaluation mechanism which can effectively encourage management personnel to take the initiative.

[0010] To achieve the above-mentioned purpose, the present application adopts the following technical solutions:

[0011] Step S1: Laboratory theoretical risk value TR calculation

[0012] S101: Establishing a laboratory inherent risk factor library, including but not limited to:

[0013] Hazardous chemical factors: type, maximum storage capacity, toxicity, flammability and explosive properties.

[0014] Instrument and equipment factors: number and power level of high-pressure, high-temperature, high-speed, radiation, laser and other equipment.

[0015] Experimental process factors: reaction types involved, operation complexity. The reaction types include: addition reaction, elimination reaction, substitution reaction (including nucleophilic substitution reaction, electrophilic substitution reaction, free radical substitution reaction), rearrangement reaction, oxidation-reduction reaction, condensation reaction, polymerization reaction (including polyaddition reaction, polycondensation reaction); metathesis reaction, combination reaction, decomposition reaction, displacement reaction, coordination reaction, oxidation-reduction reaction; enzyme catalysis reaction, fermentation reaction, photosynthesis, respiration, biotransformation reaction; homogeneous reaction, heterogeneous reaction (including gas-liquid reaction, gas-solid reaction, liquid-solid reaction, etc.), multiphase reaction (including gas-liquid-solid reaction, etc.); thermal initiation reaction, photo-initiation reaction, catalytic reaction (including homogeneous catalytic reaction, heterogeneous catalytic reaction, enzyme catalytic reaction), high-pressure reaction, low-pressure / vacuum reaction, electrochemical reaction; synthesis reaction, separation / purification reaction, conversion reaction, degradation reaction, modification reaction; environmental and personnel factors: laboratory area, ventilation conditions, number and flow rate of permanent personnel.

[0016] The weight distribution is based on expert opinions and historical accident statistics, ensuring that high-risk laboratories obtain higher TR values and reflecting the fair benchmark.

[0017] S102: Collecting laboratory static attribute data. Collect environmental data through sensors and cameras in the laboratory, and configure specific data of the above-mentioned risk factors for each laboratory.

[0018] S103: Calculating the theoretical risk value TR. The weighted scoring formula is used to automatically calculate TR:

[0019] TR = Σ (weight of risk factor i * score of the factor in laboratory j)

[0020] Weights are determined based on expert review or historical accident statistical analysis, reflecting the contribution of each factor to the overall risk. Scores are assigned based on the actual configuration of the laboratory.

[0021] Output: Each laboratory receives a quantified TR value that remains stable over a long period, serving as the "performance difficulty coefficient" for the laboratory's management personnel.

[0022] Furthermore, the theoretical risk value weighting factor and weighting allocation are as follows:

[0023] Hazardous chemical factor TR_chemical (weight: 0.35)

[0024] The chemical is rated based on its toxicity, flammability, explosiveness, corrosivity, and storage capacity.

[0025] Low risk: 1 point, weighting coefficient 1.0.

[0026] Medium risk: 3 points, weighting coefficient 1.5.

[0027] High risk: Score 5 points, weighting coefficient 2.0.

[0028] Furthermore, low-risk chemicals include water and salt solutions; medium-risk chemicals include organic solvents and weak acids; and high-risk chemicals include highly toxic chemicals and explosives.

[0029] The calculation formula is: TR_chemical = Σ(Chemical type score × Storage quantity coefficient) × 0.35. Instrument and equipment factor TR_equipment (weight: 0.30)

[0030] Risk levels are assessed based on factors such as equipment power, temperature, pressure, and radiation.

[0031] Low-risk equipment: 1 point.

[0032] Medium-risk equipment: 3 points.

[0033] High-risk equipment: 5 points.

[0034] Furthermore, low-risk equipment includes computers and printers; medium-risk equipment includes centrifuges and ovens; and high-risk equipment includes autoclaves and lasers.

[0035] The calculation formula is: TR_equipment = Σ(equipment risk score × quantity coefficient) × 0.30. Experimental process factor TR_process (weight: 0.20)

[0036] Scoring is based on the complexity of the experiment, the exothermic reaction, and the pressure change:

[0037] Low-risk process: 1 point.

[0038] Medium-risk process: 3 points.

[0039] High-risk process: 5 points.

[0040] Furthermore, low-risk processes include data calculation and sample observation; medium-risk processes include distillation and extraction; and high-risk processes include high-pressure synthesis and high-temperature reaction.

[0041] The calculation formula is: TR_process = Σ(Process Risk Score × Operation Frequency Coefficient) × 0.20. Environmental and personnel factor TR_environment (weight: 0.15)

[0042] Scoring is based on laboratory area, ventilation conditions, personnel density, and training level:

[0043] Low-risk environment: 1 point.

[0044] High-risk environment: 5 points.

[0045] Furthermore, low-risk environments include spacious, well-ventilated areas with adequate staff training; high-risk environments include crowded areas with poor ventilation and high staff turnover.

[0046] The calculation formula is: TR_environment=Σ(environment score×personnel coefficient)×0.15.

[0047] The total formula for calculating TR is:

[0048] TR=TR_chemical+TR_equipment+TR_process+TR_environment

[0049] Step S2: Dynamic perception and collection of actual risk data

[0050] S201: Collect dynamic data through the device.

[0051] Monitoring records: Record personnel violations and potential safety hazards.

[0052] Sensor recording: Real-time monitoring of potential environmental hazards.

[0053] Log integration: Obtain reported potential hazards and their status from the repair reporting system and inspection records.

[0054] Furthermore, violations include not wearing goggles; potential hazards include blocked safety passages and clutter in front of fire-fighting equipment; and environmental hazards include excessive concentrations of combustible gases and abnormal temperature and humidity.

[0055] S202: Store dynamic data. Collected events are stored in a unified database. Each record must include at least: laboratory number, event type, event severity, discovery time, rectification status, and rectification completion time.

[0056] Step S3: Calculation of Actual Risk Value (AR)

[0057] S301: Determine the assessment cycle. Set the performance assessment cycle to run from 00:00 on the 1st of each month to 24:00 on the last day of that month.

[0058] S302: Summarize dynamic data within the period. Extract all relevant event records of the target laboratory within this period from the database.

[0059] S303: Calculate AR value based on dynamic weight algorithm.

[0060] AR = Σ(Base deduction for event i * Efficiency weight * Recurrence weight)

[0061] Basic deduction: Preset based on event type and severity.

[0062] Efficiency weight: Set based on rectification time. For example, rectification within 24 hours has a weight of 0.5, rectification between 24 and 72 hours has a weight of 1.0, and rectification exceeding 72 hours has a weight of 1.5. This parameter directly assesses the response speed of management personnel.

[0063] Recurrence weight: If the same type of hidden danger occurs repeatedly in a short period of time, the weight is 1.5 for the second occurrence and 2.0 for the third. This parameter assesses the manager's ability to eradicate problems.

[0064] Furthermore, the weighting factors for the actual risk value are allocated as follows:

[0065] Basic deduction weight:

[0066] A base deduction value is preset based on the type and severity of the hazard:

[0067] Basic safety hazard: Deduct 1 point.

[0068] Medium-level hazard: Deduct 3 points.

[0069] High-level hazard: Deduct 5 points.

[0070] Furthermore, low-level hazards include items not being neatly arranged; medium-level hazards include not wearing protective equipment; and high-level hazards include fire and chemical leaks.

[0071] Rectification efficiency weight (weight range: 0.5~2.0):

[0072] Based on the time frame from the discovery of a potential hazard to its complete rectification:

[0073] Rectification within 24 hours: Weight 0.5 (reward for high efficiency).

[0074] Rectification within 24 - 72 hours: Weight 1.0 (standard).

[0075] Rectification exceeding 72 hours: Weight 1.5 (penalty for delay).

[0076] No rectification: Weight 2.0 (severe penalty).

[0077] Recurrence weight (weight range: 1.0 - 3.0):

[0078] Set according to the number of occurrences of the same type of hidden danger within the assessment period:

[0079] First occurrence: Weight 1.0.

[0080] Second occurrence: Weight 1.5.

[0081] Third and subsequent occurrences: Weight 2.0.

[0082] If the hidden danger involves major safety risks, the recurrence weight is increased to 3.0.

[0083] Furthermore, major safety hazards include fire hazards.

[0084] Total AR calculation formula:

[0085] AR = Σ (basic deduction × rectification efficiency weight × recurrence weight)

[0086] Output: Each laboratory obtains a quantified AR value within the assessment period.

[0087] Step S4: Calculation of management efficiency coefficient (MEF) and determination of performance level

[0088] S401: Calculate the management efficiency coefficient (MEF).

[0089] MEF = TR / AR

[0090] S402: Determine the final performance level based on MEF.

[0091] Compare the MEF value with the preset performance threshold and automatically output the performance level:

[0092] Performance level A (excellent): MEF ≥ MEF_Excellent. Indicates excellent management efficiency and outstanding safety performance under the challenge of inherent high risks.

[0093] Performance level B (qualified): MEF_Qualified ≤ MEF < MEF_Excellent. Indicates effective management and a risk control level that matches the inherent risks.

[0094] Performance level C (to be improved): MEF < MEF_Qualified. It indicates insufficient management effectiveness and failure to effectively control the risks in its jurisdiction.

[0095] Furthermore, MEF_Excellent and MEF_Qualified are preset thresholds, ranging from 0.5 to 1.5.

[0096] Step S5: Output of performance appraisal results and differential application

[0097] According to the adjudicated performance level, differential control and incentives are carried out as follows:

[0098] For performers with level A performance: Send commendation notices, associate performance bonuses, and grant priority for resource applications.

[0099] For performers with level B performance: Push standardized improvement suggestions and best practices.

[0100] For performers with level C performance: Automatically increase the monitoring frequency of their management laboratories and require their superior supervisors to intervene and formulate improvement plans.

[0101] Furthermore, the present invention proposes a management effectiveness model with adjustable weights, which is used to adaptively adjust the weights of each risk factor in the theoretical risk value TR and the actual risk value AR, so as to keep the performance score reasonable and fair under different laboratory scenarios. The management effectiveness model with adjustable weights includes:

[0102] 1. Data layer:

[0103] It is used to store laboratory static risk factor data, dynamic hidden danger event data, and historical performance result data, including:

[0104] Scores of indicators such as hazardous chemicals, instrument equipment, experimental processes, environment, and personnel in each laboratory; basic deductions, rectification time, and recurrence times of various hidden danger events; MEF values, accident rates, complaint records, expert manual evaluation results, etc. within the historical assessment period.

[0105] 2. Rule and initial weight layer:

[0106] According to existing expert experience and regulatory standards, give the initial weight vector of each risk factor

[0107]

[0108] The initial weights of the TR portion correspond to the initial values ​​of hazardous chemical factors, instrument and equipment factors, experimental process factors, and environmental and personnel factors. The initial weights of the AR portion correspond to factors such as basic deductions, rectification efficiency, and recurrence. This layer also specifies the range of weight values ​​and business constraints: each weight is non-negative and the sum is 1; the weight of the hazardous chemical factor is not less than the weight of the environmental and personnel factor.

[0109] 3. Weight Learning and Regulation Layer:

[0110] This layer uses historical data and a constrained optimization algorithm to iteratively adjust the weights, resulting in an updated weight vector w. *

[0111] While minimizing deviations from the initial expert weighting system, this layer aims to ensure the final scoring results are more consistent with the "actual management effectiveness" and reduce scoring biases between different types of laboratories. It includes: an objective function construction module; a constraint management module; a weight iteration and update module; and an online fine-tuning module for incrementally updating weights based on the latest periodic data.

[0112] 4. Rating Service Layer:

[0113] The updated weight vector is written into the calculation formulas of TR and AR to form a dynamically adjustable MEF calculation engine, providing a unified interface service for performance appraisal systems and mobile applications, and realizing automatic calculation and grade output of TR / AR / MEF.

[0114] Through the aforementioned adjustable weight management efficiency model, this invention can not only ensure the fairness of the assessment system during the design phase, but also automatically adjust the weights based on actual data during long-term operation, avoiding structural unfairness caused by a single, rigid weight system, thereby further improving the rationality and credibility of performance scoring.

[0115] Furthermore, the core formula principle is explained as follows:

[0116] 1. Definition of weight vector

[0117] Taking TR as an example, the weights of each risk factor are defined as an adjustable vector:

[0118] w TR =(w c ,w e ,w p ,w enc )

[0119] Among them, hazardous chemical factor w c Instrument and equipment factor w e Experimental process factor w p and environmental and human factors wenv ,satisfy:

[0120] w c +w e +w p +w env =1,w c ,w e ,w p ,w env If ≥0, then the total TR calculation formula is rewritten as:

[0121] TR=TR chemical ·w c +TR equipment ·w e +TR process ·w p +TR environment ·w env

[0122] Among them, TR chemical TR is the TR value for hazardous chemical factors. equipment TR is the TR value of the instrumentation factor. process TR is the experimental process factor. environment The TR value represents the environmental and human factors.

[0123] Similarly, regarding the "basic deduction" in AR... base "Rectification efficiency w" eff "Repeated occurrence w" rep "Equal factors, define the AR weight vector w" AR :w AR =(w base ,w eff ,w rep )

[0124] It is used to dynamically adjust the speed of rectification and the sensitivity to recurrence in different organizations or at different times.

[0125] 2. The principle of integrating expert weighting and data-driven weighting

[0126] This invention employs an "expert + data fusion" weight construction method. For any weight vector w, the following formula is used:

[0127] w * =(1-α)w( 0) +αW (data) ,0≤α≤1

[0128] in:

[0129] w (0) The initial weights given by the experts;

[0130] w (data) The corrected weights, obtained by fitting historical data, are derived by constructing a loss function, calculating its gradient, and then performing constrained iterative optimization.

[0131] α is the fusion coefficient, set by the model administrator or supervisor based on management maturity. Initially, it is set to 0.2–0.3, increasing to 0.5 or higher later.

[0132] 3. Data-driven weight optimization objective design

[0133] To improve the rationality and fairness of performance scoring, this invention constructs a comprehensive loss function L(w) that takes into account the following objectives:

[0134] Reasonableness objective: The deviation between the scoring results and the "actual management effectiveness indicators" should be minimized. "Actual management effectiveness indicators" can be constructed using metrics such as accident rate, number of serious accidents, number of compliance complaints, and expert evaluation scores. For example, a comprehensive effectiveness score y is constructed for each laboratory. i .

[0135] Fairness objective: The MEF distribution should conform as closely as possible to expectations among laboratories with different inherent risk levels. This can be reflected by constraining the difference in the mean MEF between high and low TR groups.

[0136] Stability objective: The new weights should not differ too much from the initial weights set by the experts, in order to ensure the interpretability and continuity of the system.

[0137] Based on the above objectives, the following example loss function is constructed:

[0138]

[0139] in:

[0140] MEF i (w) represents the management efficiency coefficient of the i-th laboratory under weight w;

[0141] f(·) is the mapping function from MEF to "effect score", which can be a linear or piecewise function;

[0142] Var group The sum of squares of the differences in the mean MEF between groups after grouping according to TR (high, medium, low) is used to measure "structural inequity".

[0143] λ1 and λ2 are trade-off parameters used to balance fairness and stability.

[0144] 4. Constraints

[0145] Weight optimization must meet several business and mathematical constraints, including but not limited to: non-negative weights and a sum of 1; for the TR part, the weight of hazardous chemicals is no less than the weight of instruments and equipment, and both are no less than the weight of environment and personnel; for the AR part, the weight of rectification efficiency and the weight of recurrence are no less than the basic deduction weight, in order to emphasize the importance of process management.

[0146] By incorporating these constraints, we can prevent the model from "overfitting" certain short-term phenomena under purely data-driven conditions, and ensure that the scoring system complies with the basic principles of safety management.

[0147] Furthermore, the algorithm and implementation process of the model are as follows:

[0148] 1. Offline training phase

[0149] When building or upgrading a model, offline training can be performed by following these steps:

[0150] S601: Sample Data Preparation

[0151] Select historical data from several assessment cycles;

[0152] For each laboratory i, collect:

[0153] Factor scores required for static TR calculation; hazard data required for dynamic AR calculation; actual accidents, penalty records, complaint information, etc.; manual evaluation of the safety management effectiveness for this period by experts or superiors. i .

[0154] S602: Initial Weight Setting

[0155] Using the fixed weight system (0.35 / 0.30 / 0.20 / 0.15) described in this invention as w (0) Enter the business constraints into the model.

[0156] Furthermore, business constraints include weight ranges and their relative sizes.

[0157] S603: Constructing the loss function

[0158] Based on the aforementioned loss function definition, calculate the MEF value of each laboratory under the current weight w, and then construct the overall loss L(w).

[0159] S604: Iterative Optimization with Constraints

[0160] The weights are iteratively updated using a programmable and easily implemented optimization algorithm:

[0161] Projective gradient descent algorithm:

[0162] 1. Calculate the gradient of the loss function with the current weights.

[0163] 2. According to

[0164] Perform a gradient update, where η is the learning rate;

[0165] 3. Projecting onto the constraint set involves adjusting it to satisfy conditions such as "sum equal to 1, non-negativity, interval constraints, and size relationship," thus obtaining a valid w. (t+1) Alternatively, an existing quadratic programming / convex optimization solver can be used to solve the problem, reducing the amount of manual coding. In this case, w (t+1) That is, w (data) The algorithm can run on a regular PC server, has low data dimensionality, manageable computational load, and is easy to integrate with existing OA or laboratory safety management platforms.

[0166] S605: Convergence and Testing

[0167] Optimization stops when the change in the loss function between two consecutive iterations is below a threshold, or when the maximum number of iterations is reached; the final weights w are then used to optimize the system. (data) The MEF distribution was verified on high-risk, medium-risk, and low-risk laboratory sets to determine whether it met fairness expectations; after expert review, the fused w... * Write it into the configuration as the official weight for the new round of assessment.

[0168] 2. Fine-tuning and continuous learning

[0169] During model operation, perform "lightweight updates" using the following strategy to ensure the model remains consistent with the latest management practices:

[0170] S701: Summary of Periodic Data

[0171] At the end of each assessment cycle, the newly added data for that cycle is automatically summarized and the sample set is updated.

[0172] S702: Minor Updates

[0173] With current weight w * As initial values, a small number of iterations are performed on the data from the most recent 2 to 3 periods to make minor adjustments; the adjustment range is controlled by a small learning rate and a larger λ2 to ensure that the weight changes are smooth and interpretable.

[0174] Furthermore, a small number of iterations may consist of 5 to 10 steps.

[0175] S703: Version Control and Rollback

[0176] The model generates a version number for each weight update and saves historical versions. If a weight update is found to cause obvious unreasonableness in practical operation, it can be manually rolled back to the previous stable version with one click.

[0177] Through the aforementioned offline training and fine-tuning mechanism, the weight control model of this invention can gradually transform the performance scoring system from "experience-based" to a dynamic and fair system of "data-driven + experience-constrained" without changing the existing business processes.

[0178] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0179] Achieving fair performance evaluation: By introducing TR values ​​to calibrate performance benchmarks, personnel who achieve outstanding management effectiveness in high-risk laboratories can receive high evaluations, fundamentally solving the problem of unfair incentives and stimulating the enthusiasm of all employees.

[0180] Positive shift in guiding behavior patterns: Because efficient rectification can improve MEF values, the assessment orientation has shifted from "avoiding problems" to "proactively and efficiently solving problems," forming a virtuous cycle driven by incentives, thereby significantly reducing the recurrence rate of problems and promoting the formation of a preventive safety culture.

[0181] Provides objective and quantitative decision-making basis: all data is collected by equipment, and the algorithm determines performance. The process is transparent and the results are reliable, which greatly reduces disputes caused by human evaluation.

[0182] Empowering Precise Management and Capability Enhancement: Detailed TR / AR / MEF data provides precise data support for performance reviews and individual capability enhancement plans. The "rectification efficiency weight" serves as a strong incentive signal, encouraging managers to optimize rectification processes and even fostering micro-innovations such as rapid response mechanisms or digital tools, thereby continuously improving management efficiency and driving the organization's continuous progress in safety management capabilities. Detailed Implementation

[0183] To provide a further understanding of the purpose, structure, features, and functions of the present invention, detailed descriptions are provided below in conjunction with embodiments.

[0184] The following uses the "Organic Synthesis Laboratory A" of the School of Chemistry of a certain university as an example to give a complete and detailed description of the implementation process of the present invention, starting from step 1.

[0185] Step 1: Calculation of the theoretical laboratory risk value (TR)

[0186] 1.1 Establish a database of inherent risk factors in the laboratory

[0187] For organic synthesis laboratory A, the following inherent risk factor library is first established according to the risk factor system in the instruction manual:

[0188] Hazardous chemical factors: involving flammable organic solvents (ethanol, acetone, acetonitrile), corrosive acids and alkalis (concentrated sulfuric acid, sodium hydroxide), and small amounts of toxic reagents (heavy metal salts).

[0189] Instrumentation and equipment: Equipped with a high-temperature oil bath, electric heating mantle, rotary evaporator, high-speed centrifuge, fume hood, etc.

[0190] Experimental process factors: include multi-step organic synthesis, involving medium- to high-risk processes such as exothermic reactions and vacuum distillation.

[0191] Environmental and personnel factors: The laboratory area is approximately 80m² 2 There are about 10 long-term users, and the staff training is good, but the staff density is too high.

[0192] The weights of each factor are as set by experts in the instruction manual:

[0193] Hazardous chemical factor weight: 0.35

[0194] Instrument and equipment factor weight: 0.30

[0195] Experimental process factor weight: 0.20

[0196] Environmental and personnel factor weights: 0.15

[0197] 1.2 Static Attribute Data Collection and Scoring

[0198] (1) Hazardous chemical factor TR_chemical

[0199] Based on the main chemicals in the laboratory, a comprehensive score was given according to toxicity, flammability and explosiveness, corrosivity, and storage quantity:

[0200] High-risk chemicals: Some explosive and highly corrosive reagents, converted to a total score of 5, with a weighting coefficient of 2.0;

[0201] Medium-risk chemicals: organic solvents and weak acids, converted to a total score of 3, with a weighting coefficient of 1.5;

[0202] Low-risk chemicals: partially buffered salt solutions, converted to a total score of 1, with a weighting factor of 1.0.

[0203] After comprehensive statistics, the results were obtained

[0204] TR c chemical=(5×2.0+3×1.5+1×1.0)×0.35≈5.43

[0205] (2) Equipment factor TR_equipment

[0206] Scoring is based on equipment risk level and quantity coefficient:

[0207] High-risk equipment: Assigned as high-risk equipment, with an overall risk score converted to 4 × 2.0;

[0208] Medium-risk equipment: Risk score converted to 6 × 1.5;

[0209] Low-risk equipment: Risk score is converted to 10 × 1.0.

[0210] but:

[0211] TR e quipment=(4×2.0+6×1.5+10×1.0)×0.30≈8.10

[0212] (3) Experimental process factor TR_process

[0213] Based on the complexity of the process and the frequency of operation:

[0214] High-risk processes: converted to 3 × 2.0;

[0215] Medium-risk process: converted to 5 × 1.5;

[0216] Low-risk process: converted to 2 × 1.0. Therefore:

[0217] TR p process=(3×2.0+5×1.5+2×1.0)×0.20≈3.10

[0218] (4) Environmental and personnel factors TR_environment

[0219] Scoring is based on factors such as area, ventilation conditions, personnel density, and training level:

[0220] Due to the relatively high population density but acceptable ventilation, the overall risk level is rated as high, with an environmental score of approximately 4 and a population coefficient of 1.5. Therefore:

[0221] TR e nvironment=(4×1.5)×0.15≈0.90

[0222] (5) Total TR value

[0223] TR=TR c hemical+TR e quipment+TR p rocess+TR e nvironment≈5.43+8.10+3.10+0.90≈17.53

[0224] The TR value remains stable over a relatively long period and is used to reflect the inherent risk and performance difficulty coefficient of Laboratory A.

[0225] Step 2: Dynamic perception and collection of actual risk data

[0226] 2.1 Monitoring Equipment Configuration

[0227] In laboratory A, various monitoring devices are deployed:

[0228] Video surveillance cameras: covering fume hoods, reagent storage areas, and main workbenches, automatically identifying potential hazards such as the wearing of safety helmets, goggles, and lab coats, and whether passageways are blocked.

[0229] Environmental sensors: including combustible gas alarms, temperature and humidity sensors, local smoke detectors, etc., are used to monitor gas leaks and environmental anomalies.

[0230] Log recording and repair system interface: It connects with the laboratory management system to automatically receive inspection records, hazard reports, and repair work orders.

[0231] 2.2 Dynamic Event Recording

[0232] During a given monthly assessment period, the system automatically collects and stores the following typical potential hazard events:

[0233] Event E1:

[0234] Type: High-risk hazard (flammable solvent bottles in the fume hood were not properly capped in time);

[0235] Detection method: Intelligent identification through video surveillance;

[0236] Basic deduction: 5 points;

[0237] Discovery time: March 5, 10:00 AM;

[0238] Rectification completion time: 16:00 on March 5 (rectification within 6 hours);

[0239] Does it recur? This is the first time this type of potential hazard has occurred within this period.

[0240] Event E2:

[0241] Type: Medium-risk hazard (some students were not wearing safety goggles during operation);

[0242] Discovery method: Surveillance video + manual review;

[0243] Basic deduction: 3 points;

[0244] Discovery time: 15:00 on March 10;

[0245] Rectification completion time: 10:00 AM, March 14 (if more than 72 hours have passed);

[0246] Does it recur? This is the second time this type of potential hazard has occurred within this period (there has been a similar record before).

[0247] Event E3:

[0248] Type: Low-level hazard (disorganized items on the lab bench, obstructing safety passageways);

[0249] Discovery method: On-site inspection records are entered into the system;

[0250] Basic deduction: 1 point;

[0251] Discovery time: 9:00 AM, March 20th;

[0252] The rectification was not fully completed by the end of the assessment period;

[0253] Has it occurred repeatedly? It has occurred multiple times within the assessment period (more than the third time).

[0254] All of the above events were written into the database in the form of fields such as "laboratory number, event type, severity, discovery time, rectification status, rectification completion time, and number of repetitions".

[0255] Step 3: Calculation of Actual Risk Value (AR)

[0256] 3.1 Determining the Assessment Cycle

[0257] The current performance evaluation period is set to the entire month of March mentioned above. The system retrieves all hazard event records for Laboratory A within this period.

[0258] 3.2 Dynamic Weight Calculation

[0259] According to the AR calculation rules in the instruction manual:

[0260] [AR = Σ(Base Deduction × Rectification Efficiency Weight × Repeat Occurrence Weight)] is calculated for the three example events above:

[0261] Event E1:

[0262] Basic deduction: 5 points (advanced hazard);

[0263] Rectification efficiency: Rectification within 24 hours → efficiency weight 0.5;

[0264] Repeated occurrence: First occurrence → Repeat weight 1.0;

[0265] Contribution: AR E 1 = 5 × 0.5 × 1.0 = 2.5

[0266] Event E2:

[0267] Basic deduction: 3 points (intermediate level hazard);

[0268] Rectification efficiency: Rectification takes more than 72 hours → efficiency weight 1.5;

[0269] Repeated occurrence: Second occurrence → Repeat weight 1.5;

[0270] Contribution: AR E 2 = 3 × 1.5 × 1.5 = 6.75

[0271] Event E3:

[0272] Basic deduction: 1 point (minor safety hazard);

[0273] Rectification efficiency: No rectification → efficiency weight 2.0;

[0274] Repeated occurrence: Third time or more → Repeat weight 2.0;

[0275] Contribution: AR E 3 = 1 × 2.0 × 2.0 = 4.0

[0276] Then, the total AR value of Laboratory A during this assessment period is:

[0277] AR = AR E 1+AR E 2+AR E 3 = 2.5 + 6.75 + 4.0 = 13.25 Step 4: Management Effectiveness Factor (MEF) and Performance Rating Determination

[0278] 4.1 Calculation of Management Efficiency Coefficient

[0279] According to the core formula of this invention:

[0280] [MEF=TR / AR]

[0281] Substitute TR and AR into the above:

[0282] [MEF≈17.53 / 13.25≈1.32]

[0283] 4.2 Setting and Determining Performance Rating Thresholds

[0284] Set the threshold according to the example in the instruction manual:

[0285] Excellent Performance (Grade A): MEF ≥ MEF_Excellent (value is 1.2);

[0286] Performance meets Grade B: MEF_Qualified (value 0.8) ≤ MEF <

[0287] MEF_Excellent;

[0288] Performance needs improvement, Grade C: MEF <MEF_Qualified。

[0289] Since the MEF of Laboratory A is approximately 1.32 ≥ 1.2, the system automatically determines its safety performance level for this assessment cycle to be Grade A (Excellent Performance).

[0290] Step 5: Application Example of Adaptive Weight Adjustment

[0291] After several assessment cycles, the system will use historical data to fine-tune the weights of each risk factor in TR and AR to further improve the fairness and rationality of the scoring. For example:

[0292] If statistics show that the accident rate of high-TR laboratories has significantly decreased after strict implementation of management measures, but the MEF remains low, then the system can appropriately improve the MEF using a constrained optimization algorithm.

[0293] "Rectification efficiency weight" or "recurrence weight" to more fully reward efficient management behavior;

[0294] The optimization process follows the principle of "expert weighting + data correction" in the instruction manual, ensuring that the new weights achieve dynamic optimization of the score while meeting business constraints such as "non-negative, sum to 1, and the weight of hazardous chemicals being greater than that of environmental factors".

[0295] As can be seen from the above embodiments, the method of the present invention can achieve the following in practical use: fair performance evaluation of laboratories with different inherent risk levels, positive incentives for efficient management behavior, and long-term dynamic optimization of the weighting system.

[0296] The present invention has been described in the above-described embodiments; however, these embodiments are merely examples for implementing the present invention. It must be noted that the disclosed embodiments do not limit the scope of the present invention. Conversely, any modifications and refinements made without departing from the spirit and scope of the present invention are within the scope of patent protection of the present invention.

Claims

1. A dynamic evaluation and management method for laboratory safety performance levels based on management effectiveness factors, characterized in that: Includes the following steps: S1: Calculate the theoretical risk value TR of the laboratory. The TR value is calculated based on the laboratory's static inherent risk factors through weight allocation and weighted comprehensive evaluation method. The weight allocation and weighted comprehensive evaluation method includes weight allocation of static inherent risk factors and weighted calculation of the allocated static inherent risk factors. S2: Real-time collection of laboratory safety hazard data through monitoring equipment, including surveillance cameras and sensors; the safety hazard data includes hazards related to hazardous chemicals, instruments and equipment, experimental processes, and the environment and personnel. S3: Based on the safety hazard data collected in step S2, calculate the laboratory's actual risk value AR within the set assessment period using a weighted comprehensive evaluation method. The calculation of the AR value includes a rectification efficiency weight and a recurrence weight. The weighted comprehensive evaluation method includes weighting the safety hazard data and weighting the safety hazard data after weighting. S4: The Management Effectiveness Factor (MEF) is calculated according to the formula MEF = TR / AR, and the MEF value is compared with a preset threshold to determine the safety performance level of the laboratory. S5: Implement differentiated management strategies based on the safety performance level determined in step S4; S6: Establish a management efficiency model with adjustable weights, and apply the management efficiency model with adjustable weights by using a constrained optimization algorithm to iteratively update the weights of each risk factor in TR and AR based on historical data from multiple assessment cycles. S7: Use the updated weight vector obtained in S6 to calculate TR, AR, and MEF in subsequent assessment cycles; the constrained optimization algorithm includes constraints and optimization objectives, and is responsible for finding a set of optimal weights; the historical data includes all TR, AR, and MEF values ​​accumulated over a period of time. The management efficiency model architecture with adjustable weights described in step S6 is as follows: S61. Data Layer: Used to store static risk factor data, dynamic hazard event data, and historical performance results data of the laboratory, including: Scores for hazardous chemicals, instruments and equipment, experimental processes, environment and personnel in each laboratory; basic deductions, rectification time and recurrence frequency for various potential incidents; MEF values, accident rates, complaint records and expert evaluation results for the historical assessment period; S62. Rules and Initialization of Weight Layer: Based on existing expert experience and regulatory standards, the initial weight vector W for each risk factor is given. 0 wK0 represents the initial weight of the risk factor; k represents the number of risk factors. The risk factors include those in the TR (Resolution-Based) section and those in the AR (Audit-Based) section. The risk factors in the TR section include hazardous chemical factors, instrument and equipment factors, experimental process factors, and environmental and personnel factors. The risk factors in the AR section include basic deductions, rectification efficiency, and recurrence factors. The initial weights of the TR section correspond to the initial values ​​of the hazardous chemicals factor, instrument and equipment factor, experimental process factor, and environmental and personnel factor; the initial weights of the AR section correspond to the basic deduction, rectification efficiency, and recurrence factor. This layer also specifies the range of weight values ​​and business constraints; the range of weight values ​​is that each weight is non-negative and the sum is 1; the business constraints are that the weight of the hazardous chemical factor is not less than the weight of the environmental and personnel factors. S63. Weight Learning and Regulation Layer: This layer uses historical data and a constrained optimization algorithm to iteratively adjust the weights, resulting in an updated weight vector. ; S64. Rating Service Layer: The updated weight vector is written into the calculation formulas of TR and AR to form a dynamically adjustable MEF calculation engine, realizing the calculation and level output of TR / AR / MEF.

2. The dynamic evaluation and management method for laboratory safety performance levels based on management effectiveness factors as described in claim 1, characterized in that: The weight vector of the TR part This includes weights for hazardous chemicals, instruments and equipment, experimental processes, and environment and personnel, with each weight being non-negative and summing to 1. The weight of the hazardous chemicals factor is not less than the weight of the environment and personnel factor; the weight vector of the AR part. It should include at least the weights of basic deduction factors, rectification efficiency factors, and recurrence factors, and can be adjusted according to the management needs of different organizations or different periods under business constraints; The weight learning and regulation layer uses a fusion of expert weights and data-driven weights to obtain the final weight vector. ,Right now: ； in, The initial weights given by the experts, The corrected weights are obtained by fitting historical data, constructing a loss function and calculating its gradient, and then performing constrained iterative optimization. α is the fusion coefficient, and its value range is set by the model administrator or supervisor.

3. The dynamic evaluation and management method for laboratory safety performance levels based on management effectiveness factors as described in claim 1, characterized in that: In step S1, the static inherent risk factors include at least one of the following: hazardous chemical factors, instrument and equipment factors, experimental process factors, and environmental and personnel factors.

4. The dynamic evaluation and management method for laboratory safety performance levels based on management effectiveness factors as described in claim 1, characterized in that: In step S1, the calculation of the TR value and the allocation of risk factor weights are as follows: The hazardous chemical factor TR_chemical has a weight of 0.35 and is scored based on the chemical's toxicity, flammability, explosiveness, corrosivity, and storage capacity. The calculation formula is as follows: TR_chemical = Σ(Chemical type score × Storage quantity coefficient) × 0.35; The instrument and equipment factor TR_equipment has a weight of 0.

30. It is scored based on the risk level of the equipment, including power, temperature, pressure, and radiation. The calculation formula is as follows: TR_equipment = Σ(Equipment Risk Score × Quantity Coefficient) × 0.30; The experimental process factor TR_process has a weight of 0.

20. Based on the experimental complexity, exothermic reaction, and pressure change scores, the calculation formula is as follows: TR_process = Σ(Process risk score × Operation frequency coefficient) × 0.20; The environmental and personnel factor TR_environment has a weight of 0.15 and is scored based on laboratory area, ventilation conditions, personnel density, and training level. The calculation formula is as follows: TR_environment = Σ(Environmental Score × Personnel Coefficient) × 0.15; The total formula for calculating TR is: TR=TR_chemical+TR_equipment + TR_process + TR_environment.

5. The dynamic evaluation and management method for laboratory safety performance levels based on management effectiveness factors as described in claim 1, characterized in that: The AR weight allocation in step S3 is as follows: Basic deduction weight: The basic deduction value is preset according to the type and severity of the hazard: 1 point for low-level hazards, 3 points for medium-level hazards, and 5 points for high-level hazards; Rectification efficiency weight: The weight is set according to the time from the discovery of the hidden danger to the completion of rectification: 0.5 for rectification within 24 hours, 1.0 for rectification between 24 and 72 hours, 1.5 for rectification after 72 hours, and 2.0 for no rectification. Recurrence weight: The weight is set according to the number of times the same type of hidden danger occurs within the assessment period: the weight is 1.0 for the first occurrence, 1.5 for the second occurrence, and 2.0 for the third and subsequent occurrences. When it is a major safety risk, the recurrence weight can be increased to 3.

0. The AR total calculation formula is: AR = Σ(basic deduction × rectification efficiency weight × recurrence weight).

6. The dynamic evaluation and management method for laboratory safety performance levels based on management effectiveness factors as described in claim 1, characterized in that: In step S3, the rectification efficiency weight increases monotonically with the increase of rectification time; the recurrence weight increases monotonically with the increase of the number of times the same type of safety hazard occurs.

7. The dynamic evaluation and management method for laboratory safety performance levels based on management effectiveness factors as described in claim 1, characterized in that: In step S4, the safety performance level is divided into at least three levels based on the Management Effectiveness Factor (MEF), specifically: Grade A for outstanding performance, determined when MEF ≥ MEF_Excellent; Grade B, which meets the performance standards, is determined when MEF_Qualified ≤ MEF < MEF_Excellent; Grade C, where performance needs improvement, is determined when MEF < MEF_Qualified; MEF_Excellent and MEF_Qualified are preset performance thresholds.

8. The dynamic evaluation and management method for laboratory safety performance levels based on management effectiveness factors as described in claim 1, characterized in that: For A-level performance, a commendation notification and priority resource allocation are triggered; for B-level performance, standardized improvement suggestions are pushed; for C-level performance, the monitoring frequency is automatically increased, high-risk experimental activities are restricted, and the supervisor is notified to intervene.

9. The dynamic evaluation and management method for laboratory safety performance levels based on management effectiveness factors as described in claim 1, characterized in that, It also includes the step of constructing a management effectiveness model with adjustable weights, wherein the management effectiveness model with adjustable weights is used to adaptively adjust the weights of each risk factor in TR and AR, including: Data layer: Used to store static risk factor data, dynamic hazard event data, and historical performance result data of the laboratory; Rules and Initial Weight Layer: Based on expert experience and regulatory standards, the initial weight vectors for each risk factor are given, and the range of weight values ​​and business constraints are set. Weight learning and adjustment layer: Based on historical data, the weights are iteratively corrected using a constrained optimization algorithm to obtain an updated weight vector; Scoring Service Layer: The updated weight vector is written into the calculation formulas of TR and AR to form a dynamically adjustable MEF calculation engine.

Images