Iec 60300 3 2 2004

NORME INTERNATIONALECEI IEC INTERNATIONAL STANDARD 60300-3-2 Deuxième éditionSecond edition2004-11 Gestion de la sûreté de fonctionnement – Partie 3-2: Guide d'application – Recueil de

Généralités

Inventaire – Cela inclut l’information prouvant qu’une entité particulière existe en exploitation, comment elle est configurée et quelles sont les autres entités qu’elle contient

Utilisation – Cela inclut l’information sur la date de la mise en exploitation de l’entité, comment cette entité fonctionne en exploitation et la date à laquelle elle a été retirée de l’exploitation

Environnement – Cela inclut l’information sur les conditions de fonctionnement de l’entité, souvent en terme de facteurs qui sont considérés comme importants pour la sûreté de fonctionnement du produit

Evénements – Cela inclut l’information relative à tout ce qui peut survenir à l’entité pendant sa vie, telles que les défaillances, les réparations, les mises à niveau, etc

It is often not feasible to obtain all the necessary data for a specific reliability task, either due to operational issues or the high cost of data collection In such cases, it is essential to assess the reasons for requiring the data and conduct a trade-off analysis between the necessity of the data and the challenges of gathering it Sometimes, collecting the data may require changes to existing operational procedures within an organization, and the resulting difficulty and cost must be justified by the benefits gained from the analysis enabled by the data.

A statistical model will always approximate reality It is advisable to use engineering judgment and a goodness-of-fit (GOF) test to assess whether the approximation yields usable results Sensitivity to prerequisites can be evaluated using simulated data, such as through the Monte Carlo method.

Inventaire

Inventory records are crucial as they provide initial status, manufacturer details, lot numbers, modification history, repair records, and other essential information This data is vital for assessing factors that influence susceptibility to various events Without this information, reliability analysis cannot identify trends unique to specific subgroups of otherwise identical products.

Various types of events, such as failures, are inherent to the individual entity, originating either from the manufacturing process or from design weaknesses These events become apparent during the entity's lifecycle.

The course of life leads to the accumulation of experiences unique to individuals, allowing for a comprehensive life analysis when each person is specifically identified by a serial number at every record However, some life analyses do not require this identification.

For internal use at MECON Limited in Ranchi and Bangalore, distributional analysis, such as Weibull analysis, is the next step in data evaluation This method requires specific data criteria to effectively identify distributions, as outlined in relevant standards Alternatively, if distributional analysis is not feasible, non-parametric analysis can be utilized This approach has more flexible criteria but typically provides less detailed information.

8 Which data can be collected ?

Inventory – This includes information proving that a particular item exists in the field, how that item is configured, and what other items that item contains

Usage refers to the timeline of an item's deployment in the field, detailing when it was introduced, how it is utilized during its operation, and the point at which it is taken out of service.

Environment – This includes information about the operating conditions of the item, often in terms of factors that are considered important to the dependability of the item

Events – This include information about any thing that has happened to the item during its life, these will include failures, repairs, upgrades, etc

In many cases, obtaining all necessary data for a dependability task is challenging due to operational constraints or high collection costs It becomes essential to evaluate the reasons for data requirements and conduct a trade-off analysis between these needs and the difficulties of data collection Additionally, gathering the data may necessitate changes to existing operational processes, and the associated costs and challenges must be weighed against the benefits gained from the dependability analysis enabled by the data.

A statistical model inherently approximates data, making it essential to apply engineering judgment and goodness of fit (GOF) tests to assess the utility of these approximations Additionally, the sensitivity to preconditions can be analyzed through simulated data, such as employing the Monte Carlo method.

Inventory records play a crucial role in documenting the original build state, manufacturer, batch number, modification state, and repair history of items This information is essential for evaluating the factors that influence susceptibility to various events Without these records, dependability analysis cannot effectively identify trends that pertain to specific sub-groups of otherwise identical items.

Many event types, such as failures, are inherent to individual items due to manufacturing flaws or design weaknesses These events are triggered by life consumption, which begins at the initial switch-on Life consumption accumulates uniquely on each item, allowing for comprehensive life analysis if each item is identified by a unique serial number However, certain life analysis methods, like M(t) analysis in IEC 60605-6, do not require this specific identification.

LICENSED TO MECON Limited - RANCHI/BANGALORE FOR INTERNAL USE AT THIS LOCATION ONLY, SUPPLIED BY BOOK SUPPLY BUREAU.

It is essential to gather information on all at-risk entities within a population This population data can be derived from inventory information The collected data is typically referred to as "operational duration," which may include factors such as operating time, calendar duration, number of cycles, mileage, and number of copies.

Sometimes, it is neither feasible nor desirable to gather information from the entire population of an entity Therefore, sampling techniques can be employed to minimize the amount of required data These sampling methods are detailed in Article 12.

Utilisation

Usage refers to the measurement of the functions required from a product or system in service at a client, including the duration and frequency of these requirements It is essential to carefully consider the data to be measured to optimize the utility of the collected information, ensuring it can be used for future analyses in similar applications rather than being restricted to a specific use To define customer operational requirements, usage data is typically presented in terms of events, occurrence levels, and duration, along with statistical significance and associated risks, which are valuable during product qualification and development validation activities.

L’utilisation peut être continue dans le temps avec un niveau fixe, continue dane le temps avec un niveau variable ou sporadique avec un niveau constant ou variable

When equipment operates continuously at 100% capacity, calculating usage is straightforward However, estimating the average usage becomes challenging when two pieces of equipment are involved—one functioning constantly and the other occasionally, such as a backup Often, specific usage data for individual equipment is unavailable, necessitating the calculation of an average usage for the equipment type This approach can also highlight issues related to the equipment's monitoring requirements While users may provide their average usage after a phone conversation, this is less likely with military communication equipment users.

The importance of usage monitoring is crucial, as future data analyses and results may be obscured by significant inaccuracies in usage representation Many pieces of equipment are equipped with elapsed time indicators (ETI) to track real-time usage; however, these indicators can also have issues and may only provide a rough estimate of actual usage time.

L’utilisation peut non seulement être fondée sur le temps, mais aussi sur un nombre d’opérations ou de cycles (par exemple, combien de fois une entité est utilisée).

Environnement

The environment can significantly impact the information regarding a product or system's life, especially when the duration and intensity of environmental stresses must be considered in the qualification actions To accurately define operational requirements, it is essential to measure the environmental component in the application, which involves understanding the environmental inputs and the components' responses to these inputs These requirements establish a foundation for accelerated testing aimed at demonstrating compliance with reliability standards.

To effectively assess risks within a population, it is essential to gather comprehensive data on all at-risk items This data can be obtained from inventory records, typically reflecting "time in the field," which may include metrics such as operating time, calendar duration, number of cycles, mileage, and number of copies.

In certain situations, gathering data from the entire population may be impractical or unnecessary, making sampling techniques an effective method to limit the required information.

Sampling techniques are described in Clause 12

Usage data is essential for understanding customer service demands, including the frequency and duration of these demands To optimize the utility of this data, it is crucial to select the right metrics, allowing for broader analyses beyond specific applications Typically, usage data consists of event occurrences and durations, which help define customer requirements with statistical significance and associated risks, aiding in product or system qualification and validation processes.

The usage may be continuous over time at a fixed level, continuous over time at a variable level or sporadic over time at either a fixed level or variable level

Calculating equipment usage is straightforward when it operates continuously at 100% However, when one piece of equipment runs continuously while another serves as an occasional backup, estimating average usage becomes challenging Often, specific usage data for individual equipment is unavailable, necessitating an average for the equipment type This estimation can be complicated by the equipment's nature; for instance, a user of a telephone exchange may readily provide average usage data, whereas a customer with military communications equipment is less likely to do so.

Accurate usage data is crucial, as it can significantly impact further analysis and lead to substantial inaccuracies in results While many pieces of equipment are equipped with elapsed time indicators (ETI) to track actual usage time, these indicators can also present challenges and may only provide an approximate measure of true usage.

Usage may be not only time based, it may also be operations or cycle based (e.g how many times an item is used)

The environment plays a crucial role in the damage experienced by products or systems, necessitating the inclusion of environmental stress duration and intensity in qualification activities Accurately defining field use requirements requires measuring the component's environment and understanding its response to these inputs These established requirements serve as a foundation for equivalent accelerated tests, ensuring compliance with reliability standards.

A more severe environment can lead to events occurring more rapidly than in a less severe one As discussed in section 8.3, multiple environmental factors are typically relevant to a specific event, and depending on the required analysis, all of these may need to be recorded The location of the measured environment is also crucial; for instance, the environment within an airplane cabin differs significantly from that of an engine.

An environmental factor associated with usage is the wear caused by starting and stopping equipment Depending on the type of machinery, these changes in stress conditions can be significant and may outweigh the effects of permanent environmental conditions.

Evénements

Les événements de retrait peuvent inclure des défaillances, des actions de maintenance, etc

Failure events can encompass system failures, secondary failures, failures in redundant systems, failures that do not lead to system failure, and masked failures In many reliability engineering techniques outlined in the listed standards, failure is the most critical event.

When seeking to understand the resources and costs associated with maintenance related to a failure, it is essential to record repair information and identify repairs with sufficient detail for analysis It is important to note that repairs can also lead to subsequent failures, in addition to resolving the current issue Therefore, maintenance data serves as a crucial source for in-depth reliability analysis.

Before conducting data analysis related to events, it is essential to categorize these events into meaningful groups for the analyst For instance, a failure event in a complex electronic system can be classified under categories such as design, manufacturing, supplier, maintenance, damage, and software, with the possibility that the failure may remain undetected The categorization can sometimes be done at a more granular level, depending on the available data and the phenomenon under investigation, such as the type of component, reference position, and mode of failure.

Le processus d’analyse d’événement commence par une classification grossière des types d’événements et l’objet de la collecte des données d’exploitation relatives aux défaillances et la caractérisation de l’utilisation

Pour les événements de défaillance, l’analyse commence par la vérification des défaillances

If no failure is detected, it results in a conclusion of "no failure found." When a failure is confirmed, a detailed analysis can begin to isolate the actual mode of failure and the underlying mechanism causing it.

To effectively characterize usage, it is essential to ensure the collection of appropriate data This can be achieved by clearly defining data requirements and forecasting comprehensive analysis before initiating the measurement program A well-structured data collection plan and implementation should directly yield usable data for analysis and provide insights into usage status.

For software, failures are often intermittent or can be resolved by resetting the software In such cases, the customer's intent and the actual actions taken on the software can be significant for event classification.

A more severe environment may cause the event to occur sooner than one that is less severe

When analyzing a specific event, it is crucial to consider various environmental aspects, as outlined in section 8.3 Accurate recording of these elements is essential, especially since the locality of the measured environment can significantly differ; for instance, the in-cabin and on-engine environments of an aircraft are distinct from one another.

An environmental factor related to usage is the damage caused by switch on and switch off

Dependent on the type of equipment, this start-up/shut-down stress could be significant and of more importance than the steady state environmental conditions

Removal events encompass various occurrences such as failures and maintenance actions Failure events can be categorized into system failures, secondary failures, failures in redundant systems, non-critical failures, and hidden failures In numerous dependability techniques outlined in the relevant standards, failures are identified as the most significant events.

To effectively analyze resources and costs related to maintenance failures, it is essential to document maintenance repair information thoroughly This documentation should provide enough detail to facilitate analysis, as repair activities can lead to future failures in addition to resolving current issues Consequently, maintenance information serves as a crucial resource for comprehensive dependability analysis.

Before conducting data analysis on events, it is essential to categorize them into meaningful groups relevant to the analyst For instance, a failure event in a complex electronic system can be classified into categories such as design, manufacture, suppliers, maintenance, damage, software, or no failure found The level of categorization may vary based on the available data and specific concerns being investigated, which could include details like component type, reference position, and failure mode.

The process of event analysis begins with a broad classification of the type of event and purpose of field data collection for failures or usage characterization

In failure event analysis, the process starts with verifying the existence of a failure If no failure is detected, it is categorized as "no failure found." However, when a failure is confirmed, a thorough fault analysis is conducted to identify the specific failure mode and mechanism responsible for the issue.

For usage characterization, it is necessary to ensure that the right sort of data is collected

Effective data collection and instrumentation should be designed based on the data needs and planning analysis conducted prior to initiating the measurement program, ensuring that the gathered data is directly usable for analysis and transformation into valuable information about usage.

Software failures can be intermittent, known as soft errors, or may be resolved by simply resetting the software In such instances, understanding the customer's intent and their interactions with the software is crucial for accurately classifying the event.

Sources de données

There are numerous data sources for operational safety, but their availability and usefulness can vary depending on the type of product and the structure of the companies Therefore, it is not feasible to provide a comprehensive list of all data sources in this standard.

Direct information is collected by the manufacturer of the entity or product, while indirect information is gathered from third parties knowledgeable about the product, such as sellers and repairers The distinction between direct and indirect information often varies based on the type of product Generally, information regarding professional entities (e.g., telecommunications switches, factory equipment) is more likely to be direct, whereas consumer product information (e.g., household appliances, audiovisual devices, mobile phones) tends to be collected indirectly Direct information is usually preferred due to the assurance of data quality through proper collection procedures, whereas data from third parties often comes with unknown quality.

Population sizes and types of operating entities can be derived from sales, orders, deliveries, and installation records Often, all these sources are available for a specific entity, allowing for a comprehensive understanding of its situation Additionally, electronic product licenses can help identify the entity/software and the initiation of usage data to be reported.

Les enregistrements des produits pour la garantie sont aussi intéressants, par exemple pour les produits de consommation et les produits médicaux

Production information encompasses elements such as the internal structure of the product, including cards, modules, and components This type of information is typically found in the manufacturing records of batches or their equivalents Product stock statuses facilitate the identification of such data, even when manufacturing is not in progress Additionally, records of services, warranties, repairs, and used spare parts can provide valuable insights into which entities have actually failed and under what circumstances.

Product retirement records provide insights into when a product is decommissioned, indicating its exclusion from the analysis population Customer claims can be valuable, especially for specific products, as they reveal information about failures, particularly intermittent ones Customer reports and feedback can enhance the dataset Additionally, claims and their insurance coverage, when available, can help identify the location and usage of an entity Including warranty cards with entities allows buyers or users to return the card upon purchase or commissioning, yielding valuable information In many market sectors, this is often the only way to obtain such information.

Sometimes, an entity can be configured to inform the manufacturer about its commissioning This type of entity typically includes telecommunications hardware or devices connected to a communication system, which can also provide the manufacturer with information about their usage and health status When the entity no longer transmits information, it is reasonable to assume that it has been taken out of service For high-value equipment, it is possible to add telecommunications functions specifically for this purpose.

Numerous dependability data sources are available, but their accessibility and usefulness can differ based on product types and organizational structures Consequently, it is impractical to enumerate all data sources within this standard.

Direct information is gathered by the manufacturer of a product, while indirect information comes from third parties with knowledge of the product, such as those involved in sales or repairs The balance between direct and indirect information varies by product type; professional items like telecommunication switches and plant equipment typically rely more on direct information, whereas consumer items such as mobile phones and appliances often depend on indirect sources Direct information is generally preferred due to the assurance of quality through proper data collection methods, whereas third-party data may have uncertain quality.

Population sizes and item types can be derived from various records, including sales, dispatch, order, delivery, and installation data Access to these records enables a comprehensive understanding of item locations Additionally, electronic product authorizations can provide insights into item or software locations and usage start dates Product registrations for warranty purposes, particularly for consumer and medical products, further enhance this information.

Production information encompasses the internal structure of a product, detailing the specific boards, modules, and components utilized This data is typically found on production job cards or similar documentation Additionally, product stock-holdings help identify manufactured items that are not yet operational Servicing records, warranty information, repair logs, and spare parts used provide valuable insights into which items have failed and the circumstances surrounding those failures.

Disposal records provide crucial information regarding the removal of products from service, indicating that they should not be included in analysis Customer complaints can help pinpoint the location of specific items and offer insights into failures, particularly intermittent ones Additionally, customer reports and comments can enhance data sets Insurance claims and coverage records are valuable for identifying the location and usage of items Furthermore, warranty cards sent with products allow purchasers to provide essential information upon purchase or service entry, often serving as the sole source of such data in many market sectors.

Certain items, particularly telecommunications devices, can be configured to automatically notify the manufacturer upon activation These devices can also report their usage and health status If an item fails to report, it may indicate that it is no longer in use For high-value equipment, a dedicated telecommunication function can be integrated specifically for this reporting purpose.

9 Méthodes d’analyse et exigences pour les données

Several IEC standards and guidelines provide instructions and support for the analysis of reliability data Table 1 lists all reliability standards that include data-related requirements, organized by their reference numbers The table specifies which methods within each standard can be utilized and identifies the necessary data to collect for implementing the proposed techniques.

La suite donne un exemple de l’utilisation des tableaux

A company may wish to assess the reliability of an electronic component under specific operational conditions By utilizing the IEC 61709 standard, it can convert the failure rate of electronic components across various environmental conditions The company notes that the data requirements include constant failure rates for an electronic component under defined conditions and information about the environment in which the electronic component is used.

To achieve constant failure rates, it is essential to utilize IEC 60605-4, which outlines the requirements for data related to operating durations before failure Additionally, to maintain a consistent failure rate, the methods specified in IEC 60605-6 must be applied, along with other requirements for data concerning operating durations before each relevant failure.

Tableau 1 – Exigences relatives aux données pour les méthodes de sûreté de fonctionnement, pourquoi les utiliser, et les références CEI

Quoi faire ? Donnée nécessaire Référence CEI Titre

Comment appliquer le concept de cycle de vie

Cỏt des éléments identifiés et cỏt total du projet

CEI 60300-3-3 Evaluation du cỏt du cycle de vie

Comment sélectionner et appliquer les techniques d’analyse de risque

Fréquence des événements identifiés, probabilité d’occurrence des événements et durée de ces occurrences

CEI 60300-3-9 Analyse du risque des systèmes technologiques

Comment présenter les données de fiabilité des composants et pièces

Nombre de défaillances des composants concernés, modes de défaillance et durée avant défaillance des composants concernés

CEI 60319 Présentation et spécification des données de fiabilité pour les composants électroniques

Comment estimer les taux de défaillance constants

Durées avant défaillance des entités (la méthode graphique nécessite un minimum de quatre observations par durée avant défaillance)

CEI 60605-4 Méthodes statistiques de distribution exponentielle

Comment établir si un taux de défaillance est constant

Durée avant défaillance pour chaque défaillance (la méthode numérique nécessite un minimum de

10 observations par durée avant défaillance; la méthode graphique en nécessite quatre)

CEI 60605-6 Tests de validité des hypothèses du taux de défaillance constant ou de l’intensité de défaillance constante

9 Analysis methods and their data requirements

Généralités

Les différentes approches de la collecte de données sont décrites de 12.2 à 12.5.

Basée sur le temps – continue ou discontinue

There are several potential methods for time-based data collection: a) continuous data collection; b) data collection over a specific period; c) data collection across multiple periods; and d) data collection using a moving period.

La collecte de données continue est menée tout au long du cycle de vie de l’entité sans interruption, comme le montre la Figure 2

In the planning phase of data collection, it is essential to clearly define responsibilities for data recording, processing, and analysis Additionally, it is important to establish authority and accountability for acting on the results Regular reviews of the entire data collection and analysis process should be conducted to ensure its effectiveness and adequacy.

Effective planning for dependability data collection involves several key considerations: first, identify the stakeholders and the specific information they seek; second, determine the types of analyses that can be conducted and the tools required; third, specify the data to be collected and the sources from which it will be obtained; fourth, establish the timing and frequency of data collection; and finally, decide on the methods for data collection, whether manual, semi-automatic, automated, or remote automated.

Establishing lasting relationships with customers is essential for effective data gathering, which involves implementing suitable methods for recording and transmitting data, providing feedback to customers, and taking corrective actions on products when necessary.

The different approaches to data collection are described in 12.2 to 12.5

12.2 Time based – continuous and discontinuous

There are several potential methods for time based data collection: a) continuous data collection; b) windowed data collection; c) multiple windowed data collection; d) rolling window data collection

Continuous data collection amasses data constantly throughout the life of an item, as shown in Figure 2

Mise en service Mise au rebut

Temps Collecte continue des données

Figure 2 – Collecte de données continue

Data collection occurs during a specific phase in the product lifecycle, such as from commissioning to the end of the warranty period.

Collecte des données sur une période

Figure 3 – Collecte de données sur une période

Entry into Service Removal from service

Windowed data collection amasses data from a single window in the product life cycle, for example, from product introduction until the end of the warranty period See Figure 3

Entry into service Removal from service

La collecte de données sur plusieurs périodes est effectuée sur plusieurs périodes pendant la vie du produit Voir Figure 4

Figure 4 – Collecte de données sur plusieurs périodes

Data collection over a moving period is akin to collection over a fixed period, with the key difference being that the start and end points are dynamic This approach ensures that older data is continuously discarded as new data is gathered, reflecting the obsolescence of products and maintaining the data acquisition system's storage capacity Non-relevant data is removed to prioritize the recording of significant events and their context Additionally, data from a fixed period can be utilized to calculate average values over operational durations, often achieved through short-term collection while assuming that observed values remain consistent throughout the entire operational period.

It is important to note that the reference to time in data collection may not be calendar-based; other references are available These time references can include operational time—the duration during which the system is functional—or the power-on duration, which encompasses both idle time and operational time, among others.

Additionally, the reference to time may not necessarily imply a duration and can instead be based on the number of operations or cycles, such as the number of car starts or kilometers driven There is often a relationship between these time references and the operational or calendar duration, which is referred to as usage.

Multiple window data collection collects data from several multiple time windows during the product life See Figure 4

Entry into service Removal from service

Figure 4 – Multiple window data collection

Rolling window data collection continuously updates the time frame for data analysis, discarding the oldest data as new information is gathered This approach helps manage obsolescence in products and optimizes memory usage in data acquisition systems by eliminating irrelevant data When significant events occur, this method allows for the capture of additional context and related events Additionally, windowed data can be utilized to calculate average values over specific operating periods, based on the assumption that short-term observations reflect the overall trends during the entire period.

When collecting data, it's important to recognize that the time metric used may not correspond to calendar time Alternative measures such as "operating time," which refers to the duration a system is actively functioning, or "powered up time," which includes both standby and operational periods, can be utilized Additionally, some time metrics may be based on operations or cycles, such as the number of trips taken or the distance a vehicle has traveled There is often a correlation between these various time metrics, particularly between operating time and calendar time, a relationship referred to as usage.

Figure 5 – Différentes références au temps

Figure 5 illustrates a typical mission profile, highlighting that the calendar duration increases throughout the entire profile, while the operational time only rises at the beginning of each mission Additionally, the cycles increase at the start of each mission.

(inversement, ils peuvent n’augmenter qu’à la fin de chaque mission) tandis que les opérations apparaissent seulement pendant la phase réelle de chaque mission.

Complète et limitée

Comprehensive data collection involves gathering information on all occurrences of the entity in operation, while limited data collection focuses on a subset, such as all entities used at a specific location or by a particular client Limited data collection may employ various sampling methods to determine which locations to select and how many entities to monitor.

To obtain information on widely consumed entities produced in large quantities, it is often necessary to send a batch to a local market where operational time and failure duration can be recorded Additionally, limiting marketing feedback to a representative market or client can streamline resource use and enhance data quality by focusing efforts on that specific sector.

Figure 5 illustrates a standard mission profile, where calendar time progresses continuously, while operating time only rises during the active mission phase Cycles increase at the beginning of each mission, or potentially at the end, while operations are confined to the actual mission duration.

Complete data collection involves gathering information on every instance of an item utilized in the field, while limited data collection focuses on a specific subset, such as items used in a particular location or by a specific customer To determine the locations and quantity of items to be monitored, limited data collection employs various sampling methods.

To gather information on mass-produced consumer items, companies often send a batch to a specific market to track usage and failure times Focusing marketing feedback on a typical market or customer allows for more efficient resource use and enhances the quality of data collected.

This technique is a form of sampling commonly used to optimize the information obtained from a dataset It is crucial to carefully select both the data collection and analysis methods while considering each other Often, data is collected without prior thought to the analysis process, which is an ineffective approach The starting point should be to determine the information needed for a specific population.

Le but principal de l’échantillonnage est de déterminer à partir de l’échantillon, l’information voulue ou le maximum d’informations possible sur la population à partir de laquelle il a été constitué

It is wise to categorize populations as follows: a) Defined and existing, such as members stored in a warehouse for a specific production line or apples on a tree Sampling from such populations can be random but is not straightforward due to the dependence of successive samples The sampling process can be simplified by replacing each member extracted from the population b) Infinite, as in the case where numbers are selected from a mathematically generated sequence It is noteworthy that sampling from a defined and existing population with replacement can be considered an infinite population since the process does not increase the quantity c) Hypothetical, as in a sequence of numbers obtained from dice rolls The series of rolls constitutes a sampling process that yields numbers existing from a non-existent population.

Les types de population répondant à la présente norme comprendront principalement ceux listés en a) et b) ci-dessus

It is noteworthy that many instances exist where members of a population are described by two (or sometimes more) mutually exclusive characteristics, such as functioning or failing, good or bad, favorable or unfavorable, and so on Members of such populations are said to possess attributes, and the sampling applied to these populations is referred to as attribute sampling When members of a population are distinguished from another based on a continuous and measurable characteristic (such as weight, time before failure, cost, etc.), the sampling process is called variable sampling.

A random sample is defined as one where each member of the population has a calculable probability of being selected There is no need to compute this probability; specifying and controlling the sampling process is sufficient to apply probability theory a) Simple random sampling occurs when each member has an equal chance of selection, and successive samples are independent b) Stratified sampling is another effective method that often surpasses simple random sampling This technique involves dividing the population into strata and taking subsamples from each layer.

In each stratum, every member of a population has an equal chance of being selected in a subsample, just like other members of that stratum This sampling method effectively distributes the sample more broadly across the population while preserving randomness within each stratum.

Sampling techniques are crucial for maximizing the information obtained from data sets It is essential to carefully choose both the data collection method and the analysis approach, ensuring they complement each other Unfortunately, many researchers collect data without considering the analysis process until after collection, which is an unsound practice The process should begin by clearly defining the information needed about the population.

The main aim of sampling is to determine from the sample, selected information or as much information as possible about the population from which it was drawn

Populations can be classified into three categories: a) Finite and existent populations, such as items from a production line or apples on a tree, where sampling may be random but not simple due to the dependence of successive draws This can be simplified by replacing each member after withdrawal b) Infinite populations, exemplified by selecting numbers from a mathematically generated sequence, where sampling from a finite population with replacement can be viewed as sampling from an infinite population since the supply is never exhausted c) Hypothetical populations, illustrated by the outcomes of rolling a die, where continuous rolls represent a sampling process that draws from a non-existent population.

The types of population dealt with in this standard will comprise mainly those listed in a) and b) above

Many populations can be categorized by mutually exclusive characteristics, such as "working or failed," "good or bad," and "favorable or unfavorable." These classifications often include terms like "good," "out-of-spec," or "catastrophically failed."

Sampling can be categorized into two types based on the characteristics of the population When members are identified by specific attributes, the process is known as sampling of attributes Conversely, when individuals are differentiated by continuous measurable traits such as weight, time-to-failure, or cost, it is referred to as sampling of variables.

A random sample is defined as one where every possible sample has a calculable probability of being selected, eliminating the need for actual calculations; instead, proper specification and control of the sampling procedure suffice for applying probability theory Simple random sampling ensures that every possible sample has an equal chance of selection, with independent successive drawings In contrast, stratified random sampling is another effective method that can offer advantages over simple random sampling in certain situations.

Tiêu đề	Guide d'application – Recueil de données de sûreté de fonctionnement dans des conditions d'exploitation
Chuyên ngành	Dependability Management
Thể loại	Standard
Năm xuất bản	2004

Định dạng
Số trang	88
Dung lượng	1,07 MB